Full Issue, Vol. 64 No. 1

Western North American Naturalist

Volume 64 Number 1 Article 21

2-20-2004

Follow this and additional works at: https://scholarsarchive.byu.edu/wnan

Recommended Citation (2004) "Full Issue, Vol. 64 No. 1," Western North American Naturalist: Vol. 64 : No. 1 , Article 21. Available at: https://scholarsarchive.byu.edu/wnan/vol64/iss1/21

This Full Issue is brought to you for free and open access by the Western North American Naturalist Publications at BYU ScholarsArchive. It has been accepted for inclusion in Western North American Naturalist by an authorized editor of BYU ScholarsArchive. For more information, please contact [email protected], [email protected]. Western North American Naturalist 64(1), ©2004, pp. 1–6

BIOGEOGRAPHIC AND CONSERVATION IMPLICATIONS OF LATE QUATERNARY PYGMY RABBITS (BRACHYLAGUS IDAHOENSIS) IN EASTERN WASHINGTON

R. Lee Lyman1

ABSTRACT.—Five implications of a biogeographic model of pygmy rabbits (Brachylagus idahoensis) in eastern Washing- ton proposed in 1991 are confirmed by 11 new late-Quaternary records. Pygmy rabbits from eastern Oregon colonized eastern Washington during the late Pleistocene and occupied their largest range during the middle and late Holocene. Disjunction of the eastern Washington population from that in eastern Oregon occurred during at least the late Holo- cene. Nineteenth-century cattle grazing and 20th-century agricultural practices reduced habitat preferred by pygmy rabbits. Conservation of the small remaining population of pygmy rabbits will necessitate altered land use practices.

Key words: agriculture, biogeography, Brachylagus idahoensis, conservation, grazing, pygmy rabbit, Washington.

Populations of pygmy rabbits (Brachylagus MATERIALS AND METHODS idahoensis) in eastern Washington are isolated from conspecific populations in southeastern Records reported by Lyman (1991) were Oregon, Nevada, and portions of adjacent states reviewed, and documents that appeared since (Fig. 1). Lyman (1991) outlined a hypothetical 1990 were examined for evidence of prehistoric biogeographic model for the populations in mammalian faunal remains in eastern Wash- eastern Washington based on 8 prehistoric ington. All but 2 of the 19 records of pygmy records. I have refined that model in light of rabbits derive from archaeological contexts 11 new prehistoric records of this diminutive (Table 1). McAllister (1995) believes that 1 of leporid from the area. those 2 from the Juniper Dunes Preserve com- Based on available data, Lyman (1991) hy- prises remains of Nuttall’s cottontail (Sylvi- pothesized that (1) pygmy rabbits colonized lagus nuttallii) that were originally misidenti- eastern Washington from Great Basin popula- fied. McAllister (1995) could not relocate the tions in eastern Oregon during the late Pleis- specimens on which the original identifications tocene; (2) eastern Washington populations were based, so this record plays no role in became disjunct from Great Basin populations analysis. For comparative purposes, 20th-cen- at the end of the Pleistocene, ca. 10,000 radio- tury records of pygmy rabbits were compiled carbon years before present (RCYBP); (3) pygmy from McAllister and Allen (1993), McAllister rabbits occupied much of the Columbia Basin (1995), and Johnson and Cassidy (1997). in eastern Washington prior to about 5500 I noted locations of all sites producing pre- RCYBP when sagebrush-dominated steppe historic remains of pygmy rabbits (Fig. 2), fre- habitats were at their maximum extent coinci- quency of pygmy rabbit remains, and evident with a period of greater-than-modern dence for determining the age of remains. All aridity; (4) after 4500–4000 RCYBP the range remains of pygmy rabbits are dated to the time of pygmy rabbits shrank as the range of big of sediment deposition, indicated by radio- sagebrush (Artemisia tridentata) decreased, carbon ages (all ages reported here are in commensurate with increased effective mois- RCYBP) determined from charcoal within the ture; and (5) agricultural practices over the last sediments or stratigraphically associated, tem- 100–150 years exacerbated the depletion of porally diagnostic artifacts. Given that pygmy sagebrush range and further depleted popula- rabbits burrow (Weiss and Verts 1984), indicated tions, thus creating the current range of pygmy ages may comprise maximum ages of remains. rabbits in eastern Washington. Many prehistoric pygmy rabbit specimens

1Department of Anthropology, 107 Swallow Hall, University of Missouri, Columbia, MO 65211.

1 2 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Fig. 2. Map of eastern Washington showing locations of 20th-century records of pygmy rabbits (+; from Johnson and Cassidy 1997, McAllister 1995, McAllister and Allen 1993) and prehistoric sites (numbers) that have produced remains of pygmy rabbits. County lines shown for refer- Fig. 1. Historic distribution of pygmy rabbit (horizontal ence. Numbers correspond to Table 1; M = Marmes hatching; after Lyman 1991). Vertical hatching shows area Rockshelter. of Figure 2. examined, however, are stained and weath- rabbits had begun colonizing eastern Wash- ered to the same degree as associated bones of ington at the end of the Pleistocene. Remains taxa believed to have been deposited at the of pygmy rabbits do not exist among 756 spec- same time as dated charcoal or artifacts. Pygmy imens identified to mammalian genus or species rabbits have relatively small home ranges (Katz- recovered from Marmes Rockshelter (45FR50) ner and Parker 1997), and all examined remains floodplain sediments (Caulk 1988, Gustafson of this species display no evidence that preda- and Wegener 1998). These sediments were tors transported them far (Lyman 1994). I there- deposited between 9500 and 10,500 RCYBP. If fore assume that location of their recovery is pygmy rabbits were not present there (Fig. 2) within a few kilometers of where the individu- at this time, then their colonization of central als lived. Frequencies of all mammalian re- Washington was not complete—the maximum mains discussed are given as the number of range of their distribution had not yet been identified specimens (NISP). attained—at the end of the Pleistocene. Although the sample is small (NISP = 427), BIOGEOGRAPHIC HISTORY faunal remains of middle-Holocene age (8000 to 4000 RCYBP) from Marmes Rockshelter The record from 45BN265 (Table 1:1) does also do not include specimens of pygmy rabbit not refute the 1st implication that pygmy rab- (Gustafson 1972). The single specimen from bits were in the process of colonizing eastern 45WT134 (Table 1:3), just east of Marmes Rock- Washington during the late Pleistocene (Fig. shelter, indicates that this taxon was in the 1); this record is 1 of only 2 positive pieces of general area at the end of the middle Holo- evidence. The single specimen of pygmy rab- cene (ca. 4000 RCYBP). bit from site 45KT1362 (Table 1:16) dates be- Together with the record from 45GR97 tween 10,700 and 10,000 RCYBP. It comprises (Table 1:2) dating to 8700 RCYBP, the 45BN265 the 2nd piece of positive evidence for the 1st (Table 1:1) and 45KT1362 (Table 1:16) records implication because it suggests that pygmy suggest pygmy rabbits were distributed across 2004] PYGMY RABBIT IN EASTERN WASHINGTON 3

TABLE 1. Prehistoric records of pygmy rabbits (Brachylagus idahoensis) in eastern Washington. Sitea Age (RCYBP) Reference 1. 45BN265 100,000–13,000 Rensberger et al. 1984 2. 45GR97 8700 Irwin and Moody 1978 3. 45WT134 4200–1000 Lyman 1990 4. 45GR445 2400–2150 Lyman n.d. 5. 45CH302 2500–150 Lyman unpublished data 6. 45AD2 <3000 Deaver and Greene 1978 7. 45FR5 1000–200 Olson 1983 8. Juniper Dunes Preserve <3000 Miller 1977b 9. 45LI150 <5500 Lyman unpublished data 10. 45LI206 1500–500 Lyman unpublished data 11. 45DO331 3000–900 Lyman 1995 12. 45AD104 1900 Lyman 1993 13. 45KT980 2300 Lyman 1998a, 1998b 14. 45KT1003 3000–1700 Lyman 1998b 15. 45KT338 1200 Flenniken et al. 1997, Lyman unpublished data 16. 45KT1362 10,700–10,000 Lyman unpublished data 17. 45YA641 unknown Lyman unpublished data 18. 45YA579 1700 Flenniken et al. 1997, Lyman unpublished data 19. 45YA533 7400, 2300 Flenniken et al. 1997 aSite number corresponds to map location in Figure 2. bSpecimen was likely misidentified; data omitted from analysis here.

at least a portion of their maximum range in lished data), suggests that the disjunction was eastern Washington during the early Holocene in place during the late Holocene (last 4000 (ca. 10,000–8000 RCYBP). Determination of RCYBP). the full extent of their range must await recov- All other records of pygmy rabbits date to ery of additional materials of late-Pleistocene the last 3000 RCYBP (Table 1). If in fact pygmy and early-Holocene age. Remains of pygmy rabbit range was decreasing after about 4000 rabbits from site 45YA533 (Table 1:19) may be RCYBP, then perhaps that range was larger than as much as 7400 years old, although their age suggested by Figure 2. For example, records is unclear (Flenniken et al. 1997). Most pygmy from eastern Kittitas County and northeastern rabbit remains from 45LI150 (Table 1:9) were Yakima County indicate pygmy rabbits occu- deposited by raptors (Lyman unpublished data). pied a range approximately 50 km west of their Because these remains were recovered from historically documented range during the late depositional contexts stratigraphically above Holocene. It can be conjectured on this basis ages of 5500 RCYBP, they must be younger. If that their remains eventually will be found in these remains date between 5500 and 4000 central Yakima County. RCYBP, they indicate pygmy rabbits occurred in the central Columbia Basin near the end of IMPLICATIONS FOR CONSERVATION the middle Holocene and they confirm the 3rd implication—that pygmy rabbits had their wid- Pygmy rabbits occupy stands of big sage- est distribution in eastern Washington during brush. They feed extensively on these plants the middle Holocene. and also use them for cover to avoid predation Given the lack of records for prehistoric (Green and Flinders 1980a, 1980b, Gabler et al. pygmy rabbits along the Lower Columbia River 2001). Paleoenvironmental data from eastern during the Holocene, the 2nd implication— Washington indicate that sagebrush was dense that eastern Washington populations became across much of the central Columbia Basin dur- disjunct from those in the northern Great Basin ing the middle Holocene, and that between at the end of the Pleistocene—cannot be about 4000 and 2000 RCYBP, a dense under- refuted. Circumstantial evidence, i.e., archae- story of grass existed (Chatters 1995, 1998). The ological samples in the McNary Reservoir area density of grass decreased after 2000 RCYBP that do not include remains of pygmy rabbit as climatic conditions took on modern charac- (Burtchard 1981, Cole 1968, Lyman unpub- teristics. This suggests that pygmy rabbits may 4 WESTERN NORTH AMERICAN NATURALIST [Volume 64 have occurred in locations during the last 2000 cate that surface sediment “strength” (kg ⋅ cm–2 years where they have not been historically necessary to penetrate) was significantly lower documented. in loci occupied by pygmy rabbits than in Katzner and Parker (1997) suggest that graz- unoccupied loci. Therefore, compaction of sed- ing has caused some loss of preferred pygmy iment by grazing cattle likely decreased the rabbit habitat. Grazing by cattle and sheep in suitability of sediment for pygmy rabbits. If eastern Washington began in the mid-19th use of an area by grazing cattle also decreased century, and “by the beginning of the 1870s, the abundance of sagebrush, then cattle would the intermountain country of the Pacific North- have had a significant influence on pygmy rab- west had become pretty generally a cattleman’s bit habitat. Alterations to local floras by 19th- country” (Oliphant 1947:220). No systematic century grazing practices were followed by census data on cattle numbers are available, tillage, which ultimately resulted in less sage- but a single estimate from the 1880s suggested brush. In addition, when irrigation began in 20,000 head in Yakima Valley (Oliphant 1932). the early 20th century, sagebrush was eradi- The Yakima River valley of Yakima County was cated by herbicides and fire and replaced by used as winter range and the Kittitas River fields and grasslands (Buechner 1953). valley of Kittitas County as summer range Comparison of 20th-century records of (Oliphant 1947). Range deterioration was al- pygmy rabbits (McAllister and Allen 1993, ready apparent in the 1880s (Buechner 1953). McAllister 1995, Johnson and Cassidy 1997) It took the form of overgrazing, reducing or with prehistoric records, irrespective of age, eliminating native grasses in various areas and suggests reduction of pygmy rabbit range (Fig. reducing vegetation cover (Daubenmire 1940, 2). The 1st formal surveys and collections of Daubenmire and Colwell 1942); reduced veg- local mammals took place in the early 20th cen- etation altered erosion (Young 1943). Prior to tury (Dice 1919, Couch 1928, Taylor and Shaw the introduction of tractors, wheat farmers used 1929, Svihla and Svihla 1940, Larrison 1944, horses to pull farm equipment. Once farmers Booth 1947, Dalquest 1948, Johnson et al. 1950). adopted the tractor, shortly after the begin- Any historic distribution map of pygmy rab- ning of the 20th century, many work horses bits is, therefore, at least partially a function of were turned out to pasture and contributed to 19th-century agricultural practices. The 5th overgrazing (Chohlis 1952). implication of Lyman’s (1991) model—that Many cattle died during harsh winters and agricultural practices depleted pygmy rabbit prompted the initiation of artificial feeding at range via habitat destruction—seems to be the end of the 19th century (Oliphant 1932). borne out. Cattle consume significantly more grass than CONCLUSION browse in local shrub-steppe habitats (Uresk and Rickard 1976, Hanley and Hanley 1982). Lyman’s (1991) biogeographic model has When concentrations of cattle were high in implications for modern management of the eastern Washington, sagebrush “suffered heav- pygmy rabbit. The extant population of pygmy ily from breakage” (Daubenmire 1970:13). Harsh rabbits in eastern Washington is restricted to winters there may have forced cattle to consume central Douglas County (McAllister 1995). even more browse, resulting in marked destruc- Holocene climatic change and modern agri- tion of sagebrush. With the replacement of cultural practices especially have significantly horses by tractors, more acreage was tilled, depleted habitat suitable for this leporid and some of which had previously been cattle and reduced the size and range of the extant popu- horse pasture (Oliphant 1948, Chohlis 1952). lation. To maintain the isolated and unique ≥ Pygmy rabbits also require deep ( 50 cm), gene pool represented by the extant popula- friable sediment, often with high sand content, tion of pygmy rabbits in eastern Washington, in which to dig burrows (Weiss and Verts 1984, steps must be taken to alter modern land- Gabler et al. 2001). It is well documented that modification activities. cattle grazing causes increased compaction of sediment (Chanasky and Naeth 1995). In some ACKNOWLEDGMENTS cases the bulk density of sediment increases as much as 20% (McCarty and Mazurak 1976). I thank J.R. Galm and S. Gough for the op- Data reported by Weiss and Verts (1984) indi- portunity to study many of the pygmy rabbit 2004] PYGMY RABBIT IN EASTERN WASHINGTON 5 remains reported here. Helpful comments on ties, Washington. Lithic Analysts Research Report 52. an early draft were provided by M.J. O’Brien 250 pp. GABLER, K.I., L.T. HEADY, AND J.W. LAUNDRÉ. 2001. A and several anonymous reviewers. habitat suitability model for pygmy rabbits (Brachy- lagus idahoensis) in southeastern Idaho. Western LITERATURE CITED North American Naturalist 61:480–489. GREEN, J.S., AND J.T. FLINDERS. 1980a. Brachylagus ida- BOOTH, E.S. 1947. Systematic review of the land mammals hoensis. Mammalian Species 125:1–4. of Washington. Doctoral dissertation, State College ______. 1980b. Habitat and dietary relationships of the of Washington (Washington State University), Pull- pygmy rabbit. Journal of Range Management 33: man. 646 pp. 136–142. BUECHNER, H.K. 1953. Some biotic changes in the state of GUSTAFSON, C.E. 1972. Faunal remains from the Marmes Washington, particularly during the century 1853– Rockshelter and related archaeological sites in the 1953. Research Studies of the State College of Wash- Columbia Basin. Doctoral dissertation, Washington ington 21:154–192. State University, Pullman. 183 pp. USTAFSON AND EGENER BURTCHARD, G.C. 1981. Test excavations at Box Canyon G , C.E., R. W . 1998. Faunal remains and three other side canyon sites in the McNary analysis. Pages 99–151 in B.A. Hicks, editor, Marmes Reservoir. Washington State University, Laboratory Rockshelter (45FR50) preliminary report: 1998 re- of Archaeology and History, Project Report 10. 122 pp. sults. Confederated Tribes of the Colville Reserva- tion report to the Walla Walla District U.S. Army CAULK, G.H. 1988. Examination of some faunal remains from the Marmes Rockshelter floodplain. Master’s Corps of Engineers. thesis, Washington State University, Pullman. 103 pp. HANLEY, T.A., AND K.A. HANLEY. 1982. Food resource partitioning by sympatric ungulates on Great Basin range- CHANASKY, D.S., AND M.A. NAETH. 1995. Grazing impacts on bulk density and soil strength in the foothills fes- land. Journal of Range Management 35:152–158. IRWIN, A.M., AND U. MOODY. 1978. The Lind Coulee site cue grasslands of Alberta, Canada. Canadian Journal (45GR97). Washington State University Archaeologi- of Soil Science 75:551–557. cal Research Center Project Report 56. 367 pp. CHATTERS, J.C. 1995. Population growth, climatic cooling, JOHNSON, M.L., P.W. CHENEY, AND T.H. SCHEFFER. 1950. and the development of collector strategies on the Mammals of the Grand Coulee, Washington. Murrelet Southern Plateau, western North America. Journal of 31:39–42. World Prehistory 9:341–400. JOHNSON, R.E., AND K.M. CASSIDY. 1997. Terrestrial mam- ______. 1998. Environment. Pages 29–48 in D.E. Walker, mals of Washington state: location data and predicted Jr., editor, Handbook of North American Indians, distributions. Volume 3 in K.M. Cassidy, C.E. Grue, volume 12: Plateau. Smithsonian Institution, Wash- M.R. Smith, and K.M. Dvornich, editors, Washington ington, DC. state gap analysis—final report. Washington Cooper- CHOHLIS, G.J. 1952. Range condition in eastern Washing- ative Fish and Wildlife Unit, University of Washing- ton fifty years ago and now. Journal of Range Man- ton, Seattle. 304 pp. agement 5:129–134. KATZNER, T.E., AND K.L. PARKER. 1997. Vegetative charac- COLE, D.L. 1968. Report on archaeological research in teristics and size of home ranges used by pygmy rab- the John Day Dam reservoir area—1967. University bits (Brachylagus idahoensis) during winter. Journal of Oregon Museum of Natural History, report to the of Mammalogy 78:1063–1072. National Park Service. 53 pp. LARRISON, E.J. 1944. An early spring reconnaissance of COUCH, L.K. 1928. Small mammals of the Yakima Valley, the birds and mammals of Northrup Canyon, Upper Washington. Murrelet 9(1):9–14. Grand Coulee, Washington. Murrelet 25:35–42. DALQUEST, W.W. 1948. Mammals of Washington. Univer- LYMAN, R.L. 1990. Zooarchaeology. Pages 98–138 in D.R. sity of Kansas Museum of Natural History Publica- Brauner, editor, Archaeological data recovery at tions 2, Lawrence. 444 pp. Hatiupuh, 45WT134, Whitman County, Washington. DAUBENMIRE, R. 1940. Plant succession due to overgraz- Oregon State University Department of Anthropol- ing in the Agropyron bunchgrass prairie of south- ogy, report to the Walla Walla District U.S. Army eastern Washington. Ecology 21:55–64. Corps of Engineers. ______. 1970. Steppe vegetation of Washington. Washing- ______. 1991. Late Quaternary biogeography of the pygmy ton State University Agricultural Experiment Sta- rabbit (Brachylagus idahoensis) in eastern Washing- tion, Technical Bulletin 62. 131 pp. ton. Journal of Mammalogy 72:110–117. DAUBENMIRE, R., AND W.E. COLWELL. 1942. Some edaphic ______. 1993. The faunal collection. Pages 39–56 in V. changes due to overgrazing in the Agropyron–Poa Morgan, editor, Cultural resource investigations at prairie of south-eastern Washington. Ecology 23: 45AD104, an upland bison site, Adams County, Wash- 32–40. ington. Eastern Washington University Reports in DEAVER, K., AND G.S. GREENE. 1978. Faunal utilization at Archaeology and History 100-81. 45AD2: a prehistoric archaeological site in the ______. 1994. Vertebrate taphonomy. University of Cam- Channeled Scablands of eastern Washington. Tebiwa, bridge Press, Cambridge. 524 pp. Miscellaneous Papers of the Idaho State University ______. 1995. Zooarchaeology of the Moses Coulee Cave Museum of Natural History 14:1–21. (45DO331) spoils pile. Northwest Anthropological DICE, L.R. 1919. The mammals of southeastern Washing- Research Notes 29:141–176. ton. Journal of Mammalogy 1:10–22. ______. 1998a. Zooarchaeology of six sites in the Yakima FLENNIKEN, J.J., J.D. GALLISON, AND F. D . M ILLER. 1997. Training Center expansion area. Pages F.1–F.41 in K. Yakima Training Center National Register eligibility Boreson, editor, Evaluation of cultural resources in testing deep alluvial sites, Yakima and Kittatas Coun- construction impact areas, Yakima Training Center 6 WESTERN NORTH AMERICAN NATURALIST [Volume 64

expansion area, Kittitas County, Washington. Eastern OLSON, D.L. 1983. A descriptive analysis of the faunal Washington University Reports in Archaeology and remains from the Miller site, Franklin County, Wash- History 100–88. ington. Master’s thesis, Washington State University, ______. 1998b. Zooarchaeology of five sites in the Yakima Pullman. 147 pp. Training Center expansion area. Pages A.3.1–A.3.46 RENSBERGER, J.M., A.D. BARNOSKY, AND P. S PENCER. 1984. in S. Gough, editor, Yakima Training Center expan- Geology and paleontology of a Pleistocene-to-Holo- sion area archaeology: investigations in the Johnson cene loess succession, Benton County, Washington. Creek drainage basin, Kittitas County, Washington. Eastern Washington University Reports in Archaeol- Eastern Washington University Reports in Archaeol- ogy and History 100–39. 105 pp. ogy and History 100–93. SVIHLA, A., AND R.D. SVIHLA. 1940. Annotated list of the MCALLISTER, K.R. 1995. Washington state recovery plan mammals of Whitman County, Washington. Murrelet for the pygmy rabbit. Washington Department of 21:53–58. Fish and Wildlife, Olympia. 73 pp. TAYLOR, W.P., AND W. T. S HAW. 1929. Provisional list of land MCALLISTER, K.R., AND H. ALLEN. 1993. Status of the mammals of the state of Washington. Occasional pygmy rabbit (Brachylagus idahoensis) in Washing- Papers of the Charles R. Conner Museum 2. State ton. Washington Department of Fish and Wildlife, College of Washington (Washington State Univer- Olympia. 31 pp. sity), Pullman. 32 pp. MCCARTY, M.K., AND A.P. MAZURAK. 1976. Soil compaction URESK, D.W., AND W.H. RICKARD. 1976. Diets of steers on in eastern Nebraska after 25 years of cattle grazing a shrub-steppe rangeland in south-central Washing- management and weed control. Journal of Range ton. Journal of Range Management 29:464–466. Management 29:384–386. WEISS, N.T., AND B.J. VERTS. 1984. Habitat and distribu- MILLER, S.M. 1977. Mammalian remains from the Juniper tion of pygmy rabbits (Sylvilagus idahoensis) in Ore- Forest Preserve, Franklin County, Washington. Mas- gon. Great Basin Naturalist 44:563–571. ter’s thesis, University of Idaho, Moscow. 40 pp. YOUNG, V.A. 1943. Changes in vegetation and soil of Palouse OLIPHANT, J.O. 1932. Winter losses of cattle in the Oregon prairie caused by overgrazing. Journal of Forestry Country, 1847–1890. Washington Historical Quar- 41:834–838. terly 23:3–17. ______. 1947. The cattle herds and ranches of the Oregon Received 7 May 2002 country, 1860–1890. Agricultural History 21:217–238. Accepted 22 November 2002 ______. 1948. History of livestock industry in the Pacific Northwest. Oregon Historical Quarterly 49:3–29. Western North American Naturalist 64(1), ©2004, pp. 7–17

FATE AND CHARACTERISTICS OF PICEA DAMAGED BY ELATOBIUM ABIETINUM (WALKER) (HOMOPTERA: APHIDIDAE) IN THE WHITE MOUNTAINS OF ARIZONA

Ann M. Lynch1

ABSTRACT.—Spruce aphid, Elatobium abietinum (Walker), is a new invasive pest in high-elevation forests of southwestern USA. Plots in the White Mountains of Arizona were evaluated over several years to assess the extent and severity of damage in high-elevation forests and to identify tree and site characteristics associated with defoliation and mortality. Large areas were defoliated in each of 4 recent outbreaks. Impact from a single defoliation episode included an overall mortality of 10.3% to Picea engelmannii Parry, 24%–41% in severely defoliated trees. Defoliation severity was much greater on P. engelmannii than on P. pungens Engelm. and was more severe in the lower canopy layers. Retention of foliage in the upper-crown third of individual trees was a critical factor in tree survivorship. Mortality was associated with defoliation severity and severe infection by Arceuthobium microcarpum (Engelmann) Hawksworth & Wiens. Picea pungens was much more susceptible to A. microcarpum than was P. engelmannii. The combined effects of high levels of defoliation and mistletoe infection were lethal, resulting in almost 70% mortality. Mortality continued to occur at least 3 years after defoliation. This aphid will affect natural disturbance regimes and tree population dynamics in mixed-conifer and spruce-fir forests of the American Southwest.

Key words: Elatobium abietinum, Arceuthobium microcarpum, Picea engelmannii, Picea pungens, invasive species, exotic, disturbance ecology, insect impact.

Spruce aphid, Elatobium abietinum (Walker) defoliation on 156,800 acres in the White (Homoptera: Aphididae), is a new invasive pest Mountains (USDA Forest Service 2001). in the interior southwestern United States Elatobium abietinum feeds on dormant, (Lynch 2003). This species probably originated mature Picea needles (Parry 1976, Jackson and in northern Europe (Hanson 1952, Bejer-Peter- Dixon 1996). Epizootics of this insect occur sen 1962, Carter and Halldórsson 1998), where during the spring in the Pacific Northwest and it is known as green spruce aphid. It has been other areas with maritime climate (Bejer- known from the Pacific Northwest coastal areas Petersen 1962, Parry 1973, Heie 1989, Day of North America since 1916 (Koot and Ruth and Crute 1990, Carter and Halldórsson 1998) 1971, Carter and Halldórsson 1998). It was and during the fall in high-elevation forests of found in urban Santa Fe, New Mexico, in 1976, the Southwest (Lynch 2003). Relatively minor where it has been an intermittent pest in the population increases can occur in maritime urban forest. The first wildland outbreak in climates in the fall (Hussey 1952, Bevan 1966). the Southwest occurred over the 1989–90 win- The host species in the Southwest are Picea ter in the White Mountains of Arizona, causing engelmannii Parry (Engelmann spruce) and P. defoliation on more than 100,000 acres (USDA pungens Engelm. (Colorado blue spruce). The Forest Service 1997). Three subsequent out- insect sucks sap from needle phloem cells, breaks occurred over the winters of 1995–96, causing needle necrosis, dehydration, and pre- 1996–97, and 1999–2000 (USDA Forest Ser- mature drop (Bevan 1966). Tree mortality is vice 1997, 1998, 2001). The range has expanded uncommon in maritime areas, where the most to include the Mogollon Mountains (just east important host species are P. sitchensis (Bong.) of the White Mountains) and Sacramento Carr. (Sitka spruce), P. glauca (Moench) Voss Mountains in New Mexico, and the Pinaleño (white spruce), and P. abies (L.) Karst (Norway Mountains and San Francisco Peaks in Arizona. spruce; Bevan 1966, Carter 1977, Day and The last outbreak was very extensive, causing McClean 1991, Seaby and Mowat 1993, Thomas

1USDA Forest Service, Rocky Mountain Research Station, 2500 South Pine Knoll, Flagstaff, AZ 86001-6381.

7 8 WESTERN NORTH AMERICAN NATURALIST [Volume 64 and Miller 1994, Straw et al. 1998). Elatobium plots), Pseudotsuga menziesii var. glauca (Mirb.) abietinum has been a chronic pest in the Pacific Franco–dominated, spruce-fir (almost all P. Northwest, but occasionally it causes severe engelmannii–A. lasiocarpa var. arizonica), and tree mortality in local areas (Koot 1992). pure spruce (either pure P. engelmannii or a I conducted this study to evaluate the ex- mix of both Picea species). Plots of relatively tent and severity of damage in high-elevation pure spruce were above 9100 feet in elevation forests and to identify tree and site character- and composed of medium- to large-sized trees istics associated with defoliation and mortality. (mean dbh of 8–14 inches) with fairly uniform Emphasis was placed on identifying tree char- diameter distributions (usually with coefficient acteristics associated with damage and mortal- of variation of the mean [CVM] of dbh < 30%). ity because the annual Aerial Detection Sur- Spruce dominance (based on density) ranged veys indicated extensive defoliation with little from 2% to 100%, with a mean of 32%. Picea apparent site or stand variability within the pungens was present on 14 plots but was the defoliated areas (Fig. 1; USDA Forest Service major component on only 4 of those. 1998, 1999). This approach will allow assessment of potential damage in subsequent out- METHODS breaks and in other mountain ranges. Western spruce dwarf mistletoe, Arceutho- The plots consist of high-elevation plots from bium microcarpum (Engelmann) Hawksworth the Fort Apache Indian Reservation Continu- & Wiens (Viscaceae), is a localized parasite of ous Forest Inventory (CFI) plot system. The Picea in Arizona and New Mexico. The most CFI system and spruce-fir population dynam- damaging disease agent in southwestern mixed- ics were described by Moran-Palma and conifer forests dominated by Picea, it is noted McTague (1997). The CFI is designed to pro- for causing an unusually high rate of mortality vide data on growth, yield, and mortality of in P. pungens (Mathiasen et al. 1986, Hawks- the entire forest rather than of individual sites worth and Wiens 1996). Evaluation of A. or stands. The CFI consists of 3-plot clusters microcarpum effects was included in the study on a 100-chain (2012-m) grid. Within each after analyses indicated that it plays an impor- cluster, circular plots of 0.20 acres (0.08 ha) tant role in the fate of trees defoliated by E. are usually located on a north–south line, 5 abietinum. chains (101 m) apart. For this study individual plots are considered individual observations STUDY SITES without regard to cluster. Variance of tree den- Thirty-seven 0.20-acre (0.08-ha) study plots sity, stocking, and dominance by P. engelman- were surveyed in 1997, 1999, 2000, and 2001. nii was as great between plots as between The plots include considerable variation in clusters, which is not surprising given the dis- elevation, species composition, density, diameter tance between plots. distribution, and dominance by spruce. Eleva- Tree- and plot-level effects were evaluated. tion ranges from 8174 feet to 9698 feet (2491 m Individual tree measurements taken in 1997 to 2956 m), with a mean of 9130 feet (2783 m). on live Picea greater than 5 inches dbh in- The sampled area excludes some of the high- cluded species, dbh, activity of defoliators and est-elevation spruce-fir forest in the White spruce beetle (Dendroctonus rufipennis Kirby Mountains, which is in the Mount Baldy Wilder- [Coleoptera: Scolytidae]), defoliation index, and ness Area, and a 10,000-acre (4047-ha) area dwarf mistletoe rating. Defoliation index was defoliated in 1995–1999 by Nepytia janetae computed as the sum of 3 crown-third ratings, Rindge (Lepidoptera: Geometridae; USDA For- where each crown-third was rated as 0, 1, 2, or est Service 1999, 2000, Lynch and Fitzgibbon 3, by 33% defoliation classes (an index of 9 in- 2004). Density varies from 15 tpa to 590 tpa dicates that each crown-third was 67%–100% (37 tph to 1458 tph), with spruce density vary- defoliated). Trees or plots with defoliation in- ing from 5 tpa to 155 tpa (12 tph to 383 tph). dices of 0–3, 4–6, and 7–9 were considered Plots include variable and relatively uniform lightly, moderately, and severely defoliated, diameter distributions, indicating that even- respectively. Severity of A. microcarpum infec- and mixed-aged stands were represented. tion was assessed using the Hawksworth (1977) The range of species composition includes 6-class dwarf mistletoe rating index (DMR). mixed-conifer (5 species were present on 14 Trees or plots with DMR values of 4, 5, and 6 2004] FATE OF PICEA AFTER ELATOBIUM DEFOLIATION 9

Fig. 1. High-elevation areas mapped during the 1996 and 1997 Aerial Detection Surveys as defoliated by E. abietinum in the White Mountains in the 1995–96 and 1996–97 defoliation episodes, respectively, or burned (hatched). Contour intervals are 200 m (656 feet.).

were considered severely infected. Picea were assumption that a uniform diameter distribu- reassessed in 1999, 2000, and 2001 for bark tion indicates a relatively even-aged stand. beetle activity, defoliation, and mortality. Addi- Foliage damage on P. engelmannii and P. tionally, the size of individual Picea relative to pungens from E. abietinum is difficult to evalu- the size of neighboring Picea was evaluated as ate with precision. Stippling and banding the dbh of the individual tree divided by the observed on E. abietinum–damaged P. sitchen- plot mean dbh for Picea, such that the smaller sis foliage in Europe and the Pacific North- and larger Picea on the plot had values < or west are not always found on P. engelmannii >1, respectively. and P. pungens in Arizona. Although fading and For each plot, elevation was recorded with yellow or red discoloration do occur, they are a GPS unit and aspect was estimated from not reliable indicators of E. abietinum feeding. U.S. Geological Survey 7.5-minute quadrangle Foliage usually changes color very little, losing maps. Plot data of tree density, dominance by some of its brightness and looking slightly individual species, and mortality were computed grayer than healthy foliage. This change is small, from tree data. Density data were converted uncertain, and unobservable under many con- to per-acre units for analysis and presentation. ditions (dusk, cloudiness, bright sun). Many Density and species dominance data for species needles are gray-green when they fall. Mini- other than Picea were taken from the 1994 and mal fading of foliage of P. engelmannii is known 1995 CFI measurements. Standard deviations to delay the detection of D. rufipennis out- and CVM of dbh were used at plot level to breaks (Furniss and Carolin 1977). I suspect evaluate uniformity of tree size, with the that some aphid-damaged foliage is retained 10 WESTERN NORTH AMERICAN NATURALIST [Volume 64 for a year or longer after foliage or tree death, 1995). Paired Student sample t tests were used as considerable feeding by Piciformes birds on to compare defoliation severity by crown-thirds. bark beetles and wood boring beetles was I used Principal Component Analysis to observed on trees with green foliage. A signifi- summarize vegetation and site data. Scores of cant amount of woodpecker feeding indicates the first 3 components were subsequently the presence of wood-boring beetles in dead computed for each plot and assessed for asso- wood or the presence of a large number of ciation with defoliation and mortality (Ise- bark beetles, such that the tree is already dead brands and Crow 1975, Nichols 1977, Gauch or dying. Defoliation estimates made during 1982). Principal components are no longer a this study include lost foliage and foliage dis- favored method for devising predictor models colored yellow or red, and are probably con- but remain a good tool for summarizing multi- servative. Also, defoliation from a sap-sucking dimensional data prone to multicolinearity, insect is not the same as defoliation from a particularly when the components are easy to leaf-chewing insect such as Choristoneura occi- interpret and normally distributed and the dentalis Freeman (Lepidoptera: Tortricidae) predictor variable is exclusive of those used in and Orgyia pseudotsugata (McDunnough) the components (Ludwig and Reynold 1988, (Lepidoptera: Lymantriidae). The needle re- Jackson 1991). This is the case here, where the plenishes and subsequently loses fluid removed vegetation-descriptor components are assessed by the aphid. Eventually the needle may die for association with insect damage. The 1st com- from necrosis or dehydration. Therefore, defo- ponent distinguished between mixed-conifer liation estimates made here are not directly plots (especially those with Populus tremu- comparable to similar levels of defoliation from loides Michx. present) and those dominated a leaf-chewing insect. by P. engelmannii. The 2nd component distin- Defoliation episodes occurred over the fall guished between the plots from highest eleva- and winter seasons of 1995–96, 1996–97, and tions with variable-diameter stems dominated by Abies lasiocarpa var. arizonica (Merriam) 1999–2000. The 1996–97 episode was more Lemm. and those dominated by Picea. The severe and extensive than the earlier episode, 3rd component summarized variability associ- and very little acreage was defoliated twice ated with warmer habitats and non-host species, (Fig. 1). Mortality and defoliation were sur- primarily Pinus ponderosa Laws and Abies veyed in 1997, 1999, 2000, and 2001. Recon- concolor (Gord. & Glend.) Lindl. Each of the naissance in 1998 indicated that little if any first 3 components described 15% to 20% of mortality occurred that year. Defoliation esti- the shared linear variability in the plot data. I mates taken in 1997 therefore represent a sin- analyzed the component scores, as well as gle defoliation event for most trees and plots. I other plot-level statistics, for associations with evaluated mortality from 2000 rather than 2001 defoliation and mortality using Pearson corre- to avoid the complication of the 1999–2000 lation coefficients (r). Significance of all tests episode. Usually, mortality to individual P. was evaluated at 0.10. engelmannii and P. pungens is not positively detectible until 2 years after the fatal event, so RESULTS mortality estimates made during 2000 include few if any trees that were killed by the Elatobium abietinum Defoliation 1999–2000 defoliation episode. Picea engelmannii was significantly more Contingency analyses with likelihood ratios defoliated than P. pungens (Z = –5.46, P < (G-statistics) were used to evaluate associa- 0.000; Figs. 2a, 2b). For all trees in the sample, tions between mortality and dbh class, DMR, mean 1997 defoliation index was 5.0 on P. and defoliation index (Sokal and Rohlf 1995). engelmannii and 2.3 on P. pungens. Thirty- Kolmogorov-Smirnov 2-sample tests (Z-statis- three percent of P. engelmannii were severely tics), which are sensitive to the shape of the defoliated, compared with 8% of P. pungens. frequency distribution as well as to central Fewer than 15% of P. pungens lost more than tendency, were used to evaluate associations 33% of their foliage. All age classes of needles between species or live/dead groups and indi- were affected except the most recently pro- vidual tree measurements such as dbh class, duced cohort. Some of the most recent needle DMR, and defoliation index (Sokal and Rohlf cohort were lost, but the amount was minimal. 2004] FATE OF PICEA AFTER ELATOBIUM DEFOLIATION 11

Fig. 2. Frequency of P. engelmannii (white) and P. pungens (gray) in 1997, and dead spruce (black) in 2000 by 1997 defoliation index (a,b) and dwarf mistletoe rating (c,d). Note that scale is the same for a and b, but different for c and d.

Defoliation severity of trees 5 inches dbh index was weakly correlated (|r| ≤ 0.4, P ≤ and larger was not directly associated with 0.10) with increased density and dominance dbh class (G = 6.85, P = 0.553) or DMR (G = by Populus tremuloides and A. lasiocarpa var. 65.3, P = 0.140). There was a weak, negative arizonica, decreased density and dominance correlation (r = –0.204, P = 0.00) between P. by Pinus ponderosa Laws., and the 2nd princi- engelmannii defoliation index and relative pal component (which distinguished fir-domi- diameter, such that defoliation index was gen- nated plots from Picea-dominated plots). Within- erally greater on those trees with dbh smaller plot variability in defoliation was low: 54% than the species’ average dbh on the same CVM on average with a range from 8% to 128%. plot. Inspection of the data showed that few P. The estimates above, made in 1997, include engelmannii with diameters less than 50% of defoliation from the 2 defoliation episodes in the plot Picea dbh average were lightly defoli- the 1995–96 and 1996–97 fall–winter seasons ated. Defoliation severity of larger trees did (Fig. 1). The latter was more extensive than not appear to be associated with relative size. the former, and there was little overlap in the areas defoliated in the 2 episodes. A later out- Defoliation was more severe in the lower break in 1999–2000 was orders of magnitude portions of the tree crowns (upper < middle, more severe and extensive. Because of the 2- and middle < lower, P < 0.02). For example, year (or greater) lag between defoliation and of trees with defoliation index of 7, 48% re- tree mortality, tree fate after the 1999–2000 tained at least 2/3 of the upper-crown foliage, outbreak cannot be evaluated yet; only defoli- while 97% lost at least 2/3 of the lower-crown ation patterns can be evaluated. Severity of foliage. This pattern was consistent in the defoliation estimates made in 2000 correlated other defoliation index classes. weakly but significantly (|r| < 0.4, P < 0.10) On a plot basis, distribution of plot frequency with increased tree and Abies density and by mean defoliation index was very similar to with decreased Pinus ponderosa density and that shown in Figures 2a and 2c for tree fre- dominance by Pinus. Additionally, defoliation quency by defoliation index. Twenty-six per- estimates made in 2001 (there is also a lag cent of plots with P. engelmannii were severely between aphid feeding and needle drop) cor- defoliated. Picea engelmannii mean defoliation related with increased Populus tremuloides 12 WESTERN NORTH AMERICAN NATURALIST [Volume 64 density and dominance, decreased elevation, and 67% of P. pungens and P. engelmannii with and P. engelmannii density and dominance. DI of 8 or 9 and DMR of 5 or 6 died by 2000. Most P. engelmannii mortality was due to Dwarf Mistletoe Effects spruce aphid damage or to the combined effects Eighty-seven percent and 48% of P. pun- of aphid damage and mistletoe infection, gens and P. engelmannii, respectively, were while recent P. pungens mortality has been infected by A. microcarpum. Picea pungens mostly due to dwarf mistletoe infection (Fig. was significantly more severely infected by A. 3a). To date, bark beetle activity has not microcarpum than was P. engelmannii (Z = increased substantially in Picea in the White –5.5, P < 0.001), with mean DMR values of Mountains. 3.2 and 1.8, respectively (Figs. 2c, 2d). Picea The pattern of crown defoliation affected pungens was present on 14 plots, 10 of which tree survivorship: mortality rates of 3.4%, also had P. engelmannii, so numbers shown 24.2%, and 41.2% were associated with defoli- here reflect greater susceptibility to A. micro- ation indices of 7, 8, and 9, respectively. carpum and not a sampling bias in favor of P. Invariably, the difference between class-7 pungens representation. Thirteen of 14 plots had trees and class-8 and -9 trees was that the top A. microcarpum–infected P. pungens (93%). crown-third was less severely defoliated. Like- Arceuthobium microcarpum was found over wise, class-8 trees were less severely defoli- the entire elevational range (8174–9698 feet) ated in the top crown-third than were class-9 covered by the study plots. Mean plot DMR trees. Lack of defoliation in the top third of of each species was not significantly correlated the crown was a good predictor of survivorship. with elevation, aspect, tree density, or density Mortality continued to occur until at least and dominance of P. pungens. 2001 (Fig. 3b), 3 or 4 years after the 1995–96 Fate of Individual Trees and 1996–97 defoliation episodes. The P. engelmannii trees that died in 2001 had high Picea engelmannii mortality was signifi- DMR values or high defoliation indices or cantly associated with defoliation index (G = 62.1, P = 0.000; Fig. 2a) and with A. micro- both in 1997 or 2000, but few had high defoli- carpum infection (G = 25.0, P = 0.000; Fig. ation indices in both events. Based on patterns 2c). Mean 1997 defoliation index was signifi- seen after the 1995–96 and 1996–97 outbreaks, cantly higher for trees that died than for those additional mortality should occur in 2002 and that survived (7.9 and 4.7, respectively; Table 1). 2003. Mortality of lightly, moderately, and severely Plot Characteristics Associated defoliated P. engelmannii was 4%, 6%, and 30% with P. engelmannii Mortality (3.4%, 24.2%, and 41% for trees with defoliation indices of 7, 8, and 9, respectively). Mean Mean plot mortality in 2000 of P. engelman- DMR of survivor P. engelmannii was 1.5, com- nii on lightly, moderately, and heavily defoli- pared with 3.4 for those that died (Table 1). ated plots was 4.8%, 9.5%, and 29.2%, respec- DMR was 4 or higher for 50% of the P. engel- tively. The mean for all plots was 10.3%. Mean mannii that died. plot mortality of P. engelmannii was positively Picea pungens mortality was significantly correlated (P ≤ 0.10) with mean P. engelmannii associated with DMR (G = 18.2, P = 0.006). defoliation, mean plot DMR, density and Mean DMR of survivors was 2.8, compared dominance of both Pinus ponderosa and P. with 4.6 for those that died (Table 1). Less strobiformis Engelm. (both individually and than 3% of P. pungens with DMR of 3 or less combined), and the 3rd principal component, died, but 19% of trees with DMR of 4 or which described warm habitats with Pinus higher died (Fig. 2d). Arceuthobium micro- (Table 2). Linear regression with 1997 defolia- carpum was one of the causal agents for 75% tion index (97DI), DMR, and pine dominance of P. pungens that died (Fig. 3a). There were (Pine) explained 54% of the variability in plot too few heavily defoliated P. pungens to deter- mortality of P. engelmannii (F = 13.7, P = mine if mortality was significantly associated 0.000, n = 32): with defoliation index (Fig. 2b). The combined effects of high levels of defo- Percent mortality = –14.0 + (2.9 * 97DI) liation and mistletoe infection were lethal: 66% + (3.2 * DMR) + (2.0 * Pine). 2004] FATE OF PICEA AFTER ELATOBIUM DEFOLIATION 13

TABLE 1. Mean 1997 defoliation index and dwarf mistletoe rating (DMR) of trees that died or survived. Survivors Dead mean (s)mean (s) n Z P P. engelmannii 1997 defoliation index 4.7 (2.7) 7.9 (1.6) 285 –6.332 0.000 DMR 1.5 (2.2) 3.4 (2.6) 285 –4.299 0.000 P. pungens 1997 defoliation index 2.2 (2.0) 3.7 (3.8) 78 –0.244 0.807 DMR 2.8 (2.1) 4.6 (2.1) 89 –1.987 0.047

Fig. 3. Cause (a) and year (b) of death for P. engelmannii (white) and P. pungens (gray). Some trees noted as killed by bark beetles were also defoliated or had high dwarf mistletoe ratings or both.

Mortality of each spruce species was not 2 N, AZ, located at 7340 feet [2237 m] eleva- significantly related to aspect, dbh class, or var- tion in the White Mountains), has been 21% iability, or to a variety of vegetation character less since 1994 than it was during 1980–1994 variables (P > 0.25; Table 2). Dendroctonus (23.2 inches vs. 29.1 inches [59 cm vs. 74 cm]; rufipennis was active in only 8% of the dead NCDC 1995). Wintertime precipitation has trees prior to death (Fig. 3a). changed the most. On average, winters from Interpretation of the results shown here 1994–95 to 2000–01 received 38% less precip- must include recognition of the dry winter itation than did winters from 1980–81 to 1993– conditions that have prevailed in recent years. 94 (8.3 inches vs. 14 inches [21 cm vs. 36 cm]). Mean annual precipitation at McNary, Arizona Autumn E. abietinum outbreaks in the south- (National Climate Data Center Station McNary western mountains may be preceded by dry 14 WESTERN NORTH AMERICAN NATURALIST [Volume 64

TABLE 2. Correlation (r) between 1997 defoliation index and 2000 mortality of P. engelmannii and plot characteristics. Correlations significant at 0.10 and 0.01 are marked by * and **, respectively. Defoliation index Percent mortality Aspect 0.113 0.133 Elevation 0.207 0.006 Mean dbh 0.091 0.098 Mean dbh of spruce 0.063 0.053 s (dbh) 0.213 0.105 s (dbh of spruce) 0.267 0.127 1997 defoliation index (of P. engelmannii) 0.389 * 0.429 ** DMR (Arceuthobium microcarpum) 0.003 0.494 ** Density of All trees 0.390 ** 0.100 Abies concolor –0.053 –0.238 Abies lasiocarpa var. arizonica 0.347 * 0.109 P. pungens –0.205 0.094 P. engelmannii 0.233 –0.182 P. engelmannii + P. pungens –0.001 –0.305 * Pinus ponderosa –0.381* 0.515 ** Pinus strobiformis 0.088 0.320 * Populus tremuloides 0.337 * 0.043 Pseudotsuga menziesii 0.094 0.168 Dominance by Abies concolor 0.264 * –0.066 Abies lasiocarpa var. arizonica 0.270 * –0.024 P. engelmannii 0.064 –0.260 P. pungens –0.242 0.071 P. engelmannii + P. pungens –0.039 –0.205 Pinus ponderosa + P. strobiformus –0.409 * 0.634 ** Populus tremuloides 0.287 * –0.017 Pseudotsuga menziesii –0.246 0.313 * PC 1 (mixed-conifer vs. P. engelmannii) 0.273 0.113 PC 2 (A. lasiocarpa vs. P. engelmannii) 0.321 * 0.067 PC 3 (warm habitats with Pinus) –0.101 0.443 **

winters (Lynch 2003). More importantly, mois- of Arizona incurred 3 outbreaks in 6 years, ture stress undoubtedly contributes to tree each with extensive, serious levels of defolia- mortality. Little moisture is received from tion. Impacts documented in this study indi- April through June, so dry winters mean that cate that this insect will affect natural distur- soil moisture is also very low for the early part bance regimes and tree population dynamics of the growing season. in mixed-conifer and spruce-fir forests. In the hosts’ native habitats, P. engelmannii DISCUSSION is more susceptible to E. abietinum than is P. pungens. Both species can be damaged in Euro- Prior to the 1989–90 outbreak in the White pean maritime habitats (Theobald 1914, Han- Mountains, E. abietinum was known as a pest son 1952), and the North American species of only in areas with mild maritime winter cli- Picea are generally considered to be more sus- mate, primarily on P. sitchensis and P. glauca ceptible than Picea from Europe and Asia (Bejer-Petersen 1962, Carter and Halldórsson (Theobald 1914, Nichols 1987, Jensen et al. 1998). In most areas where E. abietinum is 1997). Trees can withstand light to moderate considered a pest species, both the aphid and amounts of defoliation, but a large proportion the host are exotic (Bevan 1984, Carter and of heavily defoliated trees will die. Elatobium Halldórsson 1998). Population increases and abietinum damage causes more mortality to P. damage occur in the spring in maritime areas, engelmannii in high-elevation habitats than it while outbreaks develop in the fall and winter does on the most important host in Europe, P. in the southwestern U.S., perhaps persisting sitchensis. In maritime habitats E. abietinum through mild winters. The White Mountains impacts height growth, radial growth, and seed 2004] FATE OF PICEA AFTER ELATOBIUM DEFOLIATION 15 and cone crops (Bevan 1966, Carter 1977, Day stands, which would be from even lower ele- and McClean 1991, Seaby and Mowat 1993, vations. Acciavatti and Weiss did not survey P. Thomas and Miller 1994, Straw et al. 1998), pungens and did not evaluate mortality. although mortality to P. sitchensis occurs in Beyond the clear patterns of species differ- New Zealand (Nicol et al. 1998) and British ence in susceptibility to both the insect and Columbia (Koot 1992) and has occurred at least disease agents, increased defoliation in the occasionally in Europe in the past (Theobald lower canopy layers, and strong relationships 1914). between tree mortality and severe mistletoe Defoliation was most severe on the smallest infection and defoliation, risk and hazard pat- trees of each plot and on the lower portions of terns associated with site and vegetation char- the tree crowns, a pattern apparently related acter were indistinct. Patterns of increased de- to canopy position, not to tree size. Retention foliation severity with cooler habitats, increased of foliage in the top crown-third was strongly presence of Abies, and decreased defoliation associated with survivorship, probably because severity with increased presence of Pinus or these trees retain a proportionally large amount warmer habitats were weak, but worthy of con- of the most recently produced foliage. At the sideration, because they were consistent among onset of this study, severe defoliation was the different defoliation episodes and analytic defined as defoliation index ≥7. In future work methods (both individual variables and the severe defoliation should be defined as defoli- relevant principal components were significant). ation index ≥8. Also noteworthy is that about These patterns suggest that aphid populations half of the dead trees did not die until 2 to 3 are greater, or persist longer, in cooler habitats. years after defoliation—a factor that should be Elatobium abietinum requires mature, dormant considered in future damage surveys. foliage for populations to increase to damaging Only one-third of trees both severely defo- levels (Parry 1976, Jackson and Dixon 1996), liated by E. abietinum and heavily infected by and perhaps trees in colder habitats enter dor- A. microcarpum survived. P. pungens was more mancy earlier in the fall. Though weak, the frequently and more severely infected by A. patterns seen here are consistent enough to microcarpum than was P. engelmannii. Mortal- warrant further study. Although a different ity of heavily infected P. pungens was 6 times approach might clarify these relationships, fac- greater than that of noninfected and lightly to tors such as autumn weather and E. abietinum moderately infected trees, as observed by population levels in early autumn are probably Mathiasen et al. (1986). With pine-infesting stronger factors contributing to outbreak sever- Arceuthobium species, mortality is primarily ity. The association between mortality and associated with DMR class 6 (Hawksworth warm, pine-supporting habitats, where defolia- and Lusher 1956). Here, mortality was 19% in tion was less severe, probably reflects greater class 4 and higher. Mathiasen et al. (1986) drought stress on those sites. found 13%, 30%, and 47% mortality in class-4, Picea engelmannii and P. pungens co-occur -5, and -6 trees. They surveyed live and dead in many stands. The greater susceptibility of P. trees, and mortality was not dated nor identi- engelmannii to E. abietinum will probably fied per species. The mortality observed in favor the reproduction of P. pungens in these this study is quite high for a 4-year period, stands. Because P. pungens is much more sus- indicating that defoliation accelerates mortal- ceptible to A. microcarpum, the severity of that ity of dwarf mistletoe–infected trees, and that problem could increase over time. In other dwarf mistletoe–related mortality is acceler- stands P. engelmannii would be replaced by ated in droughts. other species depending on habitat, but pri- While A. microcarpum infection rates were marily by Abies lasiocarpa var. arizonica at much lower on P. engelmannii, Acciavatti and higher elevations and by Pseudotsuga men- Weiss (1974) observed even lower levels (3.4%). ziesii at lower elevations. Acciavatti and Weiss (1974) surveyed the Accumulation of fine fuels from dehydrated entire elevational range of P. engelmannii and and dead needles, dead twigs and branches, seldom found A. microcarpum above 10,000 and dead trees will increase fire hazard. The feet. Data reported here are from plots at 9640 recurrence of even fairly frequent outbreaks feet and lower, and Mathiasen et al. (1986) will result in a near-constant presence of de- stated that their survey was of mixed-conifer hydrated fine fuels distributed throughout the 16 WESTERN NORTH AMERICAN NATURALIST [Volume 64 forest canopy and on the ground. This will fur- I thank Michelle Frank, Maury Williams, Jill ther destabilize ecosystem processes in south- Wilson, and the anonymous reviewers for many western high-elevation forests that have already helpful comments, discussions, and reconnais- been affected by grazing, logging, and fire ex- sance; the numerous people who assisted with clusion, especially with regard to fire regimes data collection; and Joyce Van de Water for in lower-elevation forests and the likelihood of preparation of the map figure. stand-replacement fires in high-elevation forests (Baisan and Swetnam 1990, Swetnam and LITERATURE CITED

Betancourt 1990, Bahre 1991, Grissino-Mayer ACCIAVATTI, R.E., AND M.J. WEISS. 1974. Evaluation of et al. 1995). dwarf mistletoe on Engelmann spruce, Fort Apache Elatobium abietinum outbreak severity varies Indian Reservation, Arizona. Plant Disease Reporter from one episode to another. This study evalu- 58:418–419. BAHRE, C.J. 1991. A legacy of change: historic human ated the damage from a single episode (a few impact on vegetation of the Arizona borderlands. plots may have been defoliated twice). Out- University of Arizona Press, Tucson. break severity may vary from year to year and BAISAN, C.H., AND T.W. S WETNAM. 1990. Fire history on a from place to place, but the general findings of desert mountain range: Rincon Mountain Wilder- ness, Arizona, U.S.A. Canadian Journal of Forest greater susceptibility of P. engelmannii, greater Research 29:1559–1569. defoliation in the lower canopy layers, and BEJER-PETERSEN, B. 1962. Peak years and regulation of high probability of mortality to severely defoli- numbers in the aphid Neomyzaphis abietina Walker. ated or severely A. microcarpum–infected trees Oikos 13:155–168. BEVAN, D. 1966. The green spruce aphis (Elatobium (Neo- or both will probably be reasonably consistent. myzaphis) abietinum Walker. Scottish Forestry 20: If E. abietinum outbreak frequency is low, 193–201. then many trees should be able to survive and ______. 1984. Coping with infestations. Quarterly Journal reproduce. However, 3 of 6 autumn/winter sea- of Forestry 78:36–40. CARTER, C. 1977. Impact of green spruce aphid on growth: sons incurred outbreaks. If outbreaks continue can a tree forget its past? Forestry Commission at this frequency, then this insect will impact Research and Development Paper 116, Edinburgh. natural disturbance regimes and tree popula- CARTER, C., AND G. HALLDÓRSSON. 1998. Origins and back- tion dynamics in mixed-conifer and spruce-fir ground to the green spruce aphid in Europe. Pages 1–10 in K.R. Day, G. Halldórsson, S. Harding, and forests. Large areas were defoliated in each N.A. Straw, editors, The green spruce aphid in west- outbreak, and significant portions of the areas ern Europe: ecology, status, impacts and prospects were defoliated severely. Eighteen percent, for management. Forestry Commission Technical 10%, and 24% of plots assessed in 1997, 2000, Paper 24, Edinburgh. DAY, K., AND S. CRUTE.1990. The abundance of spruce and 2001, respectively, had mean defoliation aphids under the influence of an oceanic climate. ratings of 7.5 or higher. Repeated defoliation Pages 25–33 in A.D. Watt, S.R. Leather, M.D. Hunter, episodes with 10% overall P. engelmannii mor- and N.C. Kidd, editors, Population dynamics of for- tality, 30% in severely defoliated areas, will lead est insects. Intercept Ltd., Andover, Hampshire, England. to reduced representation of P. engelmannii in DAY, K.R., AND S. MCCLEAN. 1991. Influence of the green the ecosystem. More severe defoliation and spruce aphid on defoliation and radial stem growth subsequent mortality in the smaller size classes of Sitka spruce. Annals of Applied Biology 11:415–423. FURNISS, R.L., AND V. M . C AROLIN. 1977. Western forest will mean a gradual reduction of ingrowth with- insects. USDA Forest Service, Miscellaneous Publi- in the forest. Also, repeated defoliation episodes cation 1339. 654 pp. will likely prevent viable seed from being pro- GAUCH, H.G., JR. 1982. Multivariate analysis in commu- duced. If this is the case, P. engelmannii repre- nity ecology. Cambridge University Press, Cambridge. sentation in the forest will diminish over time. 298 pp. GRISSINO-MAYER, H.D., C.H. BAISAN, AND T.W. S WETNAM. 1995. Fire history in the Pinaleño Mountains of south- ACKNOWLEDGMENTS eastern Arizona: effects of human-related disturbances. Pages 399–407 in L.F. DeBano, G.J. Gott- fried, R.H. Research was supported by the USDA For- Hamre, P.E. Ffolliott, and A. Ortega-Rubio, technical coordinators, Biodiversity and management of the est Service, Rocky Mountain Research Sta- Madrean Archipelago: the Sky Islands of southwest- tion. I am grateful to the White Mountain ern Untied States and northwestern Mexico. USDA Apache Tribe, Bureau of Indian Affairs at Fort Forest Service, General Technical Report RM-GTR- Apache Indian Reservation, and USDA Forest 264, Fort Collins, CO. HANSON, H.S. 1952. The green spruce aphid, Neomyza- Service Region 3 Arizona Zone Entomology phis abietina Walker. Report on Forest Research 1951: and Pathology for cooperation and assistance. 98–104. 2004] FATE OF PICEA AFTER ELATOBIUM DEFOLIATION 17

HAWKSWORTH, F.G. 1977. The 6-class dwarf mistletoe rat- NICHOLS, J.S.A. 1987. Damage and performance of the ing system. USDA Forest Service, General Technical green spruce aphid, Elatobium abietinum, on twenty Report RM-48, Fort Collins, CO. 7 pp. spruce species. Entomologia Experimentalis et HAWKSWORTH, F.G., AND A.A. LUSHER. 1956. Dwarf mis- Applicata 45:211–217. tletoe survey and control on the Mescalero-Apache NICHOLS, S. 1977. On the interpretation of principal com- Reservation, New Mexico. Journal of Forestry 54: ponents analysis in ecological contexts. Vegetatio 384–390. 34:191–197. HAWKSWORTH, F.G., AND D. WIENS. 1996. Dwarf mistletoes: NICOL, D., K. ARMSTRONG, S.D. WRATTEN, P.J. WALSH, biology, pathology, and systematics. USDA Forest N.A. STRAW, C.M. CAMERON, C. LAHMANN, AND C.M. Service, Agriculture Handbook 709, Washington, DC. FRAMPTON. 1998. Genetic diversity of an introduced 410 pp. pest, the green spruce aphid Elatobium abietinum HEIE, O.E. 1989. Aphids in Denmark in the spring follow- (Hemiptera: Aphididae) in New Zealand and the ing the mild winter of 1988–1989. Entomologiske United Kingdom. Bulletin of Entomological Research Meddelelser 57(3):173–175. 88:537–543. HUSSEY, N.W. 1952. A contribution to the bionomics of the PARRY, W.H. 1973. Observations on the flight periods of green spruce aphid (Neomyzaphis abietina. Walker). aphid in a Sitka spruce plantation in north-eastern Scottish Forestry 6:121–130. Scotland. Bulletin of Entomological Research 62: ISEBRANDS, J.G., AND T.R. CROW. 1975. Introduction to 391–399. uses and interpretation of principal component analy- ______. 1976. The effect of needle age on the acceptabil- sis in forest biology. USDA Forest Service, General ity of Sitka spruce needles to the aphid, Elatobium Technical Report 83-17, North Central Station, St. abietinum (Walker). Oecologia (Berl.) 23:297–313. Paul, MN. SEABY, D.A., AND D.J. MOWAT. 1993. Growth changes in JACKSON, D.L., AND F. G . D IXON. 1996. Factors determin- 20-year-old Sitka spruce Picea sitchensis after attack ing the distribution of the green spruce aphid, Elato- by the green spruce aphid Elatobium abietinum. bium abietinum, on young and mature needles of Forestry 66:371–379. spruce. Ecological Entomology 21:358–364. SOKAL, R.R., AND F. J . R OHLF. 1995. Biometry: the princi- JACKSON, J.E. 1991. A user’s guide to principal compo- ples and practice of statistics in biological research. nents. John Wiley & Sons, Inc., New York. 569 pp. 3rd edition. W.H. Freeman and Company, New York. JENSEN, J.S., S. HARDING, AND H. ROULUND. 1997. Resis- 887 pp. tance to the green spruce aphid (Elatobium abiet- STRAW, N.A., G. HALLDÓRSSON, AND T. B ENEDIKZ. 1998. inum Walker) in progenies of Sitka spruce (Picea Damage sustained by individual trees: empirical sitchensis (Bong) Carr.). Forest Ecology and Manage- studies on the impact of the green spruce aphid. ment 97:207–214. Pages 15–31 in K.R. Day, G. Halldórsson, S. Harding, KOOT, H.P.1992. Spruce aphid. Forest Pest Leaflet, Forestry and N.A. Straw, editors, The green spruce aphid in Canada, Forest Insect and Disease Survey, Pacific western Europe: ecology, status, impacts and prospects Forestry Centre, Victoria, British Columbia. for management. Forestry Commission Technical KOOT, H.P., AND D.S. RUTH. 1971. Spruce aphid in British Paper 24, Edinburgh. Columbia. Forest Insect and Disease Survey Pest SWETNAM, T.W., AND J.L. BETANCOURT. 1990. Fire–south- Leaflet 16, Department of Fisheries and Forestry, ern oscillation relations in the southwestern United Canadian Forestry Service, Victoria, British Colum- States. Science 249:1017–1020. bia. 5 pp. THEOBALD, F.V. 1914. Notes on the green spruce aphis LUDWIG, J.A., AND J.F. REYNOLDS. 1988. Statistical ecology: (Aphis abietina Walker). Annals of Applied Biology a primer on methods and computing. John Wiley & 1:22–36. Sons, New York. 337 pp. THOMAS, R.S., AND H.G. MILLER. 1994. The interaction of LYNCH, A.M. 2003. Spruce aphid in high elevation habi- green spruce aphid and fertilizer applications on the tats in the Southwest. Pages 60–63 in S.L.C. Fos- growth of Sitka spruce. Forestry 67:329–341. broke, K.W. Gottschalk, editors, Proceedings of the USDA FOREST SERVICE. 1997. Forest insect and disease USDA interagency research forum on gypsy moth conditions in the Southwestern Region, 1996. R3- and other invasive species, 2002. USDA Forest Ser- 97-1, USDA Forest Service, Southwestern Region, vice, General Technical Report NE-300, Northeast- Albuquerque, NM. ern Research Station, Newtown Square, PA. ______. 1998. Forest insect and disease conditions in the LYNCH, A.M., AND R.A. FITZGIBBON. 2004. Observations Southwestern Region, 1997. R3-98-01, USDA Forest on the life history of Nepytia jocnetae in Arizona. Service, Southwestern Region, Albuquerque, NM. Southwestern Entomologist: in press. ______. 1999. Forest insect and disease conditions in the MATHIASEN, R.L., HAWKSWORTH, F.G., AND C.B. EDMINSTER. Southwestern Region, 1998. R3-99-01, USDA Forest 1986. Effects of dwarf mistletoe on spruce in the Service, Southwestern Region, Albuquerque, NM. White Mountains, Arizona. Great Basin Naturalist ______. 2000. Forest insect and disease conditions in the 46:685–689. Southwestern Region, 1999. R3-00-01, USDA Forest MORAN-PALMA, P., AND J.P. MCTAGUE. 1997. Stand dynam- Service, Southwestern Region, Albuquerque, NM. ics of the spruce-fir forest in east-central Arizona. ______. 2001. Forest insect and disease conditions in the Western Journal of Applied Forestry 12(2):55–61. Southwestern Region, 2000. R3-01-01, USDA Forest NATIONAL CLIMATE DATA CENTER (NCDC). 1995. Coop- Service, Southwestern Region, Albuquerque, NM. erative summary of the day, TD3200—period of record through 1993. National Climatic Data Center, Received 4 April 2002 Asheville, NC. Volume 4—Arizona, Nevada. Later Accepted 28 March 2003 records were obtained from the NCDC web site. Western North American Naturalist 64(1), ©2004, pp. 18–26

DENSITY AND BIOMASS OF REDBAND TROUT RELATIVE TO STREAM SHADING AND TEMPERATURE IN SOUTHWESTERN IDAHO

Bruce W. Zoellick1

ABSTRACT.—Density and biomass of redband trout (Oncorhynchus mykiss gairdneri) relative to stream temperature were examined in headwater reaches of Big Jacks and Little Jacks Creeks in southwestern Idaho. Stream shading was greater (mean of 80% versus 46%) and solar insolation was lower (mean of 7.9 versus 15.1 mJ ⋅ m–2 ⋅ day–1) in Little Jacks Creek (P < 0.04); otherwise the 2 streams were similar (e.g., width, depth, gradient, median substrate size). Maximum water temperatures increased with distance from headwater springs in both streams (P ≤ 0.07) but increased more rapidly and to higher levels (24°–26°C) in Big Jacks Creek. Daily maximum water temperatures (23 km downstream of headwater springs) during July 1996 were lower in Little Jacks Creek (ranged from 18° to 22°C) than in Big Jacks Creek (20.2° to 26°C, P < 0.001). Daily temperature fluctuations also differed between streams, averaging 3.6°C for Little Jacks Creek and 7.8°C for Big Jacks Creek (P < 0.001). Redband trout density and biomass were greater in Little Jacks Creek (means of 0.8 fish ⋅ m–2 and 25.0 g ⋅ m–2) compared to Big Jacks Creek (0.3 fish ⋅ m–2 and 8.9 g ⋅ m–2, P = 0.01). Trout density was negatively correlated with increases in water temperature (P = 0.03) and solar insolation (P = 0.09) in both streams. Trout biomass increased with stream shading and was negatively correlated with solar insolation (P < 0.1). Warmer water temperatures in Big Jacks Creek were likely due to historical summerlong livestock grazing, which dras- tically reduced riparian shading.

Key words: redband trout, Oncorhynchus mykiss gairdneri, water temperature, density, biomass, stream shading, solar insolation, desert streams, southwest Idaho.

Platts and Nelson (1989) hypothesized that 1994). However, redband trout stocks in low- summer water temperature increases limit elevation desert streams in the Snake River salmonid populations in open-canopied streams basin in southwestern Idaho and northern in the Great Basin and that deleterious effects Nevada have probably evolved adaptations to of elevated temperatures offset increases in temperature extremes (Behnke 1992). In partic- invertebrate abundance generated from greater ular, redband trout inhabiting tributary streams primary production. Li et al. (1994) tested this to the Snake and Owyhee Rivers tolerate max- hypothesis in streams inhabited by interior imum stream temperatures of 28°–29°C rainbow (redband) trout (Oncorhynchus mykiss (Behnke 1992, Zoellick 1999). Adaptations to gairdneri) in eastern Oregon. Trout biomass extremes in temperature and flow undoubtedly was negatively correlated with solar insolation have allowed populations in desert basins to and maximum stream temperature in streams persist through time even in extreme drought in the John Day River basin, but was not cor- conditions when flows become intermittent related with invertebrate biomass. Near-lethal (Behnke 1992). While increased stream tem- water temperature levels in open-canopied peratures may not limit distribution of desert- streams likely impose high metabolic costs on adapted populations of redband trout to the redband trout, offsetting higher food availabil- extent of other trout stocks, their abundance ity (Li et al. 1994). likely declines with temperature increases be- Persistence of redband trout in warm-tem- cause of increased metabolic costs (Li et al. perature stream reaches in the John Day River 1994). basin did not necessarily require physiological To ascertain whether redband trout popula- adaptations to temperature extremes; trout tions in lower-elevation sagebrush desert basins were thought to behaviorally thermoregulate respond similarly to temperature increases com- by moving to cold-water microhabitats when pared with populations in the John Day River temperatures approached 23°–25°C (Li et al. basin, I studied relationships between trout

1Lower Snake River District, U.S. Bureau of Land Management, 3948 Development Avenue, Boise, ID 83705.

18 2004] REDBAND TROUT ABUNDANCE AND STREAM TEMPERATURE 19 abundance and stream shading, solar insolation, and stream temperature in Big Jacks and Little Jacks Creeks in southwestern Idaho. These streams support redband trout stocks tolerant of extreme fluctuations in temperature (Zoellick 1999). Headwater reaches are physically similar (elevation, geomorphology, stream flows, and channel types) with the exception of the amount of stream shading provided by riparian shrubs. I hypothesized fewer trout were present in Big Jacks Creek per unit area than in Little Jacks Creek because of elevated stream temperatures due to lower levels of stream shading in Big Jacks Creek. Objectives were to (1) compare redband trout abundance (density and biomass) in reaches of Big Jacks and Little Jacks Creeks that differed only in amount of stream shading from riparian shrubs, (2) quantify solar insolation and water temperatures of the 2 streams, and (3) relate abundance of redband trout in the 2 streams to stream temperature, solar insolation, and stream shading.

STUDY SITE Fig. 1. Location of Big Jacks and Little Jacks Creeks in Big Jacks and Little Jacks Creeks flow southwestern Idaho, their watershed boundaries, study northeasterly from the Owyhee Mountains to reaches (shaded), fish sample sites, and area map (inset). C.J. Strike Reservoir on the Snake River near the town of Bruneau in southwestern Idaho (Fig. 1). Drainage basins of Big Jacks and Lit- through 1995, resulting in annual removal of tle Jacks Creeks are 633 km2 and 260 km2, 80%–100% of the current year’s growth of respectively. Elevations range from 750 m to young shrubs and herbaceous vegetation. Mean 1920 m, and the basins are predominantly vertical cover of young shrubs (<2 m tall) was vegetated with big sagebrush (Artemisia tri- unchanged from 1984 to 1994; median residual dentata) shrub-steppe communities. The upper stubble heights of herbaceous vegetation ranged perennial segments of the 2 streams are located from 2.5 cm to 3.8 cm during 1993–1995 (U.S. in canyons 30–270 m deep, carved through ry- Bureau of Land Management unpublished data). holite lava, with narrow floodplains and stream Approximately one-half or more of the ripar- substrates dominated by cobble-sized rocks. ian shrubs on Big Jacks Creek were replaced Flows in both streams were intermittent at with herbaceous vegetation because of live- downstream ends of the watersheds (Zoellick stock eliminating or preventing the reestab- 1999; Fig. 1), and surface flows were not con- lishment of shrubs after their removal by nected during 1995–96. Stream channels were beaver (Castor canadensis) or scouring flows. moderately confined by side valley slopes and Upstream, steeper gradient (2%–4%) seg- had gradients of 1.5%–4% (B stream types; ments of both study reaches were dominated Rosgen 1994). I conducted the study on the by red-twig dogwood (Cornus sericea). Lower upper 23 km of each stream (Fig. 1), starting at segments (>11 km downstream of headwater the headwater springs. springs) of the Little Jacks Creek study reach Livestock grazing has been excluded from were predominantly vegetated with arroyo wil- Little Jacks Creek since 1976, and stream low (Salix lasiolepsis), and other willow species banks were densely vegetated with riparian (S. lasiandra, S. exigua, S. amygdaloides) were shrubs. In contrast, Big Jacks Creek was grazed present. Middle to lower segments (>8 km summerlong by cattle from at least the 1970s downstream of headwater springs) of Big Jacks 20 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Creek were vegetated with remnant stands of sites at the lower ends of study segments was willows (primarily S. lasiolepsis and S. exigua) not as rigorous a statistical approach as ran- and herbaceous plant communities dominated domly selecting sample sites within study by Kentucky bluegrass (Poa pratensis), golden- reaches, but it allowed for maximizing the rod (Solidago spp.), and scouring rush (Equise- effect of upstream riparian canopy on water tum arvense). temperature (Li et al. 1994). Fish sample sites were 61–82 m long and METHODS were composed of multiple habitat units (pools, runs, and riffles). I sampled Little Jacks The 2 study reaches were stratified into 5 Creek in August of both 1995 and 1996 (3 sites segments each (3–8 km long), based on stream each year) and Big Jacks Creek in August– gradient and composition and canopy cover of September 1996. Trout were captured during riparian plant communities determined from 2 to 3 electrofishing passes, and population 7.5-minute topographic maps and 1:24,000 sizes were estimated using the Zippin capture- scale color aerial photographs. I measured can- removal model (Zippin 1958). All trout were yon width, depth, steepness of side slopes, weighed. and valley bottom widths on 7.5-minute topo- Stream (wetted channel) width and average graphic maps at 1 randomly selected site per depth were measured at fish sample sites on segment. Canyon aspect was calculated as the 10 cross-section transects located 6.1 m apart. compass bearing of the line connecting up- I calculated average depth for each cross sec- stream and downstream ends of study reaches. tion transect by the method of Overton et al. Average stream gradient for each segment was (1997). Individual transect widths and depths calculated from topographic maps. In August were treated as subsamples, and means of the 1995 and August 1996, I walked the length of widths and depths for the 10 transects were each segment to verify that stream types (Ros- the sample units. Stream flows of Little Jacks gen 1994) did not change within a segment. I Creek were similar between years; base flows then measured stream (bankfull) and floodplain measured at the lower end of the study reach dimensions at 1 reference site per segment to in September were 0.07 m3 ⋅ s–1 in 1995 and classify the stream type of each segment (Ros- 0.09 m3 ⋅ s–1 in 1996 (Zoellick 1999). gen 1996). Wolman (1954) pebble counts were used to Redband trout density (number ⋅ m–2) and sample substrate composition at fish sample biomass (g ⋅ m–2) were estimated with an elec- sites. Ten pebbles were sampled on each of 10 trofisher at 1 site per stream segment. An ex- cross-section transects that were located 6.1 m ception was the downstream-most segment of apart, for a total of 100 pebbles per site. I cal- the Big Jacks Creek study reach because ini- culated the median (50th) particle size for tial stratification indicated the gradient of this each sample site. segment (0.01) was lower than other segments. A Solar Pathfinder (Platts and Nelson 1989, Therefore, I compensated by sampling fish Li et al. 1994) was used to measure solar inso- abundance at 1 randomly selected site in Little lation where fish were sampled. For the down- Jacks Creek study reach and 2 randomly stream-most segment of Big Jacks Creek, selected sites in Big Jacks Creek study reach, which did not have a fish sample site, solar so that 6 sites were sampled per study reach. insolation was measured at the downstream Not sampling fish abundance in the down- end of the segment. The Solar Pathfinder stream-most segment (and thus the segment identifies the amount of solar insolation inter- with likely the warmest stream temperatures) cepted by local shade-producing objects of Big Jacks Creek probably decreased the (streamside shrubs and trees, canyon walls, power to detect differences in fish abundance etc.) and estimates the average daily thermal relative to temperature, but ensured all fish input falling on the stream surface for each sites were located in similar stream types. month of the year by integrating the effects of Most systematically placed fish sites were azimuth, topographic altitude, height of vege- located at the downstream ends of each study tation, aspect, latitude, hour angle, and time of segment. Access was limited in the upper half year (Platts et al. 1987, Tait et al. 1994). I cal- of each canyon, and sample sites for 1 segment culated the mean percent of solar input unim- of each study reach were established in the peded by shade from riparian shrubs and middle of those segments. Selecting sample topographic features (i.e., canyon walls) for 10 2004] REDBAND TROUT ABUNDANCE AND STREAM TEMPERATURE 21

TABLE 1. Watershed characteristics and geomorphology of Big Jacks and Little Jacks Creeks study reaches, southwestern Idaho. Feature Little Jacks Creek Big Jacks Creek Geologic parent material rhyolite lava rhyolite lava Elevation of headwater spring (m) 1673 1670 Elevation of downstream end of reach (m) 1079 1286 Aspect of canyon (degrees) 41 70 Mean canyon widtha (m) 974 ± 111 490 ± 77 Mean canyon deptha (m) 271 ± 14 152 ± 14 Canyon side slopes (%) >60 >60 Mean width of valley bottoma (m) 19 ± 4 24 ± 7 a ± Mean sx–, n = 5 for each study reach.

points (starting at the downstream end of a streams to examine correlations among stream fish site and spaced 6.1 m apart up the stream shading, solar insolation, water temperature, and at the center of the channel) per site, and and trout abundance. Measurements of maxi- by determining the number of cloud-free days mum temperature at fish sample sites in 1997 each month for the study area (calculated for were used as indices of stream temperatures the nearby city, Boise, ID; Platts et al. 1987). during 1995–96. Correlations were also used to The height of the solar pathfinder above the examine change in maximum stream tempera- stream surface was standardized at 0.3 m. To ture with distance from headwater springs. avoid pseudoreplication, insolation levels at each of the 10 points per site were treated as RESULTS subsamples. I calculated insolation as mega- Geomorphology, Shading, and joules per square meter per day, averaged for Solar Insolation the months of June through September, and calculated the percentage of insolation shaded Watershed characteristics and geomorphol- by riparian shrubs and topographic features of ogy of the 2 streams were similar with the each site for those months. For stream seg- exception of some differences in canyon size ments with 2 fish sample sites, insolation for and aspect (Table 1). All segments of the study that segment was estimated from the down- reaches comprised B stream types (Rosgen stream-most fish site. 1994), with cobble-dominated substrates (Table Temperature recorders (Stowaways; Onset, 2). Diameters of 50th particle size of substrate Inc.) were placed in the streams at the lower materials were similar between Big Jacks and end of the 2 study reaches and within 100 m Little Jacks Creeks (Table 2). Bankfull channel of headwater springs in late June 1996, which dimensions and stream gradients were similar monitored water temperatures through Sep- between the 2 study reaches. Additionally, tember. Water temperatures were recorded wetted channel widths and depths were simi- every 1.6–2 hours. In June 1997, I placed max- lar between streams (Table 2). Little Jacks imum-registering thermometers at each of the Creek canyon was deeper, but wider, than that 1995 and 1996 fish sample sites, and also placed of Big Jacks Creek (Table 1). Consequently, temperature recorders in headwater springs stream shading from topographic features (pri- and at the lower end of the 2 study reaches. marily canyon walls) did not differ between Stream temperatures were monitored through the 2 streams (Table 3). September 1997; the maximum thermometers Stream shading and insolation levels were were read once (at the end of the monitoring similar for the upstream, most confined, steep- period). gradient segments of the study reaches (Fig. 2). Differences between study reaches in red- Stream shading from riparian shrubs increased band trout abundance (density and biomass), in the less confined, downstream segments of physical habitat parameters, percent shading, the Little Jacks Creek reach, while shading stream temperatures, and solar insolation lev- decreased and insolation greatly increased in els were examined using t tests and analysis of the lower segments of the Big Jacks study reach. variance. I combined data sets from both Solar insolation was significantly greater and 22 WESTERN NORTH AMERICAN NATURALIST [Volume 64

TABLE 2. Stream channel and floodplain measurements of Big Jacks and Little Jacks Creeks study reaches, southwestern Idaho, 1996.

______Big Jacks Creek ______Little Jacks Creek – – a Channel dimension x sx– x sx– n P-value

BANKFULL CHANNEL Width (m) 6.5 0.7 6.8 0.6 5 0.73 Maximum depth (m) 0.52 0.04 0.54 0.06 5 0.76 Average depth (m) 0.27 0.02 0.27 0.03 5 0.96 Entrenchment ratiob 1.8 0.09 1.8 0.2 5 0.94 Width/depth ratio 23.8 1.8 26.0 3.1 5 0.54 Flood-prone widthb (m) 11.9 1.5 11.8 1.2 5 0.95 WETTED CHANNEL Stream width (m) 4.2 0.2 4.2 0.5 6 0.95 Stream depth (m) 0.16 0.02 0.12 0.01 5 0.14 Diameter 50th particle (mm) 62 10.5 67 2.9 6 0.26 GRADIENTc 0.017 0.003 0.021 0.002 5 0.37 aSample size per study reach; t tests were used to examine for differences between means. bRosgen (1994) channel classification system; all segments of the 2 study reaches were classified as B channel types. cGradients of segments of study reaches were calculated from USGS 7.5-minute topographic maps.

TABLE 3. Insolation (mJ ⋅ m–2 ⋅ day–1), percent stream shade (total from topographic features and vegetation), and shading from deciduous vegetation and topographic features for study reaches of Big Jacks and Little Jacks Creeks, southwestern Idaho, 1995–96.

______Big Jacks Creek ______Little Jacks Creek – – a Parameter x sx– x sx– n P-value Insolation 15.1 2.4 7.9 0.9 5 0.04 Stream shading 46.4 10.9 80.4 4.5 5 0.03 Shading from deciduous vegetation 26.7 7.9 57.0 11.4 5 0.06 Shading from topographic features 19.7 4.6 23.4 9.1 5 0.73 aNumber of sites (1 per segment of each study reach) where insolation and shading were measured; t tests were used to examine for differences between means.

total shade and deciduous vegetation shade heating is shown by maximum temperatures were lower in the Big Jacks study reach com- occasionally converging for the 2 streams, prob- pared with Little Jacks Creek (Table 3). ably during days with cloud cover or thunderstorms. Streamside vegetation on Little Jacks Stream Temperature Creek also apparently buffered drops in mini- Stream temperatures measured within 100 mum temperatures by intercepting heat re- m of headwater springs averaged 11.4 ± radiating from the stream surface at night. Daily ° ± ± ° 0.08 C ( sx–, n = 300) and 11.4 0.02 C (n = minimum temperatures in Little Jacks Creek 516) during July 1996 in Little Jacks and Big were slightly higher than those for Big Jacks Jacks Creeks, respectively. Maximum temper- Creek and remained higher than those in Big atures of 10°C were measured at Big Jacks Jacks Creek during a drop in overall tempera- and Little Jacks Creeks headwater springs in tures over several days (Fig. 3). 1997. Maximum stream temperatures and Maximum temperatures increased with dis- temperature fluctuations at the lower ends of tance from headwater springs for both Big the study reaches in 1996 were significantly Jacks (r = 0.67, n = 8, P = 0.07) and Little greater in Big Jacks Creek (Table 4). Daily max- Jacks (r = 0.77, n = 7, P = 0.03) Creeks, but imum temperatures in Big Jacks Creek were increased more quickly and to greater temper- consistently 2°–4°C higher than Little Jacks atures in Big Jacks Creek. In 1997 maximum Creek temperatures (Fig. 3). The effect of solar temperatures observed at fish sample sites in 2004] REDBAND TROUT ABUNDANCE AND STREAM TEMPERATURE 23

Fig. 2. Stream shading (bars) and solar insolation (lines) Fig. 3. Daily maximum and minimum stream tempera- relative to distance from headwater springs for Big Jacks tures at the downstream end of the Big Jacks Creek (solid Creek (solid bar, line) and Little Jacks Creek (hatched bar, lines) and Little Jacks Creek (dashed lines) study reaches, dashed line), southwestern Idaho, 1995–96. southwestern Idaho, July–August 1996.

Little Jacks Creek did not exceed 22°C, while maximum temperatures in Big Jacks Creek increased with stream shading (r = 0.68, P = increased to 24.5°C by 8.3 km from headwater 0.09). Trout biomass decreased with maximum springs and remained at or above 24°C to the stream temperature, but the correlation was lower end of the study reach. Mean daily max- not significant (r = –0.53, P = 0.47). imum temperatures in July 1997 at the lower end of Big Jacks (21.8°C) and Little Jacks DISCUSSION (18.8°C) study reaches were about 2°C lower than in 1996 (F1,108 = 58.88, P < 0.001). Platts and Nelson (1989) showed that salmonid biomass in desert streams in the Trout Abundance Great Basin is negatively related to solar inso- Redband trout density was significantly lation levels. They hypothesized that trout greater in Little Jacks Creek (0.8 ± 0.1 fish ⋅ abundance in desert streams is limited by the m–2, n = 6) than in Big Jacks Creek (0.3 ± 0.1 deleterious effects of increased stream tem- fish ⋅ m–2, n = 6; t = 3.16, P = 0.01). Density peratures and temperature fluctuations. In this ranged from 0.6 to 1.3 fish ⋅ m–2 in Little Jacks study density and biomass of redband trout Creek and 0.1 to 0.8 fish ⋅ m–2 in Big Jacks were significantly greater in a well-shaded Creek. Trout biomass was also greater in Little stream than in a stream with significantly higher Jacks Creek (25.0 ± 1.9 g ⋅ m–2, n = 6) than in maximum stream temperatures and insolation Big Jacks Creek (8.9 ± 1.8 g ⋅ m–2, n = 6; t = levels. Trout density was negatively correlated 6.22, P < 0.001). Biomass ranged from 18.0 to with increases in maximum stream tempera- 30.7 g ⋅ m–2 in Little Jacks Creek and 6.1 to ture and solar insolation in both streams. Bio- 17.7 g ⋅ m–2 in Big Jacks Creek. mass also declined with increasing tempera- Trout density in Little Jacks and Big Jacks ture, but the relationship was not significant Creeks was negatively correlated with maxi- because several Big Jacks Creek sites had mum stream temperature (r = –0.76, P = lower trout biomass than expected. Density 0.03; Fig. 4) and solar insolation (r = –0.68, n and biomass were greatest in a stream reach = 12, P = 0.09), and increased with stream with 80% stream shading and maximum tem- shading, but the correlation was not significant peratures ≤22°C. Similarly, Li et al. (1994) found (r = 0.62, n = 12, P = 0.19). Redband trout that high-desert streams in eastern Oregon biomass was negatively correlated with solar with greater riparian canopy had higher stand- insolation (r = –0.71, n = 12, P = 0.06) and ing crops of interior rainbow (redband) trout 24 WESTERN NORTH AMERICAN NATURALIST [Volume 64

TABLE 4. Stream temperatures and daily temperature fluctuations (°C) at the lower end of the Big Jacks and Little Jacks Creeks study reaches, southwestern Idaho, July 1996. Parameter Big Jacks Creek Little Jacks Creek na P-value Maximum temperature 26.0 22.0 1 — Maximum daily fluctuation 10.7 5.0 1 — ± ± ± Average maximum ( sx–) 23.7 0.3 20.6 0.2 28 <0.001 Average daily fluctuation 8.3 ± 0.3 4.0 ± 0.2 28 <0.001 Average minimum 15.3 ± 0.3 16.6 ± 0.2 28 <0.001 aSample size per stream; t tests were used to examine for differences between means.

and lower daily maximum temperatures (range 16°–23°C compared with 26°–31°C). In this study riparian shrub cover on Big Jacks Creek was removed by historical summerlong cattle grazing. Livestock grazing impacts on instream habitat would add to the negative impact of elevated stream temperatures on trout. However, because the streams were dominated by cobble substrates and had confined flood plains with most pools formed by lateral scouring at bedrock or boulders, livestock impacts to trout habitat other than changes in vegetation composition were minor. Stream banks and channels were stable as evi- denced by similar channel shape and form and median substrate particle size between grazed and ungrazed streams. One possible difference between habitats of the 2 streams was Fig. 4. Correlation of redband trout density with maxi- the amount of near-bank vegetative cover that mum daily temperature for all reaches of Big Jacks and Little Jacks Creeks combined (maximum daily tempera- provided feeding and security cover for trout, tures were measured in 1997). but this difference was unlikely the major causative factor for the almost threefold difference in trout density and biomass between the shaded and open-canopy (grazed) streams. approached 23°–25°C by moving to coldwater Redband trout populations in this study microhabitats, and that persistence of redband responded to temperature increases similarly trout in warm-temperature stream reaches in to populations in the John Day River basin in the John Day River basin did not necessarily eastern Oregon, despite the fact that redband require physiological adaptations to tempera- trout from this study (inhabiting sagebrush ture extremes. desert basins) are thought to have evolved In desert streams in the John Day River adaptations to temperature extremes (Behnke basin, primary productivity increased in open- 1992). Redband trout stocks in both streams in canopy reaches as did invertebrate abundance this study have been documented to survive (Li et al. 1994, Tait et al. 1994). However, trout short-term exposure to maximum tempera- abundance was not correlated with increased tures of 29°C, temperatures of >26°C for up invertebrate abundance. In these open-canopy to 4.4 hours, and daily temperature fluctua- streams, maximum temperatures were ele- tions of up to 11°C (Zoellick 1999). Behnke vated to 26°–31°C, imposing higher metabolic (1992) observed redband trout foraging in a costs on trout than could be offset by increases pool with no flow at a temperature of 28.3°C in food supply (Li et al. 1994). Elevated stream in northern Nevada. In contrast, Li et al. (1994) temperatures also may have affected the avail- thought trout in the John Day basin behav- ability of prey. Tait et al. (1994) found open- iorally thermoregulated when temperatures canopy reaches of streams in eastern Oregon 2004] REDBAND TROUT ABUNDANCE AND STREAM TEMPERATURE 25 supported greater periphyton abundance and These species were absent from the cooler, invertebrate biomass, but much of the increased closed-canopy study reach of Little Jacks Creek invertebrate biomass was that of a large-bod- but were present in warmer, downstream ied caddisfly (Dicosmoecus), which was less reaches of Little Jacks Creek. available to and infrequently eaten by trout Implications for Salmonid Management and other small fish. in Desert Streams In contrast, removal of riparian canopy on streams in the Cascade and Coast Ranges of To maintain and ultimately improve habitat the Pacific Northwest has been associated with for trout in the Great Basin and desert por- increases in salmonid abundance (Murphy et al. tions of the Columbia Basin, streams should 1981, Murphy and Hall 1981, Hawkins et al. be managed to preserve riparian shrubs and 1982, 1983). In these nondesert streams, canopy trees and increase their canopy cover to pro- removal increased primary productivity and vide suitable stream temperatures for salmon- biomass of invertebrates without elevating ids. Brook trout (Salvelinus fontinalis) in south- maximum stream temperatures above 22°C ern Ontario (Barton et al. 1985), desert-adapted (Murphy et al. 1981, Hawkins et al. 1982, 1983). redband trout from this study, and redband Similarly, removal of canopy cover on small trout from the south central portion of the streams in southeast Alaska increased primary Columbia Basin (Li et al. 1994, Tait et al. 1994) productivity along with invertebrate and declined in abundance with increasing stream salmonid biomass (Hetrick et al. 1998a, 1998b, temperatures resulting from open canopies. Keith et al. 1998). However, during periods of Probable causes were higher metabolic costs low stream flows and sunny weather, increased imposed by temperature elevations, competi- solar input associated with open-canopy reaches tion with warm-water fishes, and changes in of southeastern Alaska streams was predicted prey availability (Li et al. 1994, Tait et al. 1994). to increase stream temperatures beyond the The importance of maintaining riparian optimum for growth of juvenile coho salmon shading of desert streams is further illustrated (Oncorhynchus kisutch; Hetrick et al. 1998b). by heating and cooling of streams relative to Thus, the effect of riparian canopy (removal or amount of riparian canopy. Keith et al. (1998) restoration) on trout production must be ex- demonstrated that even relatively short (20–76 amined relative to the stream temperatures m long) open-canopy reaches can substantially that are attained. increase water temperatures (up to 6°C) in Decreased abundance of redband trout in small streams in southeast Alaska (average warmer stream reaches in desert basins may widths of 1.6–2.8 m). Presence of open-canopy be due in part to competition with warm-water reaches does not necessarily result in elevated fishes. Reeves et al. (1987) found that redside stream temperatures through all downstream shiner (Richardsonius balteatus) and juvenile reaches. Water warmed by solar input in open- steelhead trout (Oncorhynchus mykiss) com- canopy reaches can cool when it flows through peted for habitat and warm water favored red- downstream reaches with closed canopies (Li side shiners. Production of trout decreased in et al. 1994, Hetrick et al. 1998b, Keith et al. warm water (19°–22°C) when redside shiners 1998). Cooling (when air temperatures exceed were present. Similarly, Baltz et al. (1982) stream temperatures) occurs when heat loss to showed that competition for cover in riffles hyporheic or groundwater exchange exceeds between speckled dace and riffle sculpin (Cot- heat gained from low insolation in closed- tus gulosus, a cold-water species) was medi- canopy reaches. ated by temperature. Tait et al. (1994) found Idaho water quality regulations designated that warm-water fishes (cyprinids and suckers) to protect cold-water aquatic life prescribe increased in abundance in warmer, unshaded that water temperatures not exceed 22°C, with reaches of streams in the John Day River a maximum daily average of ≤19°C (Idaho basin. Similarly, redside shiner, speckled dace Department of Environmental Quality 2000). (Rhinichthys osculus), and bridgelip suckers During 1996–97, water temperatures in Little (Catostomus columbianus) were common in Jacks Creek met these criteria, while they the warmer, open-canopy reach of Big Jacks were not met in the lower reaches of Big Jacks Creek that was examined in this study (U.S. Creek. Differences in temperature and trout Bureau of Land Management unpublished data). abundance between the 2 streams indicate 26 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Idaho’s cold-water criteria were appropriate salmonids to riparian and instream cover modifica- for protecting redband trout populations. tions in small streams flowing through second- growth forests of southeast Alaska. Transactions of the American Fisheries Society 127:889–907. ACKNOWLEDGMENTS LI, H.W., G.A. LAMBERTI, T.N. PEARSONS, C.K. TAIT, J.L. LI, AND J.C. BUCKHOUSE. 1994. Cumulative effects J. Nelson, D. Kearns, M. Rasmussen, T. Koch, of riparian disturbances along high desert trout and S. Duke assisted with fish and habitat streams of the John Day basin, Oregon. Transactions of the American Fisheries Society 123:627–640. sampling. M. McCoy’s assistance in preparing MURPHY, M.L., AND J.D. HALL. 1981. Varied effects of the figures is greatly appreciated. I thank H.W. clear-cut logging on predators and their habitat in Li, W.S. Platts, and J.E. Williams for their small streams of the Cascade Mountains, Oregon. comments on earlier drafts of the manuscript. Canadian Journal of Fisheries and Aquatic Sciences 38:137–145. MURPHY, M.L., C.P. HAWKINS, AND N.H. ANDERSON.1981. LITERATURE CITED Effects of canopy modifications and accumulated sediment on stream communities. Transactions of BALTZ, D.M., P.B. MOYLE, AND N.J. KNIGHT. 1982. Com- the American Fisheries Society 110:469–478. petitive interactions between benthic stream fishes, OVERTON, C.K., S.P. WOLLRAB, B.C. ROBERTS, AND M.A. riffle sculpin, Cottus gulosus, and speckled dace, RADKO. 1997. R1/R4 (Northern/Intermountain Rhinichthys osculus. Canadian Journal of Aquatic Regions) fish and fish habitat standard inventory Science 39:1502–1511. procedures handbook. USDA Forest Service, Inter- BARTON, D.R., W.D. TAYLOR, AND R.M. BIETTE. 1985. mountain Research Station, General Technical Dimensions of riparian buffer strips required to Report INT-GTR-346, Ogden, UT. 73 pp. maintain trout habitat in southern Ontario streams. PLATTS, W.S., C. ARMOUR, G.D. BOOTH, D. GORDON, M. North American Journal of Fisheries Management BRYANT, J.L. BUFORD, P. CUPLIN, ET AL. 1987. Meth- 5:364–378. ods for evaluating riparian habitats with applications BEHNKE, R.J. 1992. Native trout of western North Amer- to management. USDA Forest Service, Intermoun- ica. American Fisheries Society Monograph 6. 275 tain Research Station, General Technical Report pp. INT-221, Ogden, UT. HAWKINS, C.P., M.L. MURPHY, AND N.H. ANDERSON. 1982. PLATTS, W.S., AND R.L. NELSON. 1989. Stream canopy and Effects of canopy, substrate composition, and gradi- its relationship to salmonid biomass in the inter- ent on the structure of macroinvertebrate communi- mountain west. North American Journal of Fisheries ties in the Cascade Range streams of Oregon. Ecol- Management 9:46–457. ogy 63:1840–1856. REEVES, G.H., F.H. EVEREST, AND J.D. HALL. 1987. Inter- HAWKINS, C.P., M.L. MURPHY, N.H. ANDERSON, AND M.A. actions between the redside shiner (Richardsonius WILZBACH. 1983. Density of fish and salamanders in balteatus) and the steelhead trout (Salmo gairdneri) relation to riparian canopy and physical habitat in in western Oregon: the influence of water tempera- streams of the northwestern United States. Canadian ture. Canadian Journal of Fisheries and Aquatic Sci- Journal of Fisheries and Aquatic Sciences 40: ences 44:1603–1613. 1173–1185. ROSGEN, D.L. 1994. A classification of natural rivers. HETRICK, N.J., M.A. BRUSVEN, T.C. BJORNN, R.M. KEITH, Catena 22:169–199. AND W.R. MEEHAN. 1998a. Effects of canopy re- ______. 1996. Applied river morphology. Wildland Hy- moval on invertebrates and diet of juvenile coho drology, Pagosa Springs, CO. salmon in a small stream in southeast Alaska. Trans- TAIT, C.K, J.L. LI, G.A. LAMBERTI, T.N. PEARSONS, AND actions of the American Fisheries Society 127: H.W. LI. 1994. Relationships between riparian cover 876–888. and the community structure of high desert streams. HETRICK, N.J., M.A. BRUSVEN, W.R. MEEHAN, AND T.C. Journal of the North American Benthological Society BJORNN. 1998b. Changes in solar input, water tem- 13:45–56. perature, periphyton accumulation, and allochtho- WOLMAN, M.G. 1954. A method of sampling coarse river- nous input and storage after canopy removal along bed material. Transactions of the American Geo- two small salmon streams in southeast Alaska. Trans- physical Union 35:951–956. actions of the American Fisheries Society 127: ZIPPIN, C. 1958. The removal method of population esti- 859–875. mation. Journal of Wildlife Management 22:82–90. IDAHO DEPARTMENT OF ENVIRONMENTAL QUALITY. 2000. ZOELLICK, B.W. 1999. Stream temperatures and the eleva- Idaho administrative procedures act 58.01.02, water tional distribution of redband trout in southwestern quality standards and wastewater treatment require- Idaho. Great Basin Naturalist 59:136–143. ments. Boise. KEITH, R.M., T.C. BJORNN, W.R. MEEHAN, N.J. HETRICK, Received 13 March 2002 AND M.A. BRUSVEN. 1998. Response of juvenile Accepted 6 February 2003 Western North American Naturalist 64(1), ©2004, pp. 27–37

RESPONSE OF TREE RING HOLOCELLULOSE δ13C TO MOISTURE AVAILABILITY IN POPULUS FREMONTII AT PERENNIAL AND INTERMITTENT STREAM REACHES

Daniel L. Potts1,2 and David G. Williams1,3

ABSTRACT.—We measured δ13C of tree ring holocellulose to assess intra- and interannual variation in integrated leaf gas exchange responses of Frémont cottonwood (Populus fremontii) to monsoonal moisture inputs in southeastern Arizona. We predicted that δ13C of trees growing along drought-susceptible intermittent reaches of this semiarid river system would be more responsive to monsoonal moisture inputs than trees found along perennial reaches, where groundwater is consistently available. We sampled stem xylem cores from 7 trees, each at an intermittent and perennial reach of the San Pedro River near Tombstone, Arizona. We identified and subdivided individual rings from 1990 to 2000. δ13C of holocellulose from these subdivisions was compared with precipitation amount, atmospheric vapor pressure deficit (Da), and δ13 90% exceedence flows (Q90) calculated from seasonal flow duration data. C values were higher at the intermittent reach than at the perennial reach. Furthermore, annual ring δ13C values at the perennial reach were not correlated with δ13 stream flow, precipitation, or Da. C values for trees at the intermittent reach were negatively correlated with mon- 2 2 soon season (1 July–15 September) Q90 (r = 0.50, P = 0.015) and positively correlated with Da (r = 0.45, P = 0.03). Shifts in δ13C between the inner- and outer-third of the annual ring were used as a measure of intra-annual variation. 2 2 These shifts were correlated with monsoon season Da (r = 0.57, P = 0.01) and Q90 (r = 0.59, P = 0.005) for trees growing along the intermittent reach. Intra- and interannual variation in integrated photosynthetic response exists at the population-scale for these native, riparian forests. Changes in monsoonal precipitation and stream flow may differentially alter photosynthetic gas exchange of P. fremontii and function of these riparian ecosystems.

Key words: Populus fremontii, carbon isotopes, North American monsoon, riparian ecosystems.

Hydrologic processes operating at local, et al. 1996), and groundwater depths can fluc- basin, and regional scales control the composi- tuate dramatically. Riparian ecosystems in arid tion and function of riparian forests in arid and and semiarid settings offer unparalleled oppor- semiarid regions (Stromberg 1993, Lines 1999, tunity to study linkages between climate, hydrol- Rood and Mahoney 2000, Shafroth et al. 2000). ogy, and plant-water relations with important Native riparian ecosystems are threatened in management implications. the semiarid southwestern United States by Interpopulation variability in transpiration groundwater pumping, land-use intensification, and water source use has been observed for and surface water diversions (Stromberg 1993). native Frémont cottonwood (Populus fremontii Alterations to the hydrologic regime such as Wats.) forests along southern Arizona’s San flood attenuation and water withdrawals induce Pedro River (Schaeffer et al. 2000, Snyder and structural and functional changes in native Williams 2000). However, these studies focused riparian forests (Stromberg et al. 1996) and only on a single growing season. Thus, the facilitate replacement by exotic species such range of interannual variability in P. fremontii as Tamarix spp. (Stromberg 1998). Water avail- water relations and its linkage to climate dy- ability to riparian trees in the southwestern namics in the upper San Pedro River basin have U.S. is extremely variable due to seasonal and not been studied. Such variability has implica- interannual drought and heterogeneous hydro- tions for catchment-scale hydrologic modeling geomorphic conditions that characterize drain- and riparian forest management (Goodrich et al. ages in these basins. Obligate riparian tree 2000). Refining the role of riparian vegetation species in these systems occur only where allu- in catchment-scale hydrologic models will be- vial groundwater depths are <3 m (Stromberg come increasingly important. Regional climate

1University of Arizona, School of Renewable Natural Resources, Tucson, AZ 85721. 2Corresponding author. Present address: Department of Ecology and Evolutionary Biology, University of Arizona, BioSciences West Rm. 431, Tucson, AZ 85721. 3Present address: Department of Renewable Resources, University of Wyoming, Laramie, WY 82070.

27 28 WESTERN NORTH AMERICAN NATURALIST [Volume 64 models predict changes in summer precipita- Carbon isotope ratios of tree ring holocellu- tion and temperature (Doherty and Mearns lose provide a record of plant photosynthetic 1999, Mearns et al. 1999), and continued valley responses to environmental variation (Francey groundwater pumping in the coming decades and Farquhar 1982, McNulty and Swank 1995, will place increased demands on limited water Livingston and Spittlehouse 1996, Walcroft et resources in the arid and semiarid Southwest. al. 1997, Brooks et al. 1998, Leffler and Evans Cottonwood water use has been explored at 1999, Waterhouse et al. 2000). Francey and a variety of spatio-temporal scales (Leffler and Farquhar (1982) provide a widely accepted Evans 1999, 2001, Sparks and Black 1999, model of carbon isotope variation in photosyn- Schaeffer et al. 2000, Snyder and Williams 2000). thetically fixed carbon. The ratio of 13C to 12C δ13 Using stable isotopes, Snyder and Williams in photosynthate ( Cp) is a function of the (2000) found that uptake of summer precipita- isotope ratio of the air surrounding the leaf δ13 tion by P. fremontii was at least in part a function ( Ca), the ratio of leaf internal to ambient of alluvial groundwater availability. During the CO2 concentration (ci/ca), the diffusive frac- 13 summer rainy season in southern Arizona (July– tionation of CO2 in air (a = 4.4‰), and the September), cottonwood trees growing along fractionation by Rubisco (b = 29‰) in the an ephemeral tributary where groundwater form: depth exceeded 4 m derived a substantial por- δ13 δ13 tion of their transpiration water from upper, Cp = Ca – a – (b – a) ci/ca (1) unsaturated soil layers. Uptake of seasonally available soil moisture was not observed in Variation in observed isotopic ratios in tree cottonwood trees growing along a perennial ring holocellulose can be attributed to envi- reach of the same stream, implying that pre- ronmental, physiological and genetic factors cipitation use varies with depth to alluvial that affect ci/ca. In the case of drought, stom- groundwater in this species. ata respond to limited water supply and high Tree rings have long been used as a tool to transpirational demand by closing and reduc- study past climate in the western United States ing the supply of CO2 in the leaf, thereby de- δ13 (Douglas 1920). Investigations of ring width in creasing ci/ca and increasing Cp (Farquhar conifers have been used to reconstruct past et al. 1989). Photosynthate bearing the iso- drought events (Meko et al. 1995, Swetnam and topic signature at the time of fixation is used Betancourt 1998), lake level fluctuations (Peter- to form holocellulose in tree rings. Fractiona- son et al. 1999), and snowpack (Peterson and tion during formation results in a systematic Peterson 1994). Several authors have attempted shift in δ13C between bulk photosynthate and to use cottonwood tree ring widths to recon- holocellulose (Berninger et al. 2000). struct stream flow with mixed success (Johnson Leffler and Evans (1999) extracted holocel- et al. 1976, Reily and Johnson 1982, Stromberg lulose from 15 annual rings (1981–1995) of 10 and Patten 1996, Dudek et al. 1998). For ex- Populus fremontii trees growing along the Rio ample, Stromberg and Patten (1996) measured Grande near Socorro, New Mexico. Their study tree ring width to relate radial growth of black revealed a significant negative correlation be- cottonwood (Populus trichocarpa) to mean tween δ13C of holocellulose and stream flow. annual stream flow in the eastern Sierra Below a threshold level of stream flow, precip- Nevada. itation became an important correlate with To achieve intra-annual resolution in the tree tree ring δ13C. Carbon isotope values of tree ring record of southeastern Arizona, Meko and ring holocellulose were sensitive to stream flows Baisan (2001) used conifer tree ring wood early in the growing season. Response to spring density and width of the latewood to recon- flows was expected based on the results of P. struct activity of the North American monsoon. deltoides branch-growth phenology in Alberta Morphology distinguishes latewood bands in (Willms et al. 1998) and general patterns of many ring porous species. Tree rings of the vernal growth in riparian trees observed by diffuse porous Frémont cottonwood lack such Brown et al. (1977). distinct features. Thus, we used carbon isotope With its bimodal pattern of winter and ratios of holocellulose from subdivisions of the summer precipitation, the San Pedro River of tree ring to quantify shifts in plant water sta- southeastern Arizona is ideal for studies on tus associated with monsoon intensity. riparian plant-water interactions in a variety of 2004] POPULUS TREE RING δ13C 29 hydrologically, geomorphically, and lithologi- the perennial reach. However, differences in cally distinct settings (Pool and Coes 1999). diameter at breast height of the sample popu- The objectives of this paper were (1) to ex- lations were not statistically significant (Stu- plore the use of holocellulose δ13C from cotton- dent’s t test, t = –1.93, df = 12, P = 0.07). We wood tree ring subdivisions as a tool to exam- sampled 2 cores from each tree to account for ine interannual variation in ci/ca to moisture potential radial variation in isotope composi- availability in the growing season, and (2) to tion. Cottonwood stems typically possess an address sensitivity in ci/ca to monsoonal mois- elliptical morphology in cross section (Mike ture inputs in cottonwood populations grow- Merigliano personal communication). For con- ing under contrasting conditions of annual sistency, tree cores were extracted 180° from stream flow. each other on the minor axis of the elliptically shaped stem. We noted tree location by GPS, METHODS elevation, and distance from main channel; we also recorded crown condition for each tree. Study Sites Tree cores were prepared according to Fritts Our research was conducted on the San (1976) with minor modification. To avoid ring- Pedro River within the San Pedro National to-ring holocellulose transfer, we prepared cores Riparian Conservation Area in southeastern with razor blades rather than sandpaper. Core Arizona. We identified 2 Frémont cottonwood preparation with razor blades had the added populations separated by less than 15 km and benefit of preserving subtle variation in xylem approximately 50 m in elevation. Populations morphology, greatly facilitating ring identifica- occur on coarse, alluvial soils and possess sim- tion. Cores were visually cross-dated and rings ilar understory vegetation. We selected a pop- were assigned a year of growth according to ulation along a gaining, perennial reach imme- the methods of Yamaguchi (1991). Cores were diately downstream of the USGS Charles- measured for width to the nearest 0.1 mm ton stream gauge (#09741000, 31°37′33″N, using a semiautomated measuring system. 110°10′26″W, elevation 1205 m). The 2nd pop- Holocellulose Extraction ulation, 15 km downstream from the 1st, is on and δ13C Analysis a broad floodplain where stream flow is lost to the underlying alluvium and flow is inter- We focused on the response of δ13C to late mittent (31°45′03″N, 110°12′02″W, elevation growing season moisture availability similar to 1152 m). the ring growth responses in latewood of high- elevation conifers of the basin (Meko and Field Sampling and Preparation Baisan 2001). Halving individual tree rings of Tree Cores could potentially obscure a late growing sea- To minimize canopy boundary layer influ- son signal with material added earlier in the ences on carbon isotope content of tree ring growing season. Dividing rings into thirds was holocellulose, we chose trees growing in simi- a compromise between the potential fidelity of lar, open stands at the populations. We selected a late growing season signal and sample mass 7 trees at each site and extracted cores at requirements for cellulose extraction. From breast height from opposite sides of the trunk each tree core we excised individual tree rings with a 5.15-mm-increment borer. Based on using a scalpel. Individual rings were sub- experience with unusable, rotten cores, we divided into thirds corresponding to inner, avoided trees exhibiting “wet wood” symp- middle, and outer portions of growth. Corre- toms (Hofstra et al. 1999). We took care to sponding tree ring portions from the same tree sample trees with full crowns, growing in sim- were combined and ground in a mill (Model ilar position relative to the active channel. Ages #3383-L10, Thomas Scientific, Swedesboro, of the trees sampled at the 2 sites varied from NJ) to pass a 40-gauge screen. 15 to 50 years and ranged in diameter at We extracted holocellulose from inner- and breast height from 28 cm to 120 cm. Trees outer-growth portions of rings corresponding suitable for sampling at the intermittent reach to the growing seasons in 1990–2000. Raw had wider variation in stem diameter, which wood contains a myriad of compounds, each resulted in an intermittent reach sample pop- with its own pathway of synthesis and set of ulation with a mean diameter larger than at isotope fractionations. Holocellulose, a major 30 WESTERN NORTH AMERICAN NATURALIST [Volume 64 component of raw wood, is a favored compound conductance in response to changing leaf water for analysis as it is immobile once formed in conditions would induce changes in ci/ca and the tree ring of a given year and is relatively δ13C (Eq. 1) via 2 processes: (1) water avail- easy to extract. We followed the procedures ability to roots and its impact on hydraulic lim- described by Leavitt and Danzer (1993), using itations from soil to leaf and (2) atmospheric a refined pouching technique suggested by demand for transpiration at the leaf. Therefore Wright and Leavitt (personal communication) we limited our analysis to 3 environmental to extract holocellulose from the raw wood of variables: (1) summer precipitation, (2) stream P. fremontii tree rings. Briefly, the process flow, and (3) atmospheric vapor pressure deficit. involved several steps over 5 days to process Historic precipitation data for Tombstone, ~45 samples of raw wood contained in heat- Arizona, 14 km from our field sites (31°43′N, sealed polyethylene pouches (Ankom Technolo- 110°04′W, elevation 1384 m), were provided gies, Fairport, NY). Pouching eliminates sam- by the Western Regional Climate Center (Reno, ple loss due to transfer and reduces the risk of NV). Based on the precipitation record at Tomb- sample contamination (Leavitt and Danzer stone, Arizona, we determined that August 1993). Organically soluble compounds were was the peak month of monsoon precipitation dissolved from the raw wood by boiling in a (87.58 mm, s = 32.48, n = 11). Temperature 2:1 solution of toluene and ethanol followed and relative humidity from Walnut Gulch by bleaching the samples in an acetic acid/ Experimental Watershed at Tombstone, Ari- sodium chlorite solution. Samples were rinsed zona for the period of climate record, 1991– in distilled water and dried. The remaining 2000, was obtained from the USDA–ARS in holocellulose was analyzed for δ13C on a con- Tucson, Arizona (~14 km from our sites, ele- tinuous flow stable isotope mass spectrometer vation 1380 m). Using temperature data, we (Delta Plus, Finnigan MAT, Inc., San Jose, CA) calculated saturation water vapor pressure (es) at the University of Arizona Department of according to Dilley (1968): Geosciences. Error variation associated with holocellulose extraction and analysis was eval- es = 0.61078 exp [17.269T / (T + 237.8)] (2) uated by including a raw wood lab standard that yielded holocellulose with a mean carbon Combined with the concurrent relative isotope value of –25.67‰ (s = 0.26, n = 30). humidity data, we calculated seasonal mean Additionally, to evaluate the precision of the atmospheric vapor pressure deficit (Da) for the mass spectrometer, an acetanilide standard early and late portions of the growing season was measured along with the holocellulose defined above for 1991–2000, according to the samples (–29.59‰, s = 0.18). Data are re- equations provided by Landsberg (1986): ported relative to the PDB carbonate standard δ using delta ( ) notation in parts per thousand Relative humidity = ea/es (3) (‰; Craig 1957). D = e – e (4) Environmental Data a s a and Analysis Growing season precipitation typically comes The growing season was divided into an in the form of intense, highly localized con- early (1 April–30 June) and late (1 July–15 vective storms during the monsoon season. September) period. April 1 approximates the Runoff from these storms produces episodic beginning of the growing season for P. fremon- stream flow several orders of magnitude greater tii on the San Pedro River following leaf-out in than base flow. Given these conditions, stream late March. The North American monsoon flow statistics such as seasonal mean discharge commences in early July in southeastern Ari- are misleading and mask important stream zona and lasts through the middle of Septem- flow variation within and between years. We ber. obtained mean daily discharge data for the The ratio of leaf internal to ambient CO2 San Pedro River’s Charleston stream gauge concentration (ci/ca) is a function of photosyn- (#09741000) from the U.S. Geological Survey thetic demand and stomatal conductance of Water Resources Division. From these data the leaf. Assuming that photosynthetic CO2 de- we calculated flow-duration curves according mand over time is constant, changes in stomatal to Searcy (1959) to summarize stream flow 2004] POPULUS TREE RING δ13C 31 during the early (April–June) and late (July–15 TABLE 1. Correlations (r) among environmental vari- September) periods of the growing season. ables collected during the late growing season for 1990–2000. Relationships significant at the P < 0.05 level Briefly, mean daily discharge data for a period are noted with an asterisk (*). of interest in a given year was assigned a rank in order of descending discharge magnitude. Total ppt. August ppt. Q90

An exceedence frequency (F) was calculated Da 0.57 0.53 0.72* for each ranked discharge value (r) according Q90 0.52 0.56 to the formula: August ppt. 0.76*

F = [r / (n + 1)] * 100 (5) where n is the number of ranked discharge each site for each year. The years were then values for the period of interest. For example, averaged to create a grand mean for each site the largest mean daily discharge value (r = 1) from 1990 to 2000. The mean δ13C value for for a 99-day period (n = 99) would have an the population growing at the gaining, peren- exceedence frequency (F) equal to 1. Like- nial reach site (–26.33‰, sx– = 0.08, n = 11) wise, the median discharge value (r = 50) for was significantly lower (Student’s t test, t = the same period would have an exceedence 3.87, df = 19, P = 0.001) than that for the frequency (F) equal to 50. By this method the population growing at the losing, intermittent 50% exceedence value (Q50) expresses the reach (–25.68‰, sx– = 0.08, n = 11). The more median stream flow for any given period of positive δ13C value for the population at the time. losing, intermittent reach is consistent with de- Flow-duration analysis provides a conve- creased ci/ca, indicative of stomatal response nient and repeatable standard for comparing to comparatively elevated atmospheric demand seasonal patterns of stream flow across years or limited water supply during the growing at the same site. By ranking and assigning a season. frequency to discharge values, flow-duration To better understand the relationship be- analysis incorporates episodic, high-flow events tween tree ring width and carbon isotope con- and sustained groundwater discharge into a tent, we compared mean ring-width measure- single calculation (Vogel and Fennessey 1995). ments with the associated mean δ13C for each Using these calculated flow-duration data from year for both populations in 1990–2000. There the Charleston stream gauge, we estimated was no relationship between mean ring width early and late growing season 90% exceedence and mean annual δ13C among cottonwood stream flow values (Q90) for each year to char- trees growing at either the losing, intermittent acterize the availability of alluvial groundwater or gaining, perennial reach (r2 = 0.10, P = at both sites. Estimated from flow-duration 0.33, and r2 = 0.09, P = 0.37, respectively). analysis, Q90 is a common, low-flow index in Environmental parameters recorded dur- water resource planning and hydropower de- ing the late growing season were significantly sign (Gordon et al. 1992). In effect, we gener- intercorrelated (Table 1). Flow duration at the ated a mean daily stream flow discharge value 90% exceedence value (Q90) correlated with for each period that is equivalent to the dis- Da during this period. Stream discharge inte- charge value exceeded 9 days out of 10 during grates highly localized, summer convective that period. storms in the watershed. The frequency and Isotope data were checked for normality and extent of these storms is in part a function of analyzed with JMP IN Statistical Discovery atmospheric moisture content. Total late grow- software version 4 (SAS Institute, Inc., Cary, ing season precipitation and August precipita- NC). We used linear regression analysis to tion were less correlated with Q90 than was examine dependence of tree ring holocellulose Da. δ13 C values on late growing season Da, Q90, Within the intermittent and perennial reach total monsoon, and August precipitation. growing populations, there was no significant relationship between outer-third holocellulose 13 RESULTS δ C variation and annual tree ring width for 1990–2000 (r2 = 0.08, P = 0.39, and r2 = We averaged early and late season δ13C 0.04, P = 0.55, respectively). Interannual δ13C values to generate a growing season average at values in outer-third holocellulose from trees 32 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Fig. 1. Relationships between outer-third tree-ring holocellulose δ13C in Populus fremontii and (a,b) August precipitation amount, (c,d) late growing season 90% exceedence flows (Q90), and (e,f) atmospheric vapor pressure deficit (Da) at perennial and intermittent reaches on the San Pedro River in southeastern Arizona.

at the losing, intermittent reach were nega- We calculated the shift in δ13C values from tively correlated with August precipitation the inner- to outer-third tree-ring subdivisions (Fig. 1b; r2 = 0.50, P = 0.015) and late season for each year. No systematic intra-annual shift 2 δ13 13 Q90 (Fig. 1d; r = 0.50, P = 0.015). C val- was found in δ C in response to late growing ues were highest during years when August season moisture conditions at the gaining, per- precipitation and late season stream flows ennial reach site (Figs. 2a, 2c, 2e). However, were low. Outer-third δ13C values were posi- there were correlations between the magni- 13 tively correlated with late season Da at this tude of the intra-annual shift in δ C and late site (Fig. 1f; r2 = 0.45, P = 0.03). There was season moisture conditions at the losing, inter- no significant correlation between tree ring mittent reach site. August precipitation (Fig. δ13 2 2 C values and total monsoon rainfall at the 2b; r = 0.37, P = 0.05), Q90 (Fig. 2d, r = δ13 intermittent reach site. C values from 0.59, P = 0.005), and late growing season Da outer-third tree ring subdivisions at the gain- (Fig. 2f; r2 = 0.57, P = 0.01) were correlated ing, perennial reach were not significantly cor- with intra-annual shifts in carbon isotope val- related with any of the environmental vari- ues for trees along the losing, intermittent ables examined (Figs. 1a, 1c, 1e). reach. There was no intra-annual shift in δ13C 2004] POPULUS TREE RING δ13C 33

Fig. 2. Differences between δ13C of tree ring holocellulose inner- and outer-third portions in relation to August precipitation for 1990–2000 (a,b); late growing season Q90 for 1990–2000 (c,d); late growing season atmospheric vapor pressure deficit (Da) for 1991–2000 (e,f) for Populus fremontii growing at perennial and intermittent reaches on the San Pedro River in southeastern Arizona.

values associated with total monsoon rainfall response to climate in populations of Populus at either site. fremontii—a species that lacks visible latewood bands. Because we compared only 2 DISCUSSION populations, each growing at different reaches of the San Pedro River, our statistical scope of This study provides a decade-long record inference is limited to these sites. Leaf gas of physiological response of a dominant ripar- exchange in P. fremontii responded signifi- ian tree species to hydrologic and climatic variation. Our data demonstrate the utility of cantly to variation in moisture conditions dur- carbon isotope analysis of tree rings to docu- ing the late growing season at the intermittent ment leaf physiological response to intra- and reach site, but not at the perennial reach site. interannual environmental variability in semi- Ring δ13C remained fairly constant at the arid riparian ecosystems. Equal subdivision of perennial reach site despite widely fluctuating tree rings for carbon isotope analysis allowed stream flow and atmospheric moisture condi- us to achieve intra-annual resolution in tree tions. 34 WESTERN NORTH AMERICAN NATURALIST [Volume 64

We found no evidence for a physiological for such a relationship with winter moisture at response to threshold values of late growing either site (data not shown). season flow for 1990–2000 among the popula- Differences in δ13C of tree ring holocellu- tions of P. fremontii that were sampled on the lose could be explained without invoking San Pedro River. Leffler and Evans (1999) re- stomatal conductance changes in response to ported that during years of abundant flow in leaf water conditions. High photosynthetic the middle Rio Grande, whole-ring holocellu- capacity can reduce ci/ca, independent of δ13 lose C of Frémont cottonwood trees did not changes in stomatal conductance. CO2 demand correlate with variation in stream flow. How- by chloroplasts is related to soil nitrogen avail- ever, in years when flow was below a threshold ability, temperature, and other factors that value, δ13C values were negatively correlated impact photosynthetic enzyme activity or con- with stream flow. Other studies have docu- centrations in the leaf. However, given simi- mented nonlinear responses of riparian vege- larities in elevation, vegetation community, tation to environmental conditions at the indi- management history, and soils at the 2 sites, it vidual and community level (Stromberg et al. seems unlikely that these factors are important 1996, Friedman and Auble 1999, Scott et al. for explaining differences. 1999). Stream flow and associated alluvial soil Intra- and interannual variation in the car- water potential at the perennial reach site may bon isotope content of atmospheric CO2 could have never reached the point beyond which also account for variation in δ13C of tree ring δ13 stomatal conductance was impacted. Converse- holocellulose. We used atmospheric CO2 C ly, alluvial soil water potentials at the intermit- data for 1990–2000 from Mauna Loa Observa- tent reach site may have been at or below the tory, Hawaii (NOAA CMDL, http://www. threshold value over the entire period covered cmdl.noaa.gov/ccgg/flask/index.html), to calcu- by our isotope sampling. Despite the close late the isotopic discrimination of cellulose ∆ proximity of the 2 populations to one another, formation ( holocellulose) from Farquhar et al. a genetic difference in drought sensitivity (1989): between populations is another possible expla- ∆ δ13 δ13 nation of these results. holocellulose = ( Cair – Cholocellulose) / δ13 Positive growth responses in Populus to (1 + Cholocellulose) (6) spring and early summer stream flow have been inferred from whole-ring δ13C (Leffler We compared early and late growing season ∆ δ13 and Evans 1999) and branch elongation (Willms holocellulose to the corresponding C value et al. 1998). We found no such relationships using regression analysis. We found high cor- δ13 δ13 ∆ between tree ring inner-third C values and relations between Cholocellulose and holocel- flow. It is possible that stored photosynthate lulose during the early and late growing sea- from prior years contributes to early season sons at both sites (Table 2). We conclude that tree ring formation (Hill et al. 1995, Robertson variation in carbon isotope composition of δ13 et al. 1997). Products of photosynthesis in the holocellulose, whether calculated as Cholo- ∆ early growing season may be allocated to the cellulose or holocellulose, does not change the growth of new leaves (Terwilliger 1997) and interpretation of the influence of intra- and fine roots (Burton et al. 2000), blurring the early interannual variability in water availability on growth isotopic signature of tree ring holocel- ci/ca. lulose. We reject a 3rd alternative hypothesis: There is no satisfying physiological expla- moisture stored in the upper-soil profile car- nation for why tree ring δ13C values would ried over from winter rains ameliorates leaf- reflect August and not monsoon total rainfall water status, and thereby masks the environ- during a growing season. These findings high- mental signal in δ13C of the rings. We tested light the difficulty of assessing the ecophysio- the hypothesis that in years of abundant winter logical impact of monsoonal precipitation at moisture, as measured by precipitation totals the catchment scale. Because of the narrow from October through March, inner-third por- spatial extent of summer convective storm tions of tree ring holocellulose δ13C would activity, storms recorded at the Tombstone tend toward greater discrimination, indicative rain gauge might miss the cottonwood popula- of greater stomatal conductance and more favor- tions growing 14 km away. Ideally, a long-term able leaf-water status. There was no evidence weather station would be situated at both sites 2004] POPULUS TREE RING δ13C 35

δ13 ∆ TABLE 2. Correlations (r) between holocellulose C and holocellulose for 1990–2000. Relationships significant at the P < 0.0001 level noted with ##. Relationships significant at the P < 0.0005 noted with #. Early growing season Late growing season Growing season shift Intermittent reach 0.94## 0.98## 0.98## Perennial reach 0.98## 0.89# 0.99##

to provide a spatially explicit record of sum- Demonstrable population-level variability in mer convective storm precipitation for the δ13C associated with stream flow and hydro- period of tree ring record. geomorphic setting provides a starting point Frequency-duration analysis provided use- for additional investigations on tree-water ful stream flow information by incorporating relations and carbon exchange in this ecosys- both sustained low flows and occasional extreme tem. Future isotope dendrochronological stud- events into a single calculation comparable ies on riparian species should carefully con- across years. To validate the utility of flow dura- sider the influence of hydrologic conditions on tion analysis in semiarid riparian ecology, we δ13C values. examined the correlation between mean late 13 growing season stream flow and δ C of outer- ACKNOWLEDGMENTS third tree ring subdivisions at the losing, intermittent reach for 1990–2000. Sensitive to ex- Nathan English, Kiyomi Morino, and Chris treme high-flow events, mean late growing sea- Eastoe provided valuable assistance in the field son stream flow failed to significantly correlate and laboratory. Dave Meko, William Wright, with δ13C of outer-third tree ring portions (r2 and Donald Potts gave valuable technical = 0.28, P = 0.09). We recommend that future advice. Steven Smith and Steven Leavitt pro- plant-water relations studies in semiarid and vided critical reviews of earlier versions of this arid riparian zones carefully consider the im- manuscript. Travis Huxman, John Marshall, portance of stream flow variability in plant and an anonymous reviewer provided valuable response. Flow-duration analysis is a tool that editorial comments. Access to the San Pedro accounts for biases with integrated seasonal National Riparian Conservation Area was pro- variables as a result of extreme events. vided by the Bureau of Land Management, The ratio of leaf internal to ambient CO2 Sierra Vista office. Research was funded by concentration and thus δ13C is regulated to a the Southwest Ground-Water Resources Pro- large degree by stomatal conductance (Far- ject of the Ground-Water Resources Program quhar et al. 1982). Stomatal conductance to- of the U.S. Geological Survey. gether with aerodynamic conductance regu- lates transpiration rate at the ecosystem scale LITERATURE CITED (Jarvis and McNaughton 1986). Understand- 13 ing intra- and interannual variation of δ C in BERNINGER, F., E. SONNINEN, T. AALTO, AND J. LLOYD. 2000. native, riparian forests has implications for the Modeling 13C discrimination in tree rings. Global scaling of water relations in this ecosystem. Biogeochemical Cycles 14:213–223. BROOKS, J.R., L.B. FLANAGAN, AND J.R. EHLERINGER. 1998. Catchment-scale models should take into Responses of boreal conifers to climate fluctuations: account the dynamic role of hydrologic setting indications from tree ring widths and carbon isotope and riparian forest water use in response to analyses. Canadian Journal of Forest Research 28: summer moisture conditions. Multiscale mod- 524–533. eling efforts such as those used by Goodrich et BROWN, D.E., C.H. LOWE, AND J.F. HAUSER. 1977. South- western riparian communities: their biotic impor- al. (2000) might benefit from refinement of the tance and management in Arizona. Pages 201–211 in evapotranspiration component by spatially inte- R.R. Johnson and D.A. Jones, technical coordinators, grating seasonal patterns of vegetation response Importance, preservation and management of ripar- with changing seasonal moisture conditions. ian habitat: a symposium. USDA Forest Service, From a management perspective, this study General Technical Report RM-43, Fort Collins, CO. BURTON, A.J., K.S. PREGITZER, AND R.L. HENDRICK.2000. adds to a growing body of evidence that hy- Relationships between fine root dynamic and nitro- drologic setting plays a key role in how riparian gen availability in Michigan northern hardwood plant populations respond to regional drought. forests. Oecologia 125:389–399. 36 WESTERN NORTH AMERICAN NATURALIST [Volume 64

CRAIG, H. 1957. Isotopic standards for carbon and oxygen LEFFLER, A.J., AND A.S. EVANS. 1999. Variation in carbon and correction factors for mass-spectrometric analy- isotope composition among years in the riparian tree sis of carbon dioxide. Geochimica et Cosmochimica Populus fremontii. Oecologia 119:311–319. Acta 12:133–140. ______. 2001. Physiological variation among Populus fre- DILLEY, A.C. 1968. On the computer calculation of vapour montii populations: short- and long-term relationships pressure and specific humidity gradients from psy- between δ13C and water availability. Tree Physiology chrometric data. Journal of Applied Meterology 7: 21:1149–1155. 717–719. LINES, G.C. 1999. Health of native riparian vegetation and DOHERTY, R., AND L.O. MEARNS. 1999. A comparison of its relation to hydrologic conditions along the Mojave simulations of current climate from two coupled River, southern California. USGS Water-Resources atmosphere-ocean global climate models against Investigations Report 99-4112, Sacramento, CA. observations and evaluation of their future climates. LIVINGSTON, N.J., AND D.L. SPITTLEHOUSE. 1996. Carbon Report to the National Institute for Global Environ- isotope fractionation in tree ring early and late wood mental Change (NIGEC), Boulder, CO. in relation to intra-growing season water balance. DOUGLAS, A.E. 1920. Evidence of climatic effects in the Plant, Cell and Environment 19:768–774. annual rings of trees. Ecology 1:24–32. MCNULTY, S.G., AND W. T. S WANK. 1995. Wood δ13C as a DUDEK, D.M., J.R. MCCLENAHEN, AND W. J. M ITSCH. measure of annual basal area growth and soil water 1998. Tree growth responses of Populus deltoides and stress in a Pinus strobes forest. Ecology 76:1581–1586. Juglans nigra to streamflow and climate in a bottom- MEARNS, L.O., I. BOGARDI, F. GIORGI, I. MATYASOVSZKY, AND land hardwood forest in central Ohio. American Mid- M. PALECKI. 1999. Comparison of climate change land Naturalist 140:233–244. scenarios generated from regional climate model FARQUHAR, G.D., J.R. EHLERINGER, AND K.T. HUBICK. 1989. experiments and statistical downscaling. Journal of Carbon isotope discrimination and photosynthesis. Geophysical Research 104:6603–6621. Annual Review of Plant Physiology and Plant Molec- MEKO, D.M., AND C.H. BAISAN. 2001. Pilot study of late- ular Biology 40:503–537. wood-width of conifers as an indicator of variability FARQUHAR, G.D., M.H. O’LEARY, AND J.A. BERRY. 1982. On of summer rainfall in the North American monsoon the relationship between carbon isotope discrimina- region. International Journal of Climatology 21: tion and the intercellular carbon dioxide concentra- 697–708. tion in leaves. Australian Journal of Plant Physiology MEKO, D.M., C.W. STOCKTON, AND W.R. BOGGESS. 1995. 9:121–137. The tree ring record of severe sustained drought. FRANCEY, R.J., AND F. D . F ARQUHAR. 1982. An explanation Water Resources Bulletin 31:789–801. of 13C/12C variations in tree rings. Nature 297: PETERSON, D.L., D.G. SILSBEE, AND K.T. REDMOND. 1999. 28–31. Detecting long-term hydrological patterns at Crater FRIEDMAN, J.M., AND G.T. AUBLE. 1999. Mortality of ripar- Lake, Oregon. Northwest Science 73:121–130. ian box elder from sediment mobilization and ex- PETERSON, D.W., AND D.L. PETERSON. 1994. Effects of cli- tended inundation. Regulated Rivers 15:463–476. mate on radial growth of subalpine conifers in the FRITTS, H.C. 1976. Tree rings and climate. Academic Press, North Cascade Mountains. Canadian Journal of For- London. 567 pp. est Research 24:1921–1932. GOODRICH, D.C., R. SCOTT, J. QI, B. GOFF, C.L. UNKRICH, POOL, D.L., AND A.L. COES. 1999. Hydrogeologic investi- M.S. MORAN, D.G. WILLIAMS, ET AL. 2000. Season gations of the Sierra Vista Subwatershed of the estimates of riparian evapotranspiration using remote Upper San Pedro Basin, Cochise County, Southeast and in situ measurements. Agricultural and Forest Arizona. U.S. Geological Survey Water Resources Meteorology 105:281–309. Investigation Report 99-4197, Tucson, AZ. GORDON, N.D., T.A. MCMAHON, AND B.T. FINLAYSON. 1992. REILY, P.W., AND W.C. JOHNSON. 1982. The effects of Stream hydrology: an introduction for ecologists. altered hydrologic regime on tree growth along the John Wiley & Sons, New York. 526 pp. Missouri River in North Dakota. Canadian Journal HILL, S.A., J.S. WATERHOUSE, E.M. FIELD, V.R. SWITSUR, of Botany 60:2410–2423. AND T. A PREES. 1995. Rapid recycling of triose phos- ROBERSTON, I., V.R. SWITSUR, A.H.C. CARTER, A.C. BARKER, phates in oak stem tissue. Plant, Cell and Environ- J.S. WATERHOUSE, K.R. BRIFFA, AND P. D . J ONES. 1997. ment 18:931–936. Signal strength and climate relationships in 13C/12C HOFSTRA, T.S., J.C. STROMBERG, AND J.C. STUTZ. 1999. ratios of tree ring cellulose from oak in east England. Factors associated with wetwood intensity of Populus Journal of Geophysical Research–Atmospheres 102: fremontii (Fremont cottonwood) in Arizona. Great 19507–19516. Basin Naturalist 59:85–91. ROOD, S.B., AND J.M. MAHONEY. 2000. Revised instream JARVIS, P.G., AND K.G. MCNAUGHTON. 1986. Stomatal con- flow regulations enables cottonwood recruitment trol of transpiration: scaling up from leaf to region. along the St. Mary River, Alberta, Canada. Rivers 7: Advances in Ecological Research 15:1–4. 109–125. JOHNSON, W., R. BURGESS, AND W. K EAMMERER. 1976. SCHAEFFER, S.M., D.G. WILLIAMS, AND D.C. GOODRICH. Forest overstory vegetation and environment on the 2000. Transpiration of cottonwood/willow forest esti- Missouri River floodplain in North Dakota. Ecologi- mated from sap flux. Agricultural and Forest Meteo- cal Monographs 46:59–84. rology 105:257–270. LANDSBERG, J.J. 1986. Physiological ecology of forest pro- SCOTT, M., P.B. SHAFROTH, AND G.T. AUBLE. 1999. Responses duction. Academic Press, London. 198 pp. of riparian cottonwoods to alluvial water table LEAVITT, S.W., AND S.R. DANZER. 1993. Method for batch declines. Environmental Management 23:347–358. processing small wood samples to holocellulose for SEARCY, J.K. 1959. Flow-duration curves. U.S. Geological stable-carbon isotope analysis. Analytical Chemistry Survey Water Supply Paper 1542-A, Washington, 65:87–89. DC. 33 pp. 2004] POPULUS TREE RING δ13C 37

SHAFROTH, P.B., J.C. STROMBERG, AND D.T. PATTEN. 2000. matic variability in the American Southwest. Journal Woody riparian vegetation response to different allu- of Climate 11:3128–3147. vial water table regimes. Western North American TERWILLIGER, V.J. 1997. Changes in the δ13C values of Naturalist 60:66–76. trees during a tropical rainy season: some effects in SNYDER, K.A., AND D.G. WILLIAMS. 2000. Water sources addition to diffusion and carboxylation by Rubisco? used by riparian trees varies among stream types on American Journal of Botany 84:1693–1700. the San Pedro River, Arizona. Agriculutural and For- VOGEL, R.M., AND N.M. FENNESSEY. 1995. Flow duration est Meteorology 105:227–240. curves II: a review of applications in water resources SPARKS, J.P., AND R.A. BLACK. 1999. Regulation of water planning. Water Resources Bulletin 31:1029–1039. loss in populations of Populus trichocarpa: the role of WATERHOUSE, J.S., A.C. BARKER, A.H.C. CARTER, L.I. stomatal control in preventing xylem cavitation. Tree AGAFONOV, AND N.J. LOADER. 2000. Stable carbon Physiology 19:453–459. isotopes in Scots pine tree rings preserve a record of STROMBERG, J.C. 1993. Frémont cottonwood–Goodding flow of the river Ob. Geophysical Research Letters willow riparian forests: a review of their ecology, 27:3529–3532. threats, and recovery potential. Journal of the Ari- WALCROFT, A.S., W.B. SILVESTER, D. WHITEHEAD, AND F. M . zona–Nevada Academy of Sciences 26:97–109. KELLIHER. 1997. Seasonal changes in stable carbon ______. 1998. Dynamics of Fremont cottonwood (Populus isotope ratios within annual rings of Pinus radiata fremontii) and saltcedar (Tamarix chinenesis) popula- reflect environmental regulation of growth processes. tions along the San Pedro River, Arizona. Journal of Australian Journal of Plant Physiology 24:57–68. Arid Environments 40:133–155. WILLMS, J., S.B. ROOD, W. WILLMS, AND M. TYREE.1998. STROMBERG, J.C., AND D.T. PATTEN. 1996. Instream flow Branch growth of riparian cottonwoods: a hydrologi- and cottonwood growth in the eastern Sierra Nevada cally sensitive dendrochronological tool. Trees 12: of California, USA. Regulated Rivers Research and 215–223. Management 12:1–12. YAMAGUCHI, D.K. 1991. A simple method for cross-dating STROMBERG, J.C., R. TILLER, AND B. RICHTER. 1996. Effects increment cores from living trees. Canadian Journal of groundwater decline on riparian vegetation of of Forest Research 21:414–416. semiarid regions: the San Pedro, Arizona. Ecological Applications 6:113–131. Received 21 October 2002 SWETNAM, T.W., AND J.L. BETANCOURT. 1998. Mesoscale Accepted 25 March 2003 disturbance and ecological response to decadal cli- Western North American Naturalist 64(1), ©2004, pp. 38–44

CONSERVATION, STATUS, AND LIFE HISTORY OF THE ENDANGERED WHITE RIVER SPINEDACE, LEPIDOMEDA ALBIVALLIS (CYPRINIDAE)

G. Gary Scoppettone1, James E. Harvey2, and James Heinrich3

ABSTRACT.—Lepidomeda albivallis (White River spinedace), a fish species endemic to the White River, Nevada, appeared headed toward extinction. In 1991 only 1 population remained, and it comprised fewer than 50 individuals in a 70-m stream reach. We monitored population recruitment and distribution and studied life history and habitat use from 1993 through 1998. We determined that L. albivallis was not reproducing and was continuing to decline, and as an emergency measure we relocated the population (14 in spring 1995 and 6 in spring 1996) downstream 200 m to a secure habitat that we judged more favorable for reproduction. The relocated population reproduced, and by September 1998 it had increased to 396 individuals that inhabited more than 1 km of stream including both pond and stream habitats. In streams they oriented near the bottom but frequently moved up in the water column to strike at drift items. Gut analysis of museum specimens indicated L. albivallis is omnivorous but feeds primarily upon aquatic invertebrates. Conservation of L. albivallis will require reestablishing additional populations within its former range.

Key words: Lepidomeda, spinedace, Plagopterini, endangered fish, life history, White River, Colorado River, largemouth bass.

Lepidomeda albivallis (White River spine- The purpose of this study was to monitor the dace) is an endangered fish species endemic L. albivallis population to gauge recruitment to the White River system, Nye and White (or verify nonrecruitment) and to generate basic Pine Counties, Nevada. It is a member of the life history information useful for species man- cyprinid tribe Plagopterini known only from agement and recovery. During the study we the lower Colorado River basin (Miller and relocated the population and monitored repro- Hubbs 1960). Plagopterini has undergone duction response and habitat use. population declines throughout its range as a result of habitat alteration and introduced STUDY AREA species (Miller 1961, 1963, Douglas et al. 1994). One species, Lepidomeda altivelis (Pahranagat The Flag Springs/Sunnyside Creek system, spinedace), is extinct, and 5 others are listed which harbors L. albivallis, is in the southern portion of the White River system and one of or have been considered for federal listing the few spring systems directly discharging (U.S. Fish and Wildlife Service 1994a, 1997). into the historical White River channel (Fig. When the type specimen of L. albivallis was 1). Most spring discharge is diverted for irri- collected in 1938, the species was abundant gation and is disjunct from the White River and had been sampled from 7 spring systems channel. Flag Springs consists of 3 springs (Miller and Hubbs 1960, La Rivers 1962). By originating within 300 m of each other. In the time of its listing (U.S. Fish and Wildlife 1991 the northernmost spring (herein North Service 1985), distribution was limited to 2 Fork) harbored the last known L. albivallis in spring systems, and in 1991 there were fewer its upper 70 m (Fig. 1); habitat consisted of than 50 individuals inhabiting a single 70-m shallow riffle (≈10 cm deep) and 2 ponds (the stream reach (U.S. Fish and Wildlife Service upper 300 m2 with a maximum depth of 1 m, 1994b). Remaining L. albivallis were large, the lower about 75 m2 with a maximum depth suggesting the population was not recruiting of 0.7 m). Restricted distribution of Lepido- and on the verge of extinction. meda albivallis in the Flag Springs system was

1Biological Resources Division, U.S. Geological Survey, 1340 Financial Blvd., Suite 161, Reno, NV 89502. 2U.S. Fish and Wildlife Service, 1340 Financial Blvd., Reno, NV 89502. 3Nevada Division of Wildlife, 1204 Avenue I, Boulder City, NV 89005.

38 2004] SPINEDACE STATUS AND LIFE HISTORY 39

Fig. 1. Map of the Flag Springs/Sunnyside Creek system, Nevada, showing Lepidomeda albivallis distribution before relocation (1991–1995) and 3 years after relocation (1998). Insets show the relationship of the Flag Springs/Sunnyside Creek system to the White River and the White River to the course of the pluvial White River.

a result of Micropterus salmoides (largemouth springs. The upper 70 m of North Fork had a bass) predation (U.S. Fish and Wildlife Service riparian corridor of willow (Salix), currant 1994b). The upper North Fork was isolated (Ribes), and wild rose (Rosa), and the upper and protected from M. salmoides by a steep 100 m of the 2 southern spring outflow channels stream grade and 2-m-high dam (Fig. 1). was lined with cottonwood (Populus) and wil- The North Fork discharged 0.03 m3 ⋅ s–1 and low. Sunnyside Creek flowed through open flowed 550 m to join the combined outflow grassland for about 4 km before discharging (0.04 m3 ⋅ s–1) of the 2 southern Flag Springs into the White River channel and then Adams- (herein South Fork) and form Sunnyside Creek McGill Reservoir (Fig. 1). Lepidomeda albival- (Fig. 1). Water temperature was approximately lis co-occurred with Catostomus clarki (desert 16.0°C at the origin of the North Fork and sucker) and Rhinichthys osculus (speckled dace) 20°–23°C at the origin of the 2 southern Flag in the Flag Springs/Sunnyside Creek system. 40 WESTERN NORTH AMERICAN NATURALIST [Volume 64

MATERIALS AND METHODS Age and Growth Species Status and Adaptive Age and growth data were generated from Management Actions museum specimens (n = 30) collected from the Flag Springs system in 1938 (UMMZ Adaptive management actions were taken 124990). Lepidomeda albivallis scales were to expand the L. albivallis population. First, small and annuli could not be determined. We M. salmoides was eradicated from the upper therefore used opercle bones to estimate age 500 m of the North Fork by electrofishing in (Scoppettone 1988). By scraping the opercle spring 1993, and a temporary fish barrier was with a scalpel, we removed the flesh and then installed to prevent reinvasion (Fig. 1). To de- used a dissection microscope to identify annuli, termine if L. albivallis was reproducing and its which were assumed to be zones where opaque distribution expanding downstream following bone met more transparent zones (Casselman this initial eradication, the North Fork was 1974). We generated a logarithmic equation to snorkeled from the temporary fish barrier illustrate the relationship of L. albivallis age to upstream to the springhead seasonally from growth (Sokal and Rohlf 1995). November 1993 to March 1995. Meanwhile, in summer 1994, personnel of the Nevada Food Habits Division of Wildlife eradicated remaining M. Food habit analysis was done on the 30 Flag salmoides in North and South Forks/Sunny- Springs fish used for age growth analysis plus side Creek to a permanent barrier about 2.5 km 14 specimens collected from Preston Big Spring downstream from the springheads (Fig. 1). in 1961, 1964, and 1965 (5F-141, 4F-1145, and The 2nd action was relocation in April and 4F-1148, Zoology Museum, University of Neva- May 1995, 1996, and 1997 of all L. albivallis da, Las Vegas). Specimens ranged from 61 mm from the upper 70 m of North Fork to a site to 96 mm FL. The anterior third of the gut 200 m downstream. Thirty standard “Gee” was examined with a dissecting scope, and minnow traps, half lined with 1-mm-mesh food items were identified and quantified by plastic screen, were baited with dry dog food percent frequency of occurrence and percent and fished within the 70-m reach. A hoop net volume (Windell 1971). We also compared gut (6.4-mm stretch mesh, 1.6 m long, with a 0.7- length to fish total length (Nikolsky 1963). m opening) was also fished in each of the 2 large pools. Traps were fished 7–10 days with- Habitat Use in the 2-month fishing period. In May 1997 In July, September, and November 1993, and the entire area within the 70-m reach was March, May, and July 1994, we quantified L. electrofished with a Smith-Root Type VII albivallis microhabitat use in the 2 headwater electroshocker to ensure all L. albivallis had ponds of North Fork and in July and October been removed. Relocated fish had access to 3 1998, and January and April 1999 in the Flag km of M. salmoides–free water and habitat Springs/Sunnyside Creek outflow (Fig. 1). Mea- similar to that used for spawning by other sured variables were total depth (water col- Plagopterini (Barber et al. 1970, Rinne 1971, umn depth at the subject fish), focal depth Blinn et al. 1998). (depth from water surface to fish), focal veloc- We monitored recruitment success and pop- ity (water velocity at fish), and mean velocity ulation expansion after L. albivallis relocation. (mean water column velocity at fish). Relative In September 1995 we began snorkeling the depth was determined by dividing focal depth North Fork, and in September 1996 and 1997 by total depth. Fish were located using mask and October 1998 we added the upper 1.5 km and snorkel. A Marsh and McBirney model of Sunnyside Creek and the South Fork. For 210 flow meter on a graduated rod was used to each monitoring period the snorkeler moved measure stream velocity and depth. We classi- upstream to count L. albivallis and estimated fied life stages as larva (<20 mm FL), juvenile fork length (FL) of 33%–100% of the fish (20–60 mm FL), and adult (>60 mm FL). sighted. Samples selected were representative of the population. These data were compared Sexual Dimorphism with lengths of a sample taken from the and Reproduction upstream 70 m of North Fork on 8 May 1992, Color differences and presence of tubercles prior to management actions. have been used to determine sex of L. albivallis 2004] SPINEDACE STATUS AND LIFE HISTORY 41

(Miller and Hubbs 1960), but these differences were not readily detectable in fish we captured from headwater ponds. Miller (1963) observed that L. vittata males have longer pectoral fins than females, extending to the pelvic insertion or beyond in males and falling short of the insertion in females. We investigated this method to sex L. albivallis using twenty 1938 Flag Springs specimens (61 m to 96 mm FL) and confirmed gender by dissecting each fish.

RESULTS Species Status and Adaptive Management Actions North Fork snorkel surveys from November 1993 through March 1995 indicated L. albivallis were restricted to the upper 70 m, not reproducing, and declining in number. Counts were 12 on 16 November 1993, 31 on 2 March 1994, 26 on 23 May 1994, 6 on 14 July 1994 and 10 January 1995, and 0 on 30 March 1995. By March 1995 aquatic vegetation was so dense that we could not determine whether L. albivallis were still present. Nevada Division of Wildlife, U.S. Fish and Wildlife Service, and U.S. Geological Survey decided that remaining L. albivallis be relocated 200 m downstream to an area of greater habitat diversity, thus enhanc- ing the probability of reproduction. We captured 20 fish (9 apparent females and 5 males in spring 1995, and 2 females and 4 unsexed in spring 1996). No L. albivallis were captured in Fig. 2. Length frequency of Lepidomeda albivallis pop- spring 1997 after trapping and electrofishing, ulations before (1992) and after (1996–1998) relocation, and we concluded that none remained. Relo- Flag Springs, Nevada. cated fish were large (≥75 mm FL), one measuring 165 mm FL, the largest L. albivallis ever reported. On 6 September 1996 we documented that Springs ranged from <1 year to 12 years of relocated L. albivallis had reproduced: 61 L. age (Fig. 3). Fish 1 year of age ranged from 42 albivallis were counted. All were young and mm to 65 mm FL and at 5 years from 70 mm smaller than fish captured in 1992 (Fig. 2). They to 98 mm FL. Growth was logarithmic with were found exclusively in the South Fork. In age, and there was a high correlation between September 1997 we counted 112 individuals, length and age (r = 0.92). The longest (107 mm ranging from 18 mm to 73 mm FL (Fig. 2); 81 FL) and oldest fish (12 years) were female; the were in the upper 500 m of the South Fork longest and oldest male was 92 mm FL and 5 and 31 in Sunnyside Creek. One year later 396 years of age. Relocated fish of both sexes were were counted, ranging from 20 mm to 110 mm presumed to be at least 11 years old. FL, with most (305) in the South Fork, but ex- Food Habits tending up the North Fork and several hundred meters downstream into Sunnyside Creek Gut samples revealed that the species is (Fig. 1). omnivorous but tends toward carnivory. This was corroborated by gut length, which aver- Age and Growth aged only 71% (s = 0.11; n = 31) of total Museum specimens collected from Flag length. Invertebrates were in 90% of guts 42 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Fig. 3. Fork length and age of Lepidomeda albivallis collected from Flag Springs, Nevada, in 1938 (n = 30).

examined and represented 71% of the volume Sexual Dimorphism of items. Most were aquatic insects (Ephem- and Reproduction eroptera, Trichoptera, Plecoptera, Hemiptera, We found ova of 2 sizes in the egg skein, aquatic Coleoptera, and aquatic Diptera), but suggesting that reproduction occurs over a Gastropoda and Turbelleria also were present. protracted period. This was corroborated by Plant material and algae were in 41% of guts museum collections that included post-larvae but accounted for only 17% of the volume con- <20 mm FL taken from spring through late sumed, and detritus made up 7% by volume. August. We also observed 13-mm larvae on 18 We observed L. albivallis in the Flag Springs May 1999, indicating reproduction in April. system feeding in or near flowing water where Thus, reproduction appeared to take place they struck at drift items. In Preston Big Spring, from at least April into July. The smallest female 2 of 14 specimens had consumed Crenichthys inspected was 61 mm FL (51 mm standard baileyi: a 91-mm FL male had taken a 25-mm length), collected in April 1965. It contained FL C. baileyi and a 61-mm FL female had 2 size classes of eggs, the largest of which taken an 18-mm FL C. baileyi. was close to maturity. This fish was the small- Habitat Use est of 20 sexed by using position of the pectoral fin relative to the origin of the pelvic fin, Lepidomeda albivallis confined to the head- and the only one for which the method was waters of the North Fork inhabited ponds unsuccessful. where they were benthically oriented and generally in mean water velocities of <2.0 cm DISCUSSION ⋅ s–1 (Table 1). In the outflow, adults inhabited higher water velocities (18.7 cm ⋅ s–1 mean water Status of L. albivallis was determined to be column and 14.9 cm ⋅ s–1 focal point) but were “critically imperiled” and conservation mea- still benthically oriented; they generally ori- sures were implemented to avoid its extinc- ented upstream but moved in various direction. In 1993 the only known L. albivallis were tions to strike at drift items. Juveniles inhabit- at the head of the North Fork of Flag Springs ing shallower water were closer to the water and were large, old, declining, and with no surface. Larvae occurred near the surface and apparent recruitment for several years. One in much shallower and slower water than that hypothesis for lack of recruitment was that L. used by adults and juveniles. albivallis were trapped upstream of the dam 2004] SPINEDACE STATUS AND LIFE HISTORY 43

TABLE 1. Characteristics of habitat used by 3 life stages of L. albivallis in North Fork ponds (1993–94) and South Fork/Sunnyside Creek outflow (1998–99), Nevada. Mean velocity Focal velocity Total depth Relative depth Mean ± s Mean ± s Mean ± s Mean ± s Life stage (cm ⋅ s–1) (cm ⋅ s–1) (cm) (%) n

NORTH FORK PONDS Adult 1.6 ± 3.1 1.4 ± 2.4 57.5 ± 8.7 81.6 ± 10.9 62 NORTH FORK/SOUTH FORK/SUNNYSIDE CREEK Larva 0.9 ± 1.5 0.5 ± 0.6 27.0 ± 9.3 22.0 ± 17.5 13 Juvenile 19.1 ± 9.2 12.9 ± 7.8 53.7 ± 17.1 70.5 ± 30.4 68 Adult 18.7 ± 9.5 14.9 ± 7.1 63.5 ± 16.1 85.1 ± 17.5 209

when it was installed in 1984 and habitat was Although the range of L. albivallis has ex- unsuitable for spawning. This suggests that panded since the early 1990s, its present habi- headwater North Fork fish were ≥11 years when tat range and diversity represent a small fraction relocated downstream. Analysis of opercle of what was previously available. Even with bones from museum specimens indicated that the habitat limitations of the Flag Springs/ L. albivallis does achieve this longevity, even Sunnyside Creek system, the population was though other Plagopterini are believed to live still increasing. Though we quantified habitat only 1 to 3 years (Minckley 1973). Tracking L. use seasonally, we are hesitant to draw conclu- albivallis over 10 years in the upper North sions on habitat preference because our data Fork suggests that they live well beyond 3 are from a period of rapid population expan- years. Only adults were observed from 1986 to sion and distribution. Habitat use for each life 1988 (Donna Withers, U.S. Fish and Wildlife stage in the Flag Springs/Sunnyside Creek sys- Service, personal communication) and from tem should be investigated again when popu- 1991 to 1996. Fishes inhabiting cool water tend lation size, year class structure, and distribu- to live longer (Reimers 1979), and cool condition have stabilized. tions in headwater ponds of the North Fork Additional populations of L. albivallis need may have contributed to the longevity. An to be reestablished within the species’ historic alternate hypothesis is that the dam modified range to avoid future threats. Requisites for stream habitat such as substrate or velocity, successful L. albivallis transplanting include thus limiting or preventing recruitment. A re- extirpation of nonnative fishes and habitat lated species, L. vittata (Little Colorado River restoration. An important component of habi- spinedace), does well in pondlike conditions tat restoration is lengthening spring outflows, but requires stream habitat with fine gravel which were cut short and diverted to artificial in which to reproduce (Blinn et al. 1998). channels, leaving an insufficient stream length Regardless of the mechanism, it is likely the to sustain L. albivallis. Until other populations species would be extinct if remaining L. albi- can be established, the Flag Springs system vallis had not been moved downstream to needs to be protected and closely monitored habitat where they successfully reproduced. for reinvasion of M. salmoides and other non- Lepidomeda albivallis survives in both pond native species. As a hedge against this threat, and stream habitat and consumes a variety of the headwater pools of North Flag Spring food items, which indicates that, like L. vittata, should be managed as a Lepidomeda albivallis it is a habitat and dietary generalist (Runck refuge. and Blinn 1993). Results of both stomach content analyses and field observations have shown ACKNOWLEDGMENTS that L. albivallis, like several other Plagop- terini, actively feeds on drift in streams (Minck- We thank the U.S. Fish and Wildlife Service ley and Carufel 1967, Barber and Minckley for funding this project. This project was con- 1983, Angradi et al. 1991). Furthermore, in the ducted under state permit S 9104 and federal family Cyprinidae a gut length shorter than permit SCOGG-3. D. Withers, M. Grader, B. body length suggests carnivory (Nikolsky Nielsen, S. Shea, C. Mace, M. Fransz, M. Jenk- 1963), also consistent with drift feeding. ins, M. Whitmore, S. Lydick, and S. Reinbold 44 WESTERN NORTH AMERICAN NATURALIST [Volume 64 assisted in data collection. R. Mills, C. Lackey, MILLER, R.R., AND C.L. HUBBS. 1960. The spiny-rayed and C. Martinez assisted in L. albivallis relo- cyprinid fishes (Plagopterini) of the Colorado River system. Miscellaneous Publication of the Museum of cation. B. Nielsen analyzed fish guts. Museum Zoology, University of Michigan 115:1–39. specimens forwarded by D.W. Nelson of the MINCKLEY, W.L. 1973. Fishes of Arizona. Arizona Game University of Michigan and R. Jennings of the and Fish Department, Phoenix. University of Nevada, Las Vegas, were ex- MINCKLEY, W.L., AND L.H. CARUFEL. 1967. The Little Colo- tremely useful in adding to the species life his- rado spinedace, Lepidomeda vittata, in Arizona. South- western Naturalist 13:291–302. tory. S. Shea aided in data analysis and K. NIKOLSKY, G.U. 1963. The ecology of fishes. Academic Swaim assisted with graphics. T. Strekal, J. Press, London and New York. Smith, M. Parker, and S. Byers reviewed the REIMERS, N. 1979. A history of a stunted brook trout pop- manuscript and made helpful suggestions. The ulation in an alpine lake: a lifespan of 24 years. Cali- fornia Fish and Game 65:196–215. manuscript was improved by several anony- RINNE, W.E. 1971. The life history of Lepidomeda molli- mous reviewers. spinis mollispinis (the Virgin River spinedace), a unique western cyprinid. Unpublished master’s the- LITERATURE CITED sis, University of Nevada, Las Vegas. 109 pp. RUNCK, C., AND D.W. BLINN, 1993. Seasonal diet of Lep- idomeda vittata, a threatened cyprinid fish in Ari- ANGRADI, T.R, J.S. SPAUDING, AND E.D. KOCH. 1991. Diel food utilization by the Virgin River spinedace, Lep- zona. Southwestern Naturalist 38:157–159. idomeda mollispinis mollispinis, and speckled dace, SCOPPETTONE, G.G. 1988. Growth and longevity of the Rhinichthys osculus, in Beaver Dam Wash, Utah. cui-ui and longevity of other catostomids and cypri- Southwestern Naturalist 36:158–170. nids in western North America. Transactions of the American Fisheries Society 117:301–307. BARBER, W.E., AND W.L. MINCKLEY. 1983. Feeding ecology of a southwestern cyprinid fish, the spikedace, Meda SOKAL, R., AND F. J . R OHLF. 1995. Biometry. 3rd edition. fulgida Girard. Southwestern Naturalist 28:33–40. Freeman, New York. U.S. FISH AND WILDLIFE SERVICE. 1985. Endangered and BARBER, W.E., D.C. WILLIAMS, AND W.L. MINCKLEY. 1970. Biology of the Gila spikedace, Meda fulgida, in Ari- threatened wildlife and plants: determination of en- zona. Copeia 1970:9–18. dangered status and designation of critical habitat for the White River spinedace. Federal Register 50: BLINN, D.W., J. WHITE, T. PRADETTO, AND J. O’BRIEN. 1998. Reproductive ecology and growth of a captive popu- 37194–37198. lation of Little Colorado spinedace (Lepidomeda vit- ______. 1994a. Endangered and threatened wildlife and tata: Cyprinidae). Copeia 1998:1010–1015. plants: animal candidate review for listing as endangered or threatened species, proposed rule. Federal CASSELMAN, J.M. 1974. Analysis of hard tissue of pike Esox lucius L. with special reference to age and Register 59:58982, 59028. growth. Pages 13–27 in T.B. Bagenal, editor, Pro- ______. 1994b. White River spinedace recovery plan. U.S. ceedings of an international symposium on the age- Fish and Wildlife Service, Portland, OR. ing of fish. European Inland Fisheries Commission ______. 1997. Title 50—Wildlife and fisheries. Part 17: En- of FAO, The Fisheries Society of the British Isles dangered and threatened wildlife and plants, CFR and The Fish Biological Association, Unwin Brothers. 17.11 & 17.12. Subpart B—List. U.S. Government Printing Office, Washington, DC. DOUGLAS, M.E., P.C. MARSH, AND W.L. MINCKLEY. 1994. Indigenous fishes of western North America and the WINDELL, J.T. 1971. Food analysis and rate of digestion. hypothesis of competitive displacement: Meda fulgida Pages 215–226 in W.E. Ricker, editor, Methods for (Cyprinidae) as a case study. Copeia 1994:9–19. assessment of production in fresh waters. IBP Hand- book 3. Blackwell Scientific Publication, Oxford and LA RIVERS, I. 1962. Fish and fisheries of Nevada. Nevada Fish and Game Commission, Carson City. Edinburgh. MILLER, R.R. 1961. Man and the changing fish fauna of the American Southwest. Papers of the Michigan Received 7 February 2002 Academy of Science, Arts, and Letters 46:365–404. Accepted 25 February 2003 ______. 1963. Distribution, variation, and ecology of Lep- idomeda vittata, a rare cyprinid fish endemic to eastern Arizona. Copeia 1963:1–5. Western North American Naturalist 64(1), ©2004, pp. 45–52

A FISH SURVEY OF THE WHITE RIVER, NEVADA

G. Gary Scoppettone1, Peter H. Rissler1, and Sean Shea1

ABSTRACT.—In spring and summer 1991 and 1992, we surveyed fishes of the White River system, Nye and White Pine Counties, Nevada, to determine the status of natives. There are 5 known native fishes to the White River: Lepi- domeda albivallis (White River spinedace), Crenichthys baileyi albivallis (Preston White River springfish), Crenichthys baileyi thermophilus (Moorman White River springfish), Catostomus clarki intermedius (White River desert sucker), and Rhinichthys osculus ssp. (White River speckled dace). All 5 had declined in range. Lepidomeda albivallis had experienced the greatest decline, with less than 50 remaining, and these were restricted to a 70-m stream reach. Rhinichthys osculus spp. was most widespread, found in 18 spring systems. Cottus bairdi (mottled sculpin) was collected for the 1st time from the White River system, where it was probably native. Protective measures should be implemented to conserve all native White River fishes to include C. bairdi.

Key words: fish survey, spinedace, sculpin, springfish, desert sucker, speckled dace, White River, Colorado River, largemouth bass.

Native fishes of the White River system, Wildlife Service 1991). Nye and White Pine Counties, Nevada, are There has been no comprehensive survey endemic, and all have declined due to habitat of White River fishes since the 1930s (Miller alteration and nonnative fish introductions and Hubbs 1960), leaving the possibility of (Deacon 1979, Courtenay et al. 1985, Miller et undiscovered populations. In this paper we al. 1989). Endemism is a result of isolation report status and distribution of White River after desiccation of the pluvial White River, native fishes. which until about 10,000 years ago flowed from interior Nevada to the lower Colorado STUDY AREA River. Today’s White River is an interior basin vestige of the pluvial White River and, because The White River is the northernmost relic of the river’s prehistoric linkage, White River of the prehistoric pluvial White River, which fishes display close taxonomic affinity with flowed from east central Nevada south to the lower Colorado River fishes (Hubbs and Miller Virgin River, and then to the Colorado River 1948). (Hubbs and Miller 1948). Two other relic waters Five native fishes were known from the are Pahranagat Creek (a.k.a. Pahranagat River) White River system: Lepidomeda albivallis and Muddy River (a.k.a. Moapa River) along (White River spinedace), Rhinichthys osculus the mid- and terminal reach, respectively (Fig. ssp. (White River speckled dace), Catostomus 1). The primary water source of the 3 relic clarki intermedius (White River desert sucker), reaches is thermal springs (Eakin 1966, Gar- Crenichthys baileyi albivallis (Preston White side and Schilling 1979). River springfish), and Crenichthys baileyi ther- The largest contributing springs to the White mophilus (Moorman White River springfish). River are in the upper and lower White River Lepidomeda albivallis was, because of its rarity valley (Fig. 1). Upper valley springs are Preston and extirpation from most of its historic range, Big, Arnoldson, Nicholas, Cold, and Indian federally listed as endangered (U.S. Fish and Springs, collectively referred to as Preston Wildlife Service 1985). By 1988 it was reported Springs (Maxey and Eakin 1949), and Lund, a from only a single spring system (D. Withers, large spring several kilometers south of these U.S. Fish and Wildlife Service, personal com- (Figs. 1, 2). Cumulative discharge for these munication). The other natives also declined springs is about 0.6 m3 ⋅ s–1 (Maxey and Eakin and were considered for listing (U.S. Fish and 1949); water temperature of Preston Springs

1Biological Resources Division, U.S. Geological Survey, 1340 Financial Blvd., Suite 161, Reno, Nevada 89502.

45 46 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Fig. 1. Map of the White River System showing distribution of major springs. Inset: White River in relation to the State of Nevada and the pluvial White River.

ranges from 20°C to 23°C and Lund Spring is Springs was the last site in which L. albivallis 18.5°C. Upper White River springs were the was known to exist (D. Withers, U.S. Fish and only localities for C. b. albivallis and the only Wildlife Service, personal communication). ones with the community of L. albivallis, R. There are 4 known nonnative fish species osculus, C. c. intermedius, and C. b. albivallis established in the White River system. Poecilia (Miller and Hubbs 1960, Williams and Wilde reticulata (guppy) established in Preston and 1981). Large springs in lower White River are Lund Springs prior to 1961 (Deacon et al. 1964). Flag, Butterfield, Hot Creek, and Moon River Micropterus salmoides (largemouth bass) was Springs (Figs. 1, 3), and their cumulative dis- stocked in Adams-McGill Reservoir (Fig. 3), charge is also about 0.6 m3 ⋅ s–1 (Maxey and Eakin 1949). Hot Creek, Moon River, and which was a source of invaders to Hot Creek Moorman Springs in the middle reach of White (Courtenay et al. 1985) and Flag Springs (D. River are the warmest (30°–34°C) and are Withers, U.S. Fish and Wildlife Service, per- inhabited by C. b. thermophilus. There was no sonal communication) until fish barriers were survey record for the Butterfield Springs sys- installed. White River and Ellison Creek sup- tem. Flag Springs harbored L. albivallis, C. c. port populations of Salmo trutta (brown trout) intermedius, and R. osculus ssp. In 1988, Flag and Oncorhynchus mykiss (rainbow trout). 2004] WHITE RIVER, NEVADA, FISH SURVEY 47

Fig. 2. Map of the upper White River system showing the large springs and sampling locations on the White River and Ellison Creek.

MATERIALS AND METHODS and agricultural diversions. Perennial reaches within Humboldt National Forest (herein White White River valley perennial water sources River Humboldt National Forest and Upper include the upper White River, Ellison Creek, Ellison Creek, respectively) were managed for Water Canyon Creek (Fig. 1), and 30 spring trout, reducing the likelihood of native fish pres- systems. Sampling was during drought years ence. Thus, we electrofished only three 50-m 1991 and 1992 when perennial water in the segments in these reaches (Fig. 2). Perennial White River was intermittent and limited to segments downstream from Humboldt National the upper 25 km. Methods included snorkeling, Forest (White River Highway 6 and Lower electrofishing (Smith Root Type VII), using Ellison Creek) had warmer water without trout standard galvanized 6-mm-mesh “Gee” traps and were sampled with greater effort. We snor- baited with dry dog food, and observing from keled a 100-m reach every kilometer for 10 km banks. We report the fish species and number along White River Highway 6 and fished a encountered at each system. total of 5 minnow traps spaced 150–200 m The upper 25 km of the White River and apart on Lower Ellison Creek (Fig. 2). Ellison Creek were the most extensive systems, We sampled sufficiently deep (>5 cm) but they were intermittent due to reservoir reaches along the length of outflows of the 30 48 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Fig. 3. Map of the lower White River system showing the large springs and their relationship to reservoirs on the Wayne E. Kirch Wildlife Management Area. Inset is map of Flag/Sunnyside Springs system showing the distribution of the last known population of Lepidomeda albivallis and distribution of Micropterus salmoides in summer 1991 and 1992.

spring systems. Typically, these areas were the Figs. 1, 3). A 2-m-high dam separated the 2 spring pool where the stream emerged, an pools, and a steep gradient separated the down- earthen ditch, and in some cases a reservoir stream pool from the rest of the North Fork along or at the terminus of the channel. and the predacious Micropterus salmoides. Rhinichthys osculus ssp. was widely distrib- RESULTS uted, inhabiting 18 spring systems compared to 4 or fewer for other natives (Table 1). It was Native Fishes extirpated from Cold and Nicholas Springs. Lepidomeda albivallis was found in only 1 Greatest number captured or observed was in (Flag Springs/Sunnyside Creek) of 7 waters Indian (1105) and Preston Big (699). from which it had been collected previously Catostomus clarki intermedius was extirpated (Table 1). Only 37 individuals were sighted, all from 4 sites where it once co-occurred with of which appeared adult (>70 mm FL). They L. albivallis, C. b. albivallis, and R. osculus ssp. occupied 2 pools in the upper 70 m of the It had limited distribution and, where found, North Fork of Flag Springs (herein North Fork; it occurred in small numbers (Table 1). The 2004] WHITE RIVER, NEVADA, FISH SURVEY 49 = extirpated, P = present but not counted, = extirpated,

———— ith an asterisk (*). ______—————— —————— ——————— Native fishes Nonnative fishes ———————— ———————— ———————— ———————— ———————— ——————— ———————— ———————— ———————— ———————— ———————— C. c. C. b. C. b. C. M. P. S. O. EE— 28—— EEEE——— L. R. —————————13 ————————16— —————— 1——14 ——————163——— ————205————— ——————201——— ————<50*—95*——— ————306————— albivallis osculus intermedius albivallis thermophilus bairdi salmoides reticulata trutta mykiss ______IVER SYSTEM IVER SYSTEM R IVER SYSTEM R R HITE HITE HITE 1. Distribution and relative abundance of fishes within the White River system, Nevada. Springs without fish are not included. E W W W Preston Town Spring)Preston Town Humboldt National Forest EHighway 6Reservoirs EReservoir EReservoir E 40*Creek 225 — 1 — — 37* >800 359* — 7* — — — — 3 — — — ABLE in Can Spring# — P ter resources T Arnoldson SpringCold Spring Upper Ellison Creek Lower Ellison Creek E — 14 46 Baker Spring# E 234 —Butterfield Springs — — 618 — — 183 20 — — — — — 25 Gardner’s Ranch Spring#Indian SpringsLund SpringM Spring#Nicholas Spring (a.k.a. Preston Big Spring —Smith Creek SpringT VYZ Springs#White River 10 — E — E 1105 —Emigrant Springs 190*Hardy Springs —Moorman Springs — 64 699 39 Silver Springs 90* 6192 206 ECamp Spring E —Flag Springs/ Sunnyside Hot Creek Spring 983 —Johnson Spring#Moon River Spring — 16 117 — — — — 84 >5000 2 — — IDDLE PPER OWER — = not found. Springs named by the authors are indicated pound or number symbol (#), and entire populations noted w U M L Wa 50 WESTERN NORTH AMERICAN NATURALIST [Volume 64 greatest number, 90, was encountered in Lund headwaters of the main stem White River Spring. This spring had little emergent vege- within the Humboldt National Forest. tation and suckers were observed under roots and banks. Co-occurring native fishes were L. DISCUSSION albivallis in northernmost Flag Springs and R. osculus ssp. at all sites. Nonnative cohabitants Lepidomeda albivallis is the rarest of White were P. reticulata and M. salmoides, each at a River native fishes, with fewer than 50 fish single site. remaining and having been extirpated from 6 Crenichthys baileyi albivallis was extirpated of 7 spring systems. Catostomus clarki inter- from Cold and Lund Springs. The greatest medius exhibited the 2nd greatest decline; it number encountered was 983 in Preston Big has been extirpated from 4 spring systems and Spring and the least was 40 in Nicholas Spring. is represented by few fish where the species Co-occurring fishes were R. osculus ssp. in occurs. The rarity of these 2 species suggests Preston, Indian, and Arnoldson Springs and P. they were most sensitive to alteration of reticulata in Arnoldson and Nicholas Springs. White River system aquatic habitats. Extirpa- We observed tens of thousands of C. b. ther- tion of both from the Preston Springs complex mophilus on 10 June 1991 in the upper 100 m occurred some time after physical isolation of of Hot Creek. The population had declined individual springs in 1973, and copper sulfate markedly to a count of only 5656 C. b. ther- treatment for aquatic plant control was proba- mophilus by 25 July 1991 after an invasion of bly a contributing factor in Preston Big Spring M. salmoides. By 12 September 1991 less than (Courtenay et al. 1985). Similarly, C. clarki ssp. 50 C. b. thermophilus were observed hiding in (Pahranagat desert sucker) and Lepidomeda emergent vegetation near the spring head. altivelis (Pahranagat spinedace) were the most Crenichthys baileyi thermophilus was the only environmentally sensitive species of Pahrana- fish at Moon River and Moorman Springs. gat Creek where habitat alteration rendered L. altivelis extinct by the 1950s and C. clarki ssp. Previously Unreported Species by the late 1960s (Minckley and Deacon 1968). On 23 July 1991 we discovered a population Rhinichthys osculus ssp., on the other hand, of Cottus in the upper 120 m of Butterfield was the most widespread and abundant species Springs. We counted only 25 hiding under both in Pahranagat Creek (G.G. Scoppettone watercress (Rorippa sp.) and estimate a total unpublished data) and the White River sys- population of probably less than 100. Carl Bond, tem, attesting to its adaptability (Moyle 2002). Oregon State University, identified 9 speci- Crenichthys baileyi is thermophilic and, mens that we had collected as an unprickled consequently, localized in distribution (Williams form of Cottus bairdi. Fish (>50 mm FL) were and Wilde 1981). However, its distribution found over sandy gravel, and smaller fish over was further reduced by spring isolation associ- sandy silt. Rhinichthys osculus ssp. was the ated with water development, and the oppor- only cohabitant. tunity for genetic mixing among populations has been eliminated (J.E. Deacon, retired, Uni- Nonnative Fishes versity of Nevada, Las Vegas, personal com- Micropterus salmoides, the most widespread munication). In this study we found C. b. albi- nonnative species in the White River system, vallis in 4 isolated spring systems compared was in 5 spring systems (Table 1). It was intro- with 6 connected springs in the 1960s (Williams duced to reservoirs along Baker, Emigrant, and Wilde 1981). Last reported in Lund Spring and Silver Springs. In Hot Creek and Flag in 1984 when it was noted as rare (Courtenay Springs systems, it apparently invaded from et al. 1985), C. b. albivallis is now extirpated reservoirs on the Wayne E. Kirch Wildlife from both Lund and Cold Springs. Because Management Area, and in both systems it had Preston and Lund Springs are no longer con- crossed barriers installed to exclude it. nected, there is no opportunity for C. b. albi- Poecilia reticulata was predominant in Lund, vallis recolonization. Similarly, loss of connec- Arnoldson, and Nicholas Springs and the only tivity between Hot Creek and Moon River fish in Cold Spring (Table 1). Oncorhynchus springs eliminated gene flow between 2 popu- mykiss inhabited Upper Ellison Creek and lations of C. b. thermophilus, and close prox- Baker Spring system. Salmo trutta was in the imity of Hot Creek Spring to recreational 2004] WHITE RIVER, NEVADA, FISH SURVEY 51 reservoirs has lead to intermittent invasions of and pelvic rays left 3.3 and right 3.4) and a M. salmoides (Williams and Wilde 1981), result- thick caudal peduncle (0.091 of SL) when ing in population crashes, the most recent compared to Bonneville Basin and Colorado occurring in summer 1991. Nevada Division River forms, thus suggesting they may be of Wildlife eliminated M. salmoides from the native. Cottus bairdi has no other representa- upper reaches of Hot Creek by summer 1993, tion in the lower Colorado River system (Lee and C. b. thermophilus now number in the et al. 1980). However, the species’ propensity thousands (J. Heinrich, Nevada Division of for inhabiting cool-water streams (Bond 1963) Wildlife, personal communication). may account for its localized distribution in the The fish community (L. albivallis, R. osculus, pluvial White River and White River valley. C. c. intermedius, and C. b. albivallis) described Of the 3 relic segments of the pluvial White by Miller and Hubbs (1960) for the Preston River, Muddy River and Pahranagat Creek Springs/Lund Spring systems no longer exists. originate from warm-water springs (26.0°– The last report of co-occurrence of the 4 fishes 33.0°C; Garside and Schilling 1979), thus pre- was in 1984 in Lund Spring (Courtenay et al. cluding the presence of C. bairdi. Only the 1985), and L. albivallis and C. b. albivallis were contemporary White River system has suffi- reported as rare even then. Since 1992 there ciently cool water (15.0°–16.5°C at Butter- has been no further C. b. albivallis or R. oscu- field, North Fork Flag Springs, and the head- lus population loss from Preston Springs water White River system) to support C. (Scoppettone and Rissler 2002). In May 2002, bairdi. We believe this is the reason they are Lund Spring continued to harbor R. osculus only along this relic segment of the pluvial and C. c. intermedius, but only about 20 large White River. It is unlikely these fish were C. c. intermedius remain, suggesting an aging transported hundreds of miles from the Col- population with little to no reproduction (M.B. orado River system and across state bound- Nielsen, U.S. Fish and Wildlife Service, per- aries to be stocked or used as bait in a small sonal communication). The only notable change eastern Nevada spring. To determine if the in fish population since 1992 was removal of population is native, further research is needed M. salmoides from Hot Creek and Flag Springs to establish its taxonomic relationship with followed by a dramatic increase in native fish other populations of C. bairdi along the Col- number (J. Heinrich, Nevada Division of Wild- orado River system. If White River C. bairdi life, personal communication). are native, we would expect the C. bairdi pop- This survey presents another species, Cottus ulation in closest physical proximity (San Juan bairdi, perhaps to be added to the White River River system) to have the closest taxonomic valley native fish assemblage. In Nevada, C. affiliation, but there would be sufficient differ- bairdi had previously been documented from ence to suggest thousands of years of isolation the Bonneville Basin and Snake River system (B. May, University of California, Davis, per- (La Rivers 1962, Bond 1963, Deacon and sonal communication). This survey indicates that White River fishes Williams 1984). Cottus bairdi also is known have experienced substantial decline and are from the Colorado River system (Lee et al. in need of conservation measures to prevent 1980). In western North America, C. bairdi further loss. The species most urgently need- has been divided into 2 subspecies (Bond ing attention is L. albivallis, which is close to 1963): C. b. semiscaber of the Columbia River extinction. However, measures to improve the and Bonneville Basin, and C. b. punctulata of status of L. albivallis also will benefit its 3 his- the upper Colorado Basin (Bond 1963, Minck- toric co-occurring species. Cottus bairdi, dis- ley et al. 1986). The White River C. bairdi is covered in this survey, needs to be protected, without prickles, characteristic of C. b. punc- but managed, in Butterfield Springs until its tulata, and serves as supporting evidence that taxonomic status can be determined. it is of the Colorado River drainage and native to the White River valley. Furthermore, 9 ACKNOWLEDGMENTS White River C. bairdi examined by Carl Bond (unpublished data) indicate a tendency for This study was conducted for the State of reduced fin ray counts (mean count: dorsal ray Nevada Division of Wildlife under contract was 15.5, anal rays 11.0, pectoral rays 14.1, NV E-1-8 Job #7, state permit 12173, and 52 WESTERN NORTH AMERICAN NATURALIST [Volume 64

U.S. Fish and Wildlife Service permit North American freshwater fishes. North Carolina SCOPGG-2. D. Withers was the catalyst for State Museum of Natural History, Raleigh. MAXEY, G.B., AND T.E. EAKIN. 1949. Ground water in the survey. We thank the following people for White River Valley, White Pine, Nye, and Lincoln their assistance in surveying waters of the Counties, Nevada. Nevada Office of the State Engi- White River valley: J. Heinrich, J. Pedretti, L. neer, Water Resources Bulletin 8. Hallock, S. Byers, and H. Lawlor. K. Swaim MILLER, R.R., AND C.L. HUBBS. 1960. The spiny-rayed assisted in preparing tables and graphics. T. cyprinid fishes (Plagopterini) of the Colorado River system. Miscellaneous Publications of the Museum Strekal, the associate editor, and anonymous of Zoology, University of Michigan 115:1–39. reviewers improved the manuscript. MILLER, R.R., J.D. WILLIAMS, AND J.E. WILLIAMS. 1989. Extinctions of North American fishes during the past LITERATURE CITED century. Fisheries (Bethesda) 14:22–38. MINCKLEY, W.L., AND J.E. DEACON. 1968. Southwestern fishes and the enigma of “endangered species.” Sci- BOND, C.E. 1963. Distribution and ecology of freshwater ence 159:1424–1433. sculpins, genus Cottus, in Oregon. Doctoral disserta- MINCKLEY, W.L., D.A. HENDRICKSON, AND C.E. BOND. tion, University of Michigan, Ann Arbor. 1986. Geography of western North American fresh- COURTENAY, W.R., JR., J.E. DEACON, D.W. SADA, R.C. water fishes: description and relationships to intra- ALLAN, AND G.L. VINYARD. 1985. Comparative status continental tectonism. Pages 519–613 in C.H. Hocutt of fishes along the course of the pluvial White River, and E.O. Wiley, editors, Zoogeography of North Nevada. Southwestern Naturalist 30:503–524. American freshwater fishes. John Wiley and Sons, DEACON, J.E. 1979. Endangered and threatened fishes of New York. the West. Great Basin Naturalist Memoirs 3:41–64. MOYLE, P.B. 2002. Inland fishes of California. University DEACON, J.E., C.L. HUBBS, AND B. ZAHRANEC. 1964. Some of California Press, Berkeley. effects of introduced fishes on the native fish fauna SCOPPETTONE, G.G., AND P.H. RISSLER. 2002. Status of the of southern Nevada. Copeia 1964:384–388. Preston White River springfish (Crenichthys baileyi DEACON, J.E., AND J.E. WILLIAMS. 1984. Annotated list of albivallis). Western North American Naturalist 62: the fishes of Nevada. Proceedings of the Biological 82–87. Society of Washington 97:103–118. U.S. FISH AND WILDLIFE SERVICE. 1985. Endangered and EAKIN, T.E. 1966. A regional interbasin ground-water sys- threatened wildlife and plants: determination of en- tem in the White River area, southeastern Nevada. dangered status and designation of critical habitat Nevada Department of Conservation and Natural for the White River Spinedace. Federal Register 50: Resources, Water Resources Bulletin 33:251–271. 37194–37198. GARSIDE, L.J., AND J.H. SCHILLING. 1979. Thermal waters ______. 1991. Endangered and threatened wildlife and of Nevada. Nevada Bureau of Mines and Geology, plants: animal candidate review for listing as endan- Bulletin 91, Mackay School of Mines, University of gered or threatened species, proposed rule. Federal Nevada, Reno. Register 56:58804–58836. HUBBS, C.L., AND R.R. MILLER. 1948. The Great Basin WILLIAMS, J.E., AND G.R. WILDE. 1981. Taxonomic status with emphasis on glacial and postglacial times. II. and morphology of isolated populations of the White The zoological evidence. University of Utah Bulletin River springfish, Crenichthys baileyi (Cyprinodonti- 38:17–166. dae). Southwestern Naturalist 25:485–503. LA RIVERS, I. 1962. Fish and fisheries of Nevada. Nevada Fish and Game Commission, Carson City. Received 7 February 2002 LEE, D.S., C.R. GILBERT, C.H. HOCUTT, R.E. JENKINS, D.E. Accepted 17 January 2003 MCALLISTER, AND J.R. STAUFFER, JR. 1980. Atlas of Western North American Naturalist 64(1), ©2004, pp. 53–58

COYOTE (CANIS LATRANS) MOVEMENTS RELATIVE TO CATTLE (BOS TAURUS) CARCASS AREAS

Jan F. Kamler1,2, Warren B. Ballard1,3, Rickey L. Gilliland4, and Kevin Mote5

ABSTRACT.—Use of 2 cattle carcass areas was determined for radio-collared coyotes (Canis latrans) in northwest Texas from January 1999 to January 2000. When 0–3 dead cattle were located at the carcass areas, resident and transient coyotes visited the carcass areas 4% and 8% of the time, respectively. However, when 30–35 dead cattle were located at 1 carcass area due to a disease epizootic, resident and transient coyotes had increased visitation rates of 19% and 63%, respectively. Resident coyotes traveled as far as 12.2 km from the center of their home ranges, suggesting that carcass areas influenced residents over a 468-km2 area. Transient coyotes traveled from as far as 20.5 km away, suggesting that carcass areas influenced transients over a 1320-km2 area. Our results indicate that carcass areas can influence coyotes over large areas and may concentrate both resident and transient coyotes in relatively small areas, at least for short periods.

Key words: coyote, Canis latrans, cattle, carrion, home range, movements, Texas.

Coyotes (Canis latrans) have omnivorous ulations can be useful information. Although diets and are opportunistic feeders (Bekoff many large-scale livestock operations and cat- 1982). Carrion is often an important part of tle feed yards dispose of dead livestock in car- coyote diets, especially in more northern areas cass areas, the use of carcass areas by coyotes (Camenzind 1978, Bekoff and Wells 1980, has rarely been reported (Danner and Smith Bowen 1981). Large amounts of carrion have 1980). We conducted a study of coyote ecology been shown to concentrate coyote numbers in in northwestern Texas, where many private winter (Camenzind 1978, Bekoff and Wells 1980, ranches contain cattle carcass areas. We Bowen 1981). However, little information exists attempted to assess the influence of 2 carcass concerning the influence of carrion concentra- areas on the surrounding coyote population, tions on movements of local coyote popula- especially when an unusually high number of tions. Furthermore, coyotes have been classi- dead cattle (n > 30) were located at 1 of the fied in their social organization as residents carcass areas. We determined visitation rates and transients (Messier and Barrette 1982, and distances traveled to carcass areas for Andelt 1985, Gese et al. 1988, Kamler and both resident and transient coyotes. Gipson 2000), and the effects of carrion concentrations on different social classes are not STUDY AREA known. For example, residents that have established territories may be less influenced by Our study site is a 9000-ha area of Rita carrion concentrations than transients. Resi- Blanca National Grasslands that is interspersed dents cannot leave their defended territories with private lands in west central Dallam to obtain carrion (Hein and Andelt 1996), as County, Texas. Rangeland vegetation consists of excursions leave their territory vulnerable to short-grass prairie species dominated by blue other coyotes (Kamler and Gipson 2000). grama (Bouteloua gracilis) and buffalograss Coyotes can be effective predators of live- (Buchloe dactyloides) that was moderately to stock, especially calves and sheep (Gier 1968, intensively grazed by cattle (Bos taurus). The Andelt 1987, Gilliland 1995, Knowlton et al. site is adjacent to an 11,000-ha private cattle 1999). Therefore, data concerning the effects ranch to the northwest and a 3400-ha private of particular ranching practices on coyote pop- cattle ranch to the south. Both cattle ranches

1Department of Range, Wildlife, and Fisheries Management, Box 42125, Texas Tech University, Lubbock, TX 79409. 2Present address: Mammal Research Institute, Polish Academy of Sciences, 17–230 Bailowieza, Poland. 3Corresponding author. 4USDA Wildlife Services, Box 60277, West Texas A&M University, Canyon, TX 79016. 5Texas Parks and Wildlife Department, 301 Main St., Suite D, Brownwood, TX 76801.

53 54 WESTERN NORTH AMERICAN NATURALIST [Volume 64 contained a shallow pit where dead cattle were tracts after death. All resident males were con- placed to be scavenged (hereafter, carcass area). sidered breeders because all were >2 years Potential mammalian prey that occurred on and appeared reproductively active (e.g., had the study site include pronghorn (Antilocapra descended testes) when captured. americana), black-tailed jackrabbits (Lepus cali- We recorded telemetry locations for each fornicus), desert cottontails (Sylvilagus audo- coyote 1–2 times per week and >12 hours bonii), black-tailed prairie dogs (Cynomys apart to establish independence (White and ludovicianus), Ord’s kangaroo rats (Dipodomys Garrott 1990). Coyote locations were triangu- ordii), ground squirrels (Spermophilus spp.), lated using 2–3 bearings obtained <5 minutes gophers (Geomys and Cratogeomys spp.), prairie apart. We radio-tracked from a vehicle using a voles (Microtus ochrogaster), hispid cotton rats null-peak telemetry system, which consisted (Sigmodon hispidus), northern grasshopper mice of dual, 4-element Yagi antennas. We conducted (Onychomys leucogaster), woodrats (Neotoma radio-tracking primarily during 1800–0900 spp.), pocket mice (Chaetodipus and Perog- hours, when coyotes were likely to be most nathus spp.), harvest mice (Reithrodontomys active (Andelt 1985). We calculated location spp.), and Peromyscus spp. (Lemons 2001). estimates using the maximum likelihood esti- mation option in the program Locate II (Pacer, METHODS Inc., Truro, Nova Scotia, Canada). Mean errors for reference collars (known locations) were 84 We captured and radio-collared 12 coyotes m (95% of errors were <145 m). (6 male, 6 female) from January to April 1999, We determined home range sizes of resi- using No. 3 Victor Soft Catch® traps (Wood- dent coyotes using the 95% minimum convex stream Corp., Lititz, PA) equipped with the polygon (MCP) method (Mohr 1947), as calcu- Paws-I-Trip™ pan tension system (M-Y Enter- lated by the Animal Movement program (Hooge prises, Homer City, PA). Trap sets were baited and Eichenlaub 1997). We calculated home with a variety of baits, urines, and lures and ranges of resident coyotes with >40 locations checked once daily. We immobilized coyotes and >9 months of radio-tracking. Because 2 with an intramuscular injection of ketamine resident coyotes were killed before 40 loca- hydrochloride and xylazine hydrochloride (10:1 tions were obtained, we estimated their mini- ratio; dosage = 1 mL ⋅ 10 kg–1 of body weight). mum home ranges with 22 and 23 locations, At time of capture we classified coyotes as respectively. We did not calculate home ranges adult (>2 years), yearling (1–2 years), or juve- for transient coyotes because they were not nile (<1 year) based on body size, reproduc- radio-tracked continuously throughout the year tive condition, and wear on teeth (Gier 1968). and had <30 total locations. Coyotes were equipped with radio-transmitter We determined the frequency of each coy- collars (Advanced Telemetry Systems, Inc., ote’s visits to the carcass areas by dividing the Isanti, MN) and released at their capture sites. number of times they were located at the car- Due to an early death, 1 female coyote was not cass areas by total number of locations. We monitored long enough to be used in analyses. calculated frequencies for 9 months of the We classified coyotes as resident or transient year (mid-January to September 1999) when based on the following criteria: resident coyotes, 0–3 dead cattle were at the carcass areas at which consisted of breeders (mated pair) and any given time (R. Wyatt and B. Burnacki, often pack associates (yearling offspring that Dalhart, TX, personal communication), and have not dispersed but do not breed), lived in for 3 months of the year (mid-October 1999 to family groups with relatively small home ranges mid-January 2000) when approximately 30–35 that overlapped little with other resident groups dead cattle were at the carcass area on the (Andelt 1985, Gese et al. 1988, Kamler and southern ranch (B. Burnacki personal commu- Gipson 2000); transient coyotes were solitary nication). Approximately 50–60 cattle died be- with relatively large, poorly defined home cause of a disease epizootic over that 3-month ranges that overlapped those of other coyotes period on the southern cattle ranch, and (Andelt 1985, Gese et al. 1988, Kamler and approximately 30–35 carcasses were in the Gipson 2000). For resident females, breeders carcass area at any given time. We determined were distinguished from pack associates by distances that resident coyotes traveled to the inspection of their nipples and reproductive carcass areas by measuring the distance from 2004] COYOTE USE OF CARCASS AREAS 55

Fig. 1. Home ranges of 7 resident coyotes (solid polygons) and estimated home ranges of 2 resident coyotes (dashed polygons) on public and private lands in Dallam County, Texas, January 1999–January 2000. Two large dots represent the carcass areas on private lands to the south (A) and north (B). Smaller dots represent centers of coyote home ranges and lines represent straight-line distances to the carcass areas.

the center of their home ranges to the carcass carcass area (Fig. 1). Two resident coyotes areas. We determined distances for transient never traveled to either the northern (14.8 and coyotes by measuring the distance between 16.9 km away) or southern (12.2 and 17.2 km their farthest locations and the carcass areas. away) carcass areas (Fig. 1). Because resident We estimated total areas that carcass areas coyotes traveled a maximum distance of 12.2 may influence resident and transient coyotes km, the carcass area affected resident coyotes by using the maximum distance traveled as over a 468-km2 area. the radius in the area-of-circle equation (area Five transient coyotes traveled a mean dis- of circle = πr2). tance of 15.9 ± 1.8 km (range = 12.8–20.5 km) to the southern carcass area, whereas no tran- RESULTS sient coyotes were recorded traveling to the northern carcass area. Because transient coy- ± The mean ( sx–) home range size of 7 resi- otes traveled a maximum distance of 20.5 km, dent coyotes (3 females, 4 males) was 12.5 ± the carcass area affected transient coyotes 0.4 km2. We also estimated minimum area of over a 1320-km2 area. use for 2 additional residents (1 female, 1 male) When 0–3 dead cattle were in the 2 carcass with 20–25 locations (Fig. 1). One resident areas during a 9-month period, both residents female was classified as a pack associate, and transients had relatively low mean visita- whereas all other residents were classified as tion rates of 4% and 8%, respectively (Table 1). breeders. Two resident coyotes, including the When >30 dead cattle were placed in the pack associate, changed social status and southern carcass area during a 3-month period, became transients, similar to changes reported both resident and transient coyotes had in- by others (Andelt 1985, Kamler and Gipson creased mean visitation rates of 19% and 63%, 2000). From the center of their home ranges, 3 respectively (Table 1). The highest visitation ± resident coyotes traveled a mean ( sx–) dis- rate among residents (16%) was exhibited by tance of 10.6 ± 1.0 km (range = 8.7–12.2 km) the only known pack associate. This pack asso- to the northern carcass area, whereas 6 resi- ciate dispersed and became a transient in dent coyotes traveled a mean distance of 7.9 ± October and subsequently spent a majority of 1.3 km (range = 3.7–11.5 km) to the southern her time (76%) at the southern carcass area. 56 WESTERN NORTH AMERICAN NATURALIST [Volume 64

TABLE 1. Visitation ratesa (%) of resident and transient coyotes to cattle carcass areas in Dallam County, Texas, 1999–2000.

______Jan.–Sept. (0–3 dead cattle) ______Oct.–Jan. (>30 dead cattle) ± ± Social class n Mean sx– n Mean sx– (%) (%) Resident 9 4.4 ± 1.8 6 19.3 ± 7.5 Transient 2 7.7 ± 1.8 2 63.1 ± 13.1 aVisitation rate = number of visits / total locations.

DISCUSSION 2000). The relatively high frequency of visitations by resident coyotes in our study suggests Cattle carcass areas can influence coyotes that benefits of obtaining large amounts of car- over relatively large areas. Visitation rates of rion must have outweighed the risk of losing both resident and transient coyotes were rela- their territory to other coyotes. tively low during most of the year when few, if Previous research indicates that resident any, dead cattle were in carcass areas. How- coyotes might temporarily leave their home ever, during periods when large amounts of ranges to obtain carrion (Bekoff and Wells 1980, carrion were present, visitation rates increased Bowen 1981). However, Hein and Andelt (1996) more than fourfold for resident coyotes and found that resident coyotes did not visit car- more than eightfold for transient coyotes. Addi- rion when it was placed outside their home tionally, during early January 2000, the United ranges. Reasons for differences among studies States Department of Agriculture’s Wildlife are unknown, although they might have been Services initiated a coyote control program at due to local food resources. Consumption of our study site for another research project. carrion by coyotes generally increases when During 2 days of aerial-shooting coyotes from local food resources decrease (Bekoff and Wells a fixed-wing aircraft, 44 coyotes were killed in 1980, Bowen 1981, Todd and Keith 1983). Thus, an approximately 2-km2 area surrounding the although we did not determine abundances of southern carcass area. Thus, large amounts of local prey species on our study site, relatively carrion caused coyotes to congregate in rela- low numbers of local prey might have encour- tively high numbers around the carcass area. aged resident coyotes to temporarily leave Although >30 dead cattle may be unusual for their home ranges to obtain additional food most private ranches, this number is not resources. unusual for cattle feed yards that commonly Transient coyotes may have benefited more dispose of dead cattle in carcass areas (Danner from carrion than residents, as transients were and Smith 1980). influenced over a larger area and had higher The pack associate had the highest visita- visitation rates than residents. Additionally, tion rate of all residents, and she later became when large amounts of carrion were present, a transient, subsequently spending most of her transients spent a majority of their time near time at the southern carcass area. This was not the carcass area, most likely because of the surprising, as previous research indicates that abundant food resources and lack of territori- pack associates are more fluid than breeders ality by other coyotes. Transient coyotes, which in their resident status (Sacks et al. 1999). can be old, disabled, or young coyotes that Pack associates often make more forays away have just dispersed, generally have lower sur- from territories than breeders, probably as a vival than residents and are often excluded precursor to finding their own territory (Sacks from optimal resources by territorial residents et al. 1999, Kamler and Gipson 2000). More (Kamler and Gipson 2000, Gese 2001). How- surprisingly, though, was that resident breed- ever, large amounts of carrion can break down ers (mated pairs) periodically left their home social structure of local resident coyotes, as ranges to visit cattle carcass areas during our local family groups cannot defend their terri- study. Resident breeders rarely leave their tories against high numbers of intruding coy- home ranges, as excursions put their territory otes (Camenzind 1978). Thus, lack of territori- at risk to other coyotes (Kamler and Gipson ality around large amounts of carrion allows 2004] COYOTE USE OF CARCASS AREAS 57 transients access to abundant and easily acces- ACKNOWLEDGMENTS sible food resources (Camenzind 1978), which could potentially increase their fitness and Research was funded by Texas Tech Uni- survival. This likely occurred at our study site, versity, Texas Parks and Wildlife Department, as the high numbers of coyotes killed near the U.S. Forest Service, U.S. Fish and Wildlife carcass area and high visitation rates of radio- Service, and USDA—Wildlife Services. We collared coyotes suggest that territoriality thank D. Bowers, B. Burnacki, C. Chisum, P. broke down in areas immediately surrounding Guile, C. Lawrence, J. Lobley, C. Murdock, the carcass site. and R. Wyatt for access to their land. We also Our results were similar to those found by thank C.C. Perchellet for assistance with the Danner and Smith (1980), who reported that a project. B.N. Sacks and an anonymous reviewer continual supply of livestock carrion from a provided helpful comments on this paper. This feed yard influenced coyote movements over a is Texas Tech University, College of Agricul- 380- to 700-km2 area, as radio-collared coy- tural Sciences and Natural Resources techni- otes traveled from as far as 15.3 km. In that cal publication T-9-953. study, immature coyotes, which are often transients, visited the carrion site 5 times more LITERATURE CITED often than adults, which are often residents. ANDELT, W.F.1985. Behavioral ecology of coyotes in south Although the supply of carrion near our study Texas. Wildlife Monograph 94:1–45. site was not continual, our results indicate that ______. 1987. Coyote predation. Pages 128–140 in M. even short-term supplies of carrion have an Novak, J.A. Baker, M.E. Obbard, and B. Malloch, edi- immediate impact on the surrounding coyote tors, Wild furbearer management and conservation in North America. Ministry of Natural Resources, population, at least with respect to overall Toronto, Ontario, Canada. movements. Additionally, our results demon- BEKOFF, M. 1982. Coyote. Pages 447–459 in J.A. Chapman strate that carcass areas may influence coyotes and G.A. Feldhamer, editors, Wild mammals of North over larger areas more than previously reported. America. Johns Hopkins University Press, Balti- more, MD. Our results also have implications for live- BEKOFF, M., AND M.C. WELLS. 1980. The social ecology of stock producers because large amounts of car- coyotes. Scientific American 242:130–148. rion can increase livestock losses in 2 ways: by BOWEN, W.D. 1981. Variation in coyote social organiza- habituating coyotes to feed on livestock flesh tion: the influence of prey size. Canadian Journal of and by increasing and concentrating local coy- Zoology 59:639–652. CAMENZIND, F.J. 1978. Behavioral ecology of coyotes on ote densities around livestock (Green et al. the National Elk Refuge, Jackson, Wyoming. Pages 1994). Habituating coyotes to feed on livestock 267–294 in M. Bekoff, editor, Coyotes: biology, be- carrion can lead to coyotes’ developing a taste havior and management. Academic Press, New York. for livestock and could consequently lead to DANNER, D.A., AND N.S. SMITH. 1980. Coyote home range, movement, and relative abundance near a cattle actual predation instead of scavenging (Fichter feedyard. Journal of Wildlife Management 44:484–487. et al. 1955, Gier 1968, Phillips and Hubert FICHTER, E., G. SCHILDMAN, AND J.H. SATHER. 1955. Some 1980). Green et al. (1994) indicated that where feeding patterns of coyotes in Nebraska. Ecological carrion is generally not available, livestock Monograph 25:1–37. GESE, E.M. 2001. Territorial defense of coyotes (Canis losses are lower, and they concluded that car- latrans) in Yellowstone National Park, Wyoming: rion removal is an important method of dam- who, how, where, when, and why. Canadian Journal age prevention to reduce livestock losses to of Zoology 79:980–987. coyotes. Our results support those of Green et GESE, E.M., O.J. RONGSTAD, AND W.R. MYTTON. 1988. Home range and habitat use of coyotes in southeast- al. (1994), inasmuch as large amounts of car- ern Colorado. Journal of Wildlife Management rion increased and concentrated local num- 52:640–646. bers of coyotes on a private ranch south of our GIER, H.T. 1968. Coyotes in Kansas. Revised. Kansas State study site. Whether coyotes became habitu- College Agricultural Experiment Station Bulletin ated to feeding on livestock flesh and conse- 393. GILLILAND, R.L. 1995. Predation impacts and management quently increased livestock predation was strategies for reducing coyote damage to cattle. Pages unknown. If coyotes did become habituated to 124–128 in D. Rollins, C. Richardson, T. Blanken- cattle flesh, then increases in livestock losses ship, K. Canon, and S. Henke, editors, Coyotes in could have occurred over a relatively large the Southwest: a compendium of our knowledge. Texas Parks and Wildlife Department, Austin. area, as coyotes traveled from as far as 20.5 km GREEN, J.H., F.R. HENDERSON, AND M.D. COLLINGE. 1994. to consume cattle carrion. Coyotes. Pages 51–76 in S.E. Hygnstrom, R.M. 58 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Timm, and G.E. Larson, editors, Prevention and MOHR, C.O. 1947. Table of equivalent populations of North control of wildlife damage. University of Nebraska, American small mammals. American Midland Natu- Lincoln. ralist 37:223–249. HEIN, E.W., AND W. F. A NDELT. 1996. Coyote visitations to PHILLIPS, M.K., AND G.F. HUBERT, JR. 1980. Winter food experimentally-placed deer carrion. Southwestern habits of coyotes in southeastern Illinois. Transac- Naturalist 41:48–53. tions of the Illinois State Academy of Science 73: HOOGE, P.N., AND B. EICHENLAUB. 1997. Animal move- 80–84. ment extension to ArcView, version 1.1. Alaska Bio- SACKS, B.N., M.M. JAEGER, J.C.C. NEALE, AND D.R. logical Science Center, U.S. Geological Survey, MCCULLOUGH. 1999. Territoriality and breeding sta- Anchorage. tus of coyotes relative to sheep predation. Journal of KAMLER, J.F., AND P. S . G IPSON. 2000. Space and habitat Wildlife Management 63:593–605. use by resident and transient coyotes. Canadian TODD, A.W., AND L.B. KEITH. 1983. Coyote demography Journal of Zoology 78:2106–2111. during a snowshoe hare decline in Alberta. Journal KNOWLTON, F.F., E.M. GESE, AND M.M. JAEGER. 1999. of Wildlife Management 47:394–404. Coyote depredation control: an interface between WHITE, G.C., AND R.A. GARROTT. 1990. Analysis of radio- biology and management. Journal of Range Manage- tracking data. Academic Press, Inc., San Diego. ment 52:398–412. LEMONS, P.R. 2001. Swift fox and coyote interactions in Received 22 August 2002 the short-grass prairie of northwest Texas: competi- Accepted 4 February 2003 tion in diets and den site activity. Master’s thesis, Texas Tech University, Lubbock. MESSIER, F., AND C. BARRETTE. 1982. The social system of the coyote (Canis latrans) in a forested habitat. Canadian Journal of Zoology 60:1743–1753. Western North American Naturalist 64(1), ©2004, pp. 59–71

OCCURRENCE AND HABITAT USE OF PASSAGE NEOTROPICAL MIGRANTS IN THE SONORAN DESERT

Paul C. Hardy1, David J. Griffin2, Amy J. Kuenzi3, and Michael L. Morrison4

ABSTRACT.—Little is known about stopover habitat use by neotropical migratory birds in the deserts of North Amer- ica. We determined distribution, abundance, and habitat associations of neotropical migrants during spring migration in the Sonoran Desert of southwestern Arizona along large washes that supported xeroriparian scrub vegetation. We detected 91 bird species during surveys, 50 (52%) of which were passage neotropical migrants. Although xeroriparian scrub covered less than 55% of the area surveyed, 97% of all detections of passage migrants were from this vegetation type. By calculating habitat breadth for each species, we classified 87% of passage migrants as xeroriparian specialists. Richness of passage migrants was strongly associated with the presence of overstory (>2.5 m) mesquite and paloverde. The highest species richness of breeding neotropical migrants was associated with width of the xeroriparian corridor. Habitat characteristics we have shown to be important to neotropical migrants can be preserved and managed by protecting xeroriparian areas, particularly those supporting mature (>2.5 m) paloverde, mesquite, desert willow, and catclaw acacia trees. Additionally, xeroriparian scrub within the creosote-bursage vegetation type may be particularly important to passage neotropical migrants.

Key words: bird abundance, desert, habitat use, migration, Southwest, xeroriparian.

Neotropical migrants generally do not store 1998). Most avian studies conducted in xero- enough lipids to fly nonstop between wintering riparian zones (i.e., arroyos or “washes” rarely and breeding areas (Blem 1980, Moore et al. receiving surface flow) of the Southwest have 1995). Consequently, suitable stopover habitat focused on breeding birds and have not that enables passage migrants to replenish lipid addressed the ecology of passage migrants deposits, repay oxygen debt, and repair dam- (Hensley 1954, Raitt and Maze 1968, Austin aged tissues is critical to successful migration 1970, Vander Wall and MacMahon 1984, Parker (Moore et al. 1993, Skagen et al. 1998). Spring 1986). In their study of vertebrate use of 2 migrants unable to restore lipids or repair water developments in southwestern Arizona, damaged tissues rapidly could arrive late on Cutler (1996) and Cutler and Morrison (1998) the breeding grounds, potentially losing terri- reported on the habitat use of 22 species of tories and mates; those unable to properly re- birds, 8 of which were neotropical migrants. cover may be exposed to increased predation In the Sonoran Desert large xeroriparian and encounter higher mortality en route (Moore zones with dense growths of shrubs and trees et al. 1993). Little is known about the ecology of west- provide feeding areas for many insectivorous ern neotropical migrants in general and neo- bird species and may also be used as travel tropical migrants of the southwestern deserts and migration corridors (Hensley 1954, Rosen- in particular (Rosenberg et al. 1991, Moore et berg et al. 1991). We studied stopover habitat al. 1995). In the southwestern U.S., over 60% use by neotropical migrants in the Sonoran of neotropical migrants are known to use mesic Desert of southwestern Arizona to (1) deter- riparian areas for stopovers or for breeding mine the distribution, abundance, and habitat (Krueper 1993). Mesic riparian areas of the associations of passage neotropical migrants Southwest provide cover, food, and water in using xeroriparian washes; and (2) develop regions where these critical elements are management recommendations for conserving scarce (Wauer 1977, Gori 1992, Skagen et al. and monitoring passage neotropical migrants.

1Feather River Land Trust, Box 1826, Quincy, CA 95971. 2San Diego National Wildlife Refuge, U.S. Fish and Wildlife Service, 13910 Lyons Valley Road, Suite R, Jamul, CA 91935. 3Biology Department, Montana Tech, Butte, MT 59701. 4Corresponding author. White Mountain Research Station, University of California, 3000 East Line St., Bishop, CA 93514.

59 60 WESTERN NORTH AMERICAN NATURALIST [Volume 64

STUDY AREA and mixed-cacti vegetation types. Each transect was 3 km long and had 10 count stations spaced We conducted our study in the Sauceda at 300-m intervals, for a total of 30 stations. Mountains of southwestern Arizona, on the We collected data at each station during 3- Barry M. Goldwater Air Force Range (BMGR), day sampling periods, spaced 10–14 days apart, Maricopa County, Arizona. The BMGR con- between late February and early June 1994 tains some of the largest undeveloped Sonoran through 1996. This interval encompassed the Desert in North America and is crossed by spring migration period. We counted birds at many large, dry washes, which support xero- each station for 5 minutes and conducted all riparian scrub vegetation. The climate of our surveys within 4 hours after sunrise. We alter- study area is typical of the Sonoran Desert and nated starting points along transects to avoid a is characterized by high summer temperatures temporal bias between visits. We recorded all (mean 38°C), warm winter temperatures (mean species of birds seen or heard, means of detec- 11°C), and low rainfall (Sellers et al. 1985). tion (auditory, visual, or both), estimated dis- The rainfall pattern is bimodal and averages tance from observer (0–25 m, 25–50 m, 50–100 225 mm per year at Ajo, Arizona, 30 km from m, 100–150 m, 150–200 m, or >200 m), location the study area. Precipitation occurs as rain of bird (paloverde, mesquite, snag; floristic/ mostly in July, August, and September during substrate model, see below), activity of bird short, intense thunderstorms and also falls as (nesting, feeding, resting), number of individ- prolonged, soaking showers from December uals observed, and general vegetation type in through March. No permanent water occurred which the bird was observed (xeroriparian scrub, on our study area. creosote-bursage, mixed-cacti, or rocks/cliffs). The study area ranges in elevation from We assigned vegetation types by visually de- 375 m to 620 m and contains vegetation from termining the dominant vegetative character- both the Lower Colorado River Valley and istics of the area. The dry xeroriparian scrub Arizona Upland subdivisions of Sonoran Desert vegetation type was dominated by tall (>2 m), scrub (Turner and Brown 1994). Typical vege- dense vegetation bordering washes. The cre- tation of the study area includes creosotebush osote-bursage vegetation type was dominated (Larrea tridentata), bursage (Ambrosia dumosa by creosotebush and bursage. The mixed-cacti and A. deltoidea), paloverde (Cercidium micro- vegetation type was dominated by yellow palo- phyllum and C. floridum), saguaro (Carnegiea verde (C. microphyllum), saguaro, and various gigantea), and cholla cactus (Opuntia spp.; small shrubs and cacti; this vegetation type Turner and Brown 1994). Large associations of generally occurred in areas with greater slopes xeroriparian scrub occur along washes within than the xeroriparian scrub and creosote-bur- the study area and are dominated by paloverde, sage vegetation types. The rocks/cliffs vegeta- ironwood (Olneya tesota), mesquite (Prosopis tion type consisted of areas that were bare velutina), catclaw acacia (Acacia greggii), desert rock or steep-sloped cliffs or both. willow (Chilopsis linearis), desert hackberry (Celtis pallida), and burro-bush (Hymenoclea Vegetation Sampling monogyra; Turner and Brown 1994). We sampled vegetation and substrate along 100-m vegetation transects that radiated from METHODS each count station using an adaptation of the point-intercept method (Karr 1968). The direc- Point Counts tion of the first transect was chosen randomly We surveyed neotropical migratory birds at and remaining transect directions were placed 3 study sites using point counts (Verner 1985, at compass increments of 45°, for a total of 8 Ralph et al. 1993). We placed 3 transects in transects radiating from each count station. large washes that were bordered by xeroripar- We located sampling points every 5 m along ian scrub and were readily accessible. One each vegetation transect, for a total of 160 transect was located at lower elevations (~380 points per count station. m) in the creosote-bursage vegetation type, At each vegetation sampling point, we visu- whereas the other 2 were located at slightly ally assigned vegetation type along the transect higher elevations (i.e., ~460 m and ~620 m) in according to general vegetation characteristics areas ecotonal between the creosote-bursage into 1 of 4 categories: (1) xeroriparian scrub, 2004] PASSAGE NEOTROPICAL MIGRANTS 61

(2) mixed-cacti, (3) creosote-bursage, or (4) 1996). Only detections <150 m from a counting rocks/cliffs. For each count station, we calcu- point were used in calculations to lessen dou- lated percent cover of each vegetation type as ble-counting of individuals; all observations the frequency of each vegetation type divided were used to determine species richness. by the total number of sampling points (see We used multiple regression analysis (Zar below vegetation type model). 1996) to determine which combination of veg- At each vegetation sampling point, we also etation and substrate variables best predicted recorded the presence of plant species in each the abundance and richness of passage mi- of 5 height categories: (1) 0.0–0.1 m (ground grants. We used pooled indices of abundance level), (2) >0.1–1.0 m (understory), (3) >1.0– and pooled richness (1994–1996) as response 2.5 m (mid-canopy), (4) >2.5–6.0 m (overstory), variables. and (5) >6.0 m (tall overstory). We calculated The scale at which variables are measured percent cover by height category for each influences the explanatory power of models to dominant plant species by dividing the fre- predict bird-habitat relationships (Morrison et quency of occurrence of live foliage for each al. 1998). Consequently, for each response height interval by the total number of sam- variable we built 3 separate multiple regres- pling points. We also calculated percent cover sion models using explanatory variables from of perennial vegetation by height category, as different scales: a cover type model, a peren- well as the total cover of perennial vegetation nial vegetation model (plant species not taken (see below perennial vegetation model). into account), and a floristic/substrate model (cover of plant species and substrate). Analyses We used stepwise selection procedures to Following Gauthreaux (1992), we classified build all multiple regression models. Prior to each species detected during point counts as 1 building the microhabitat models, we reduced of 2 types of neotropical migrants, or as non- the data set by eliminating all variables for migratory. List A neotropical migrants included which the frequency of occurrence was small those species that breed in North America and (≤5 cases per variable) and then tested all vari- spend their entire nonbreeding season primar- ables for multicolinearity (Pearson r). We reily south of the U.S. List B neotropical migrants tained 1 member of each highly intercorrelated included those species that breed and winter pair (r ≥ 0.7) judged to be more biologically extensively in North America but also have significant and easiest to measure (Norusis populations that winter south of the U.S. All 1990). To control for potential differences be- other species were classified as nonmigratory. tween transects, an indicator variable for tran- Based on literature review and personal sect and all 2-way interaction terms involving observation, we further classified neotropical transect were also considered for entry into all migrants as passage or nonpassage. Species models (Belsley et al. 1980). We used a P-value classified as passage migrants were not pre- of 0.25 to determine which variables entered sent during the 1st sampling period each year, into the model and a P-value of 0.10 to deter- seldom sang, and did not persist in the study mine which variables were removed (Belsley area for extended periods (present ≤4 weeks). et al. 1980). For all MR models, we evaluated These species were usually detected individu- the assumptions of linearity and homogeneity ally or in small (<10 individuals) mixed- or of variances by examining scatterplots of stan- single-species flocks. The analyses described dardized residuals, and the assumption of nor- below are for passage neotropical migrant mality of residuals by examining a histogram of species only. residuals (Belsley et al. 1980). When necessary, We calculated indices of passage migrant we used log, square root, and arcsine transforma- abundance (number of individuals detected tions (Norusis 1990) to meet these assumptions. per number of surveys) for each count station. We used Levins’ (1968) measure of habitat We calculated these indices for each year of breadth to classify passage migrants by degree survey data and for the pooled data set (1994– of specialization. We first controlled for differ- 1996 combined). We calculated species rich- ences in the availability of vegetation types by ness (total number of passage migrant species weighting (dividing) the number of birds de- detected) at each count station for each year tected in a given vegetation type by the pro- separately and for the pooled data set (1994– portion of each vegetation type along transects. 62 WESTERN NORTH AMERICAN NATURALIST [Volume 64

This generated a distribution of detections for creasing cover of overstory perennial vegeta- each species that would be expected if survey tion and with decreasing ground cover of effort were equal among vegetation types. A perennial vegetation. species with a proportion of detections across The floristic/substrate model to predict 2 the 4 vegetation types that matched the avail- species richness was moderately strong (R adj ability of vegetation types had the broadest = 0.583, P < 0.001) and included variables for possible habitat breadth (B = 4.0), whereas a overstory mesquite, overstory desert willow, species restricted to any 1 of the 4 vegetation rock, and vegetative litter. Increasing species types had the narrowest possible habitat richness was significantly associated with in- breadth (B = 1.0). creasing cover of overstory mesquite and desert We used goodness-of-fit G-tests (Zar 1996) willow and with decreasing cover of rock and to determine if passage migrants used micro- litter. phyllous tree species (i.e., catclaw acacia, desert INDICES OF ABUNDANCE.—The vegetation willow, ironwood, mesquite, and paloverde) type model explained a substantial amount of 2 out of proportion to their availability. We mea- variation in passage migrant abundance (R adj sured availability as the percent cover of each = 0.62, P < 0.001). After controlling for differ- tree species relative to the total percent cover ences in cover of vegetation types among tran- of all tree species, and considered locations sects, we noted that increasing abundance of of individual birds to be the sample units. passage migrants was significantly associated When significant differences were indicated, with decreasing cover of the creosote-bursage we used Bailey’s simultaneous confidence in- vegetation type. tervals (Cherry 1996) to identify which tree The perennial vegetation MR model ex- species were used disproportionately. plained a moderate amount of variation in pas- α 2 A level of significance of = 0.10 was used sage migrant abundance (R adj = 0.32, P < for all statistical analyses. We used α = 0.10 0.001). Species abundance significantly in- instead of α = 0.05 to improve the power of creased with increasing cover of overstory our tests (Eberhardt and Thomas 1991). We perennial vegetation. used program SPSS/PC+ v5.0.1 (SPSS, Inc. The floristic/substrate model explained a 1992) to perform all statistical analyses. substantial amount of variation in passage 2 migrant abundance (R adj = 0.61, P < 0.001) RESULTS and contained variables for overstory mesquite and rock. Increasing abundance of pas- Point Counts sage migrants was significantly associated with We detected 91 species during point counts increasing cover of overstory mesquite and on the transects. Forty-six species (50.5%) with decreasing cover of rock. were classified as list A neotropical migrants, HABITAT BREADTH.—Eighty-seven percent 28 species (30.8%) as list B neotropical migrants, of passage migrants (39 of 45; aerial detections and 17 (18.7%) as nonmigrants (Table 1). Fifty >20 m high not included) were classified as (57%) neotropical migrants were further clas- xeroriparian scrub specialists (Table 2). Black- sified as passage migrants (Table 2). chinned Hummingbird, Gray Flycatcher, and Western Kingbird used both xeroriparian Habitat Associations scrub and creosote-bursage vegetation types. SPECIES RICHNESS.—The vegetation type Townsend’s Warbler and Lesser Goldfinch, model to predict passage migrant richness was although occurring primarily in xeroriparian 2 weak (R adj = 0.250, P = 0.005), although scrub, also used creosote-bursage and mixed- increasing species richness was significantly cacti. Bullock’s Oriole was unique in roughly associated with increasing cover of xeroripar- dividing use between xeroriparian scrub and ian scrub vegetation. mixed-cacti. The perennial vegetation model to predict USE OF TREE SPECIES.—Passage migrants 2 species richness was moderately strong (R adj (all species combined) used tree species out of = 0.351, P = 0.001) and included variables for proportion to their availability (χ2 = 38.02, 4 overstory perennial vegetation and ground cover df, P < 0.0001); desert willow trees were used of perennial vegetation. Increasing species (13.1% use) more than expected (6.1% avail- richness was significantly associated with in- ability), and catclaw acacia trees were used 2004] PASSAGE NEOTROPICAL MIGRANTS 63

(22.8%) less then expected (31.6%). Paloverde become increasingly important to both breed- (~23% overall detections for all species), ing and passage neotropical migrants. mesquite (~17%), and catclaw acacia (~15%) Presence and Richness were the dominant or co-dominant plant of Bird Species species used by the migrants analyzed (Table 1). Although of relatively moderate use over- The pooled richness (all 3 years combined) all, flowers of desert willow were used exten- of spring passage migrants was considerably sively by all hummingbirds except the Black- greater than that found in spring by Hensley chinned, which concentrated in saguaro. Most (21 species [1954]) or Vander Wall (29 species relatively abundant (i.e., >25 detections) [1980]) at Organ Pipe Cactus National Monu- species, however, used a wide variety of plant ment (OPCNM), approximately 50 km south species (Table 1). of our study area. This disparity occurred despite the fact that both studies on OPCNM DISCUSSION encompassed the entire spring migration period and included surveys within xeroriparian scrub The variety of neotropical migrants using areas (Vander Wall’s study areas were on bajadas the xeroriparian scrub vegetation type demon- or slopes, but included smaller washes). Dif- strates its importance to these birds on our ferences may be due, in part, to the longer study area. Eighty-seven percent of passage duration (3 years instead of 2) and more inten- neotropical migrants were classified as xerori- sive effort of our study. Cutler (1996) and Cut- parian scrub specialists. Indeed, most (>90%) ler and Morrison (1998), whose study was con- detections of passage migrants were from the ducted during the same years as ours, reported xeroriparian scrub vegetation type. Passage similar richness (approximately 54 of 130 total migrant richness increased as the amount of species) for passage migrants not including xeroriparian scrub cover increased. These find- species associated with free-standing water at ings agree with studies of avian use of mesic water-development sites with extensive mes- riparian areas in the desert Southwest. Johnson quite bosques (woodlands) on the Cabeza Prieta et al. (1977) found that of 77 total breeding National Wildlife Refuge (CPNWR) approxi- neotropical migrants in the Southwest, 58 mately 55 km southwest of our study area. (75.3%) were obligate or preferential riparian However, only 2 species of passage neotropi- species. Stevens et al. (1977) found that 10.6 cal migrants (Wilson’s Warbler and Yellow- times the number of passage neotropical rumped Warbler) could be categorized as “xero- migrants per hectare were found on riparian riparian specialists.” This difference in species plots than on adjacent, nonriparian plots. richness between their study and ours is likely Ohmart and Anderson (1982) reported that of due to the fact that Cutler and Morrison (1998) 308 avian species regularly occurring in the conducted surveys throughout the year. Pas- Sonoran Desert, 56 (18%) were obligate ripar- sage migrants not observed by either Hensley ian, 197 (65%) were facultative riparian, and or Vander Wall, but observed by us, were 55 (18%) were nonriparian species. At mesic Black Swift, Vaux’s Swift, Black-chinned Hum- riparian sites in southeastern Arizona, Skagen mingbird, Calliope Hummingbird, Western et al. (1998) reported greater passage migrant Wood-Pewee, Willow Flycatcher, Hammond’s species richness at isolated oases than at larger, Flycatcher, Dusky Flycatcher, Vermilion Fly- continuous riparian corridors. However, because catcher, Tree Swallow, Cliff Swallow, Swain- of different mechanisms of migration between son’s Thrush, Gray Vireo, Solitary Vireo, Vir- species and the fact that riparian vegetation is ginia’s Warbler, Common Yellowthroat, Spotted naturally disjunct in the Southwest, Skagen et Towhee, Lincoln’s Sparrow, Yellow-headed al. (1998) stressed the importance of both types Blackbird, and American Goldfinch. Cutler of riparian vegetation to migrating passerines. (1996) and Cutler and Morrison (1998) did not Xeroriparian scrub areas are important to observe Black Swift, Calliope Hummingbird, avian species and with the widespread destruc- Gray Vireo, Virginia’s Warbler, or Spotted Tow- tion and desertification of mesic riparian areas hee. All of these species were uncommon in the arid Southwest continuing at rapid rates migrants during our study. The fact that these (Rea 1983, Krueper 1993), preservation and birds were not observed during previous stud- enhancement of xeroriparian scrub areas may ies at OPCNM and CPNWR, south of our 64 WESTERN NORTH AMERICAN NATURALIST [Volume 64

TABLE 1. Percent of detectionsa by plant species and substrate type, Barry M. Goldwater Range, 1994–1996. Migratory statusb is indicated (Gauthreaux 1992). Passage migrants are indicated with an asterisk. Species LOC1(%)c LOC2(%)d LOC3(%)e Detections

LIST A NEOTROPICAL MIGRANTS White-winged Dove SA(49.5) PV(20.9) IW(12.1) 91 Zenaida asiatica Black-chinned Hummingbird* SA(66.7) CP(33.3) — 3 Archilochus alexandri Costa’s Hummingbird DW(38.9) MQ(14.4) PV(12.2) 90 Calypte costae Calliope Hummingbird* DW(100.0) — — 2 Stellula calliope Allen’s Hummingbird* DW(100.0) — — 2 Selasphorus sasin Olive-sided Flycatcher* DW(100.0) — — 1 Contopus cooperi Western Wood-Pewee* CC,IW,MQ(33.3) — — 3 C. sordidulus Willow Flycatcher* PV(66.7) MQ(33.3) — 3 Empidonax traillii Hammond’s Flycatcher* CC(50.0) CR,MQ(25.0) — 4 E. hammondii Dusky Flycatcher* CC,DW(50.0) — — 2 E. oberholseri Gray Flycatcher* PV(40.0) CC,DW,MQ(20.0) — 5 E. wrightii Western Flycatcher Complexf* MQ(40.5) CC(23.8) PV(21.4) 42 Empidonax spp. Vermilion Flycatcher MQ,PV(50.0) — — 2 Pyrocephalus rubinus Ash-throated Flycatcher PV(39.6) IW(19.8) MQ(15.3) 111 Myiarchus cinerascens Brown-crested Flycatcher CR,DW,IW,MQ(17.8) — — 28 M. tyrannulus Western Kingbird* DW,MQ,PV(33.3) — — 3 Tyrannus verticalis House Wren* CC(60.0) DW,MQ(20.0) — 5 Troglodytes aedon Phainopepla MQ(35.3) PV(27.1) CC(18.2) 170 Phainopepla nitens Bell’s Vireo DW(46.1) MQ(23.1) PV(15.4) 13 Vireo bellii Gray Vireo* PV(100.0) — — 1 V. vicinior Solitary Vireo Complexg* CC,IW,MQ,PV(25.0) — — 4 V. solitarius Warbling Vireo* PV(37.5) CC(25.0) CR,DW,MQ(12.5) 8 V. gilvus Orange-crowned Warbler* MQ(41.2) CC(26.5) PV(20.6) 34 Vermivora celata Nashville Warbler* MQ(54.5) PV(36.4) IW(9.1) 11 V. ruficapilla Lucy’s Warbler MQ(31.3) PV(28.1) IW(21.1) 128 V. luciae Yellow Warbler* PV(30.8) CC,DW,IW,MQ(15.4) — 13 Dendroica petechia Black-throated Gray Warbler* PV(35.7) CC,IW(21.4) — 14 D. nigrescens Townsend’s Warbler* MQ(36.7) CC(23.3) PV(20.0) 30 D. townsendi Hermit Warbler* CC,IW,PV(33.3) — — 3 D. occidentalis 2004] PASSAGE NEOTROPICAL MIGRANTS 65

TABLE 1. Continued. Species LOC1(%)c LOC2(%)d LOC3(%)e Detections MacGillivray’s Warbler* CC,GR,MQ,PV(33.3) — — 4 Oporornis tolmiei Common Yellowthroat* CC,PV(50.0) — — 2 Geothlypis trichas Wilson’s Warbler* CC(24.2) MQ(23.2) DW(19.0) 95 Wilsonia pusilla Western Tanager* IW(44.4) DW(33.3) CC,MQ(11.1) 9 Piranga ludoviciana Black-headed Grosbeak* PV(28.6) DW,IW(21.4) — 14 Pheucticus melanocephalus Lazuli Bunting* CC,DH,MQ,PV(25.0) — — 4 Passerina amoena Varied Bunting CC,DW(33.3) MQ,WB(16.7) — 6 P. versicolor Green-tailed Towhee* GR(44.4) CC,MQ(22.2) — 9 Pipilo chlorurus Chipping Sparrow GR(38.5) CC,IW,MQ(15.4) — 13 Spizella passerina Brewer’s Sparrow GR(52.6) CR(15.8) PV(10.5) 19 S. breweri Lark Sparrow* GR,PV(50.0) — — 2 Chondestes grammacus Lark Bunting CC(100.0) — — 1 Calamospiza melanocorys Lincoln’s Sparrow* CR(100.0) — — 1 Melospiza lincolnii Yellow-headed Blackbird* IW(100.0) — — 1 Xanthocephalus xanthocephalus Hooded Oriole PV(59.6) SA(12.3) IW(10.5) 57 Icterus cucullatus Bullock’s Oriole* IW(40.0) OC,PV,SN(20.0) — 5 I. bullockii Scott’s Oriole SA(54.5) PV(25.0) IW(9.1) 44 I. parisorum

LIST B NEOTROPICAL MIGRANTS Turkey Vulture IW(34.2) MQ,PV(21.1) — 38 Cathartes aura Sharp-shinned Hawk* CR(100.0) — — 1 Accipiter striatus Cooper’s Hawk* PV(50.0) MQ,SN(25.0) — 4 A. cooperii Red-tailed Hawk SA(50.0) RO(28.6) GR(14.3) 14 Buteo jamaicensis American Kestrel SA(87.5) PV(8.3) RO(4.2) 24 Falco sparverius Prairie Falcon SA(100.0) — — 1 F. mexicanus Mourning Dove PV(31.7) GR,MQ(19.5) — 41 Zenaida macroura Anna’s Hummingbird* CC,DW(50.0) — — 2 Calypte anna Say’s Phoebe CR(50.0) MQ,PV(25.0) — 4 Sayornis saya Rock Wren GR(53.6) RO(35.7) CR,IW,PV(3.6) 28 Salpinctes obsoletus Bewick’s Wren* MQ(50.0) CC,GR(25.0) — 4 Thryomanes bewickii Winter Wren SH(100.0) — — 1 Troglodytes troglodytes 66 WESTERN NORTH AMERICAN NATURALIST [Volume 64

TABLE 1. Continued. Species LOC1(%)c LOC2(%)d LOC3(%)e Detections Ruby-crowned Kinglet* PV(35.3) CC(23.5) MQ(17.6) 17 Regulus calendula Hermit Thrush* PV(100.0) — — 2 Catharus guttatus American Robin MQ,CC(40.0) CR(20.0) — 5 Turdus migratorius Northern Mockingbird MQ(30.7) PV(20.4) SA(19.3) 88 Mimus polyglottos Sage Thrasher CC,CR,DW,MQ(25.0) — — 4 Oreoscoptes montanus Bendire’s Thrasher IW,PV,SA(33.3) — — 3 Toxostoma bendirei Loggerhead Shrike SA(38.9) MQ,PV(16.7) — 36 Lanius ludovicianus Yellow-rumped Warbler* PV(46.7) MQ(33.3) CC,GR,IW(6.7) 15 Dendroica coronata Spotted Towhee* CC,GR(50.0) — — 2 Pipilo maculatus Vesper Sparrow CH,PV(50.0) — — 2 Pooecetes gramineus Black-throated Sparrow GR(41.7) CR(29.8) PV(9.5) 84 Amphispiza bilineata White-crowned Sparrow GR(50.0) MQ(15.9) IW(11.4) 44 Zonotrichia leucophrys Dark-eyed Junco* DW,GR(50.0) — — 2 Junco hyemalis Brown-headed Cowbird PV(39.1) CC(21.7) SA(17.4) 23 Molothrus ater Lesser Goldfinch* MQ(41.7) CC,DW,IW(1.7) — 12 Carduelis psaltria American Goldfinch* CR(100.0) — — 1 C. tristis

NONMIGRANTS Harris’ Hawk PV,SA(44.4) MQ(11.1) — 9 Parabuteo unicinctus Gambel’s Quail GR(80.2) MQ,SA(5.0) — 121 Callipepla gambelii Greater Roadrunner GR(88.9) RO(11.1) — 9 Geococcyx californianus Gila Woodpecker SA(51.2) PV(17.8) MQ(16.3) 129 Melanerpes uropygialis Gilded Flicker SA(76.9) IW(6.7) PV,SN(3.8) 104 Colaptes chrysoides Ladder-backed Woodpecker MQ(35.5) SA(22.6) PV(16.1) 31 Picoides scalaris Common Raven RO(50.0) IW,SA(25.0) — 4 Corvus corax Verdin PV(41.9) MQ(18.1) IW(13.2) 265 Auriparus flaviceps Cactus Wren SA(34.6) PV(14.1) CC(10.3) 78 Campylorhynchus brunneicapillus Canyon Wren RO(66.7) GR(33.3) — 12 Catherpes mexicanus Black-tailed Gnatcatcher PV(30.8) MQ(17.9) IW(17.3) 156 Polioptila melanura Curve-billed Thrasher SA(25.0) MQ(19.7) PV(11.8) 76 Toxostoma curvirostre Crissal Thrasher GR(57.1) MQ(28.6) DW,PV(14.3) 7 T. crissale 2004] PASSAGE NEOTROPICAL MIGRANTS 67

TABLE 1. Continued. Species LOC1(%)c LOC2(%)d LOC3(%)e Detections Northern Cardinal MQ(26.8) PV(22.0) CC,DW(17.1) 41 Cardinalis cardinalis Pyrrhuloxia CC(33.3) PV(18.2) MQ(15.2) 33 C. sinuatus Canyon Towhee GR(33.9) CC(21.0) PV(14.5) 62 Pipilo fuscus House Finch SA(46.2) PV(15.4) IW(12.8) 39 Carpodacus mexicanus aGroups of individuals of the same species counted as n = 1 detection. Detections of high-flying individuals (>20 m) and nocturnal species are not included in Table 1. bList A contains species that breed in North America and spend their nonbreeding season primarily south of the U.S. This list contains species generally recognized as neotropical migrants. List B is composed of species that breed and winter extensively in North America, although some populations winter south of the U.S. (adapted from Gauthreaux 1992). cDominant plant species or substrate in which species was most frequently observed: CC = catclaw acacia, CP = chuparosa, CR = creosote, DW = desert willow, GR = ground, IW = ironwood, MQ = mesquite, OC = ocotillo, PV = paloverde spp., RO = rock, SA = saguaro, SH = unknown shrub spp., SN = snag, WB = wolfberry (Lycium spp.). dCo-dominant plant species or substrate in which species was frequently observed. eCo-dominant plant species or substrate in which species was frequently observed. fIncludes Pacific-slope (Empidonax difficilis) and Cordilleran (E. occidentalis) Flycatchers; all individuals of known identity were Pacific-slope Flycatchers. gIncludes Plumbeous Vireo (Vireo plumbeus) and Cassin’s Vireo (Vireo cassinii).

study area, suggests that they are uncommon greater cover of tall mesquite than was avail- migrants through the region as well. able overall included MacGillivray’s, Nashville, Orange-crowned, and Wilson’s Warblers, and Indices of Abundance Western Flycatcher. For the pooled data set, the most abundant Mesquite bosques in the southwestern spring passage migrants, in descending order, deserts have been shown to produce an abun- were Wilson’s Warbler, Western Flycatcher, dance of arthropods and to receive heavy use Orange-crowned Warbler, and Townsend’s by insectivorous passage and breeding migrants Warbler. Phillips et al. (1964) reported each of (Ohmart and Anderson 1982, Rosenberg et al. these species to be common spring migrants 1991, Cutler and Morrison 1998). On a per-tree in the Sonoran Desert of southwestern Ari- basis, mesquite provides one of the richest zona. All 4 species were commonly observed pollen and nectar sources in the Sonoran in spring at OPCNM by Hensley (1954) and Desert (Ohmart and Anderson 1982). Simpson Vander Wall (1980), and at CPNWR by Cutler et al. (1977) reported that mesquite produces (1996). more pollen per floral unit than any other insect-pollinated desert tree in North Amer- Habitat Associations ica. A large number of insects use this rich High passage migrant richness was strongly food resource while it is available (Simpson et associated with tall (mid-canopy and over- al. 1977). On our study area the flowering of story) height classes of catclaw acacia, mes- mesquite coincided with spring migration quite, and paloverde. Given this result, it is during all 3 years of our study (personal obser- not surprising that the presence of several pas- vation). sage migrant species was also associated with Passage migrants as a group (all species tall height classes of mesquite (MacGillivray’s combined) used catclaw acacia significantly Warbler, Orange-crowned Warbler, Ruby- less than expected. However, when we exam- crowned Kinglet, Western Flycatcher, Wilson’s ined the percent of detections by location, we Warbler, and Yellow-rumped Warbler), and tall saw that 23% of all detections of passage height classes of paloverde (Black-throated migrants were from catclaw acacia, and that Gray Warbler, Lazuli Bunting, Sharp-shinned approximately 22% of all detections of breed- Hawk, and Western Flycatcher). In addition, ing neotropical migrants were from catclaw. In passage migrants often selected areas with 1995 and 1996 the flowering cycle of catclaw dense cover of tall mesquite. Passage migrants coincided with the period of heaviest migra- present at count stations with significantly tion (mid-April through mid-May; personal 68 WESTERN NORTH AMERICAN NATURALIST [Volume 64

TABLE 2. Habitat distributions of passage neotropical migrant landbirds on the Barry M. Goldwater Range, Arizona, 1994–1996. Numbers represent the percentage of detectionsa within each vegetation type weighted by the estimated availability of each vegetation type in the study area. Species were classified as single vegetation type specialists if their habitat breadth was ≤1.3, two-vegetation type specialists if between 1.3 and 2.3, and broad generalists if ≥2.3. Migratory statusb is indicated (Gauthreaux 1992).

c ______Vegetation Type Species (N) XR CB MC Breadthd

A. XERORIPARIAN SPECIALISTS List A Migrants Calliope Hummingbird 2 100.0 — — 1.00 Allen’s Hummingbird 2 100.0 — — 1.00 Olive-sided Flycatcher 1 100.0 — — 1.00 Western Wood-Pewee 5 100.0 — — 1.00 Willow Flycatcher 3 100.0 — — 1.00 Hammond’s Flycatcher 4 100.0 — — 1.00 Dusky Flycatcher 3 100.0 — — 1.00 Western Flycatchere 61 95.9 4.1 — 1.09 House Wren 5 100.0 — — 1.00 Blue-gray Gnatcatcher 1 100.0 — — 1.00 Gray Vireo 2 100.0 — — 1.00 Solitary Vireof 4100.0 — — 1.00 Warbling Vireo 8 100.0 — — 1.00 Orange-crowned Warbler 40 100.0 — — 1.00 Nashville Warbler 13 100.0 — — 1.00 Yellow Warbler 17 86.2 — 13.8 1.30 Black-throated Gray Warbler 14 100.0 — — 1.00 Hermit Warbler 3 100.0 — — 1.00 MacGillivray’s Warbler 10 100.0 — — 1.00 Common Yellowthroat 4 100.0 — — 1.00 Wilson’s Warbler 118 93.7 — 6.3 1.13 List A Migrants Western Tanager 11 100.0 — — 1.00 Black-headed Grosbeak 17 100.0 — — 1.00 Lazuli Bunting 4 100.0 — — 1.00 Green-tailed Towhee 16 100.0 — — 1.00 Lark Sparrow 2 100.0 — — 1.00 Lincoln’s Sparrow 1 100.0 — — 1.00 Yellow-headed Blackbird 2 100.0 — — 1.00

observation). The large, aromatic flowers of often miss mesquite and paloverde because of catclaw attract an abundance of pollinating their open growth forms. Although percent insects (Ohmart and Anderson 1982). Indeed, cover of a plant species is often used to approx- over 90% of the individual birds detected in imate its availability in habitat selection stud- catclaw were observed foraging (unpublished ies, for species like catclaw acacia, it may be a data), suggesting that catclaw is an important poor approximation. The interior of catclaw feeding site for many species of neotropical may be so thick that it is unavailable to some migrants. This apparent contradiction between species of birds, or more likely it may obscure observed/actual use and reported results may the observer’s view, making a bird less detect- be due to the growth form of catclaw, which able. Thus, less of the tree is actually available tends to grow very thickly throughout its than that estimated by percent cover. height, often forming an impenetrable mass of Costa’s Hummingbird, the only breeding branches. This growth form contrasts with the neotropical migrant hummingbird on our study more open growth form of mesquite and palo- area, and 3 species of passage migrant hum- verde. The point-intercept method we used to mingbirds (Calliope, Allen’s, and Rufous), were determine percent cover tends to seldom miss detected from desert willow trees more than catclaw acacia due to its thickness, but may any other plant species (Table 1). Male Costa’s 2004] PASSAGE NEOTROPICAL MIGRANTS 69

TABLE 2. Continued.

c ______Vegetation Type Species (N) XR CB MC Breadthd List B Migrants Northern Harrier 1 100.0 — — 1.00 Sharp-shinned Hawk 4 100.0 — — 1.00 Cooper’s Hawk 7 100.0 — — 1.00 Anna’s Hummingbird 2 100.0 — — 1.00 Bewick’s Wren 4 100.0 — — 1.00 Ruby-crowned Kinglet 23 89.6 — 10.4 1.23 Hermit Thrush 2 100.0 — — 1.00 Yellow-rumped Warbler 19 100.0 — — 1.00 Spotted Towhee 3 100.0 — — 1.00 Dark-eyed Junco 2 100.0 — — 1.00 American Goldfinch 2 100.0 — — 1.00

B. TWO-VEGETATION TYPE (XERORIPARIAN SCRUB AND CREOSOTE-BURSAGE) GENERALISTS List A Migrants Black-chinned Hummingbird 6 57.1 42.9 — 1.96 Gray Flycatcher 6 57.1 42.9 — 1.96 Western Kingbird 7 61.6 38.4 — 1.90 Townsend’s Warbler 36 84.4 9.3 6.3 1.38 List B Migrants Lesser Goldfinch 23 77.0 13.7 9.3 1.61

C. TWO-VEGETATION TYPE (MIXED-CACTI AND XERORIPARIAN SCRUB) GENERALISTS List A Migrants Bullock’s Oriole 9 43.7 — 56.3 1.97 aGroups of individuals of the same species counted as n = 1 detection. Detections of high-flying individuals (>20 m) and nocturnal species not included in Table 2. bList A contains species that breed in North America and spend their nonbreeding season primarily south of the U.S. This list contains species generally recognized as neotropical migrants. List B is composed of species that breed and winter extensively in North America, although some populations winter south of the U.S. (adapted from Gauthreaux 1992). Vaux’s Swift, and Tree, Violet-green, Northern Rough-winged, and Cliff Swallow were not included because of their aerial behavior (could not assign to a vegetation type). cVegetation type: XR = xeroriparian scrub, CB = creosote-bursage, and MC = mixed-cacti; no use of rocks or cliffs was observed (and is not included in the table). d ∑ 2 Habitat breadth = 1/ pi , where pi = proportion of weighted detections in vegetation type i. eIncludes Pacific-Slope and Cordilleran Flycatchers; all individuals of known identity were Pacific-slope Flycatchers. fIncludes Plumbeous and Cassin’s Vireos.

Hummingbirds most frequently sang from this wood trees were Black-throated Gray Warbler, tree species (unpublished data). Desert willow Western Tanager, Black-headed Grosbeak, and trees occurred primarily along 1 transect and Bullock’s Oriole. Lucy’s Warbler was a com- were in bloom during the time spring passage mon breeding neotropical migrant that used migrant hummingbirds were observed (per- cavities in ironwood for nesting (personal sonal observation). Desert willow flowers are a observation). major source of nectar for hummingbird species Conservation and in the Sonoran Desert (Calder 1993, 1994), and Management Implications the abundance of desert willow trees along 1 transect may help explain why 14 of 19 obser- We have established a baseline for monitor- vations of passage migrant hummingbirds ing natural and human-influenced changes in occurred in this area. the abundance of neotropical migrants and Ironwood was important as a foraging and their habitat on our study area. We recommend nesting substrate for some neotropical migrants that land managers, including the U.S. Air Force (personal observation). Passage neotropical and adjacent land managers such as the Bur- migrants frequently observed foraging in iron- eau of Land Management, use our protocols to 70 WESTERN NORTH AMERICAN NATURALIST [Volume 64 monitor neotropical migrants, sample vegeta- Downard, R. Barry, and N. Brown for their tion along transects, and establish additional outstanding assistance in the field. We thank J. transects when necessary. Bird abundance and Kelly and an anonymous referee for critiquing habitat association data will aid land managers an earlier draft. in identifying areas critical to neotropical migrants. LITERATURE CITED The habitat characteristics illustrated as important to neotropical migrants can be pre- AUSTIN, G.T. 1970. Breeding birds of desert riparian habi- served and managed by protecting xeroripar- tat in southern Nevada. Condor 72:431–436. BELSLEY, D.A., E. KUG, AND R.E. WELSCH. 1980. Regres- ian areas, particularly areas that support mature sion diagnostics. Wiley and Sons, New York. (>2.5 m) paloverde, mesquite, desert willow, BLEM, C.R. 1980. The energetics of migration. Pages 175– and catclaw acacia trees, or have the potential 224 in S.A. Gauthreaux, Jr., editor, Animal migration, to support these species. Our data suggest that orientation, and navigation. Academic Press, New York. BRITTINGHAM, M.C., AND S.A. TEMPLE. 1983. Have cow- xeroriparian scrub within the creosote-bur- birds caused forest songbirds to decline? Bioscience sage vegetation type may be particularly im- 33:31–35. portant to passage neotropical migrants, where- CALDER, W. 1993. Rufous Hummingbird (Selasphorus as the width of riparian areas may be an impor- rufus). Pages 1–20 in A. Poole and F. Gill, editors, tant factor for breeding neotropical migrants. The birds of North America 53. American Ornitholo- gists’ Union, Washington, DC. When considering the use of washes by breed- ______. 1994. Allen’s Hummingbird (Selasphorus rufus). ing and passage neotropical migrants, man- Pages 1–22 in A. Poole and F. Gill, editors. The birds agers may choose to limit or close washes to of North America 135. American Ornithologists’ recreational use and vehicular traffic either Union, Washington, DC. seasonally or permanently (Luckenbach 1977). CHERRY, S. 1996. A comparison of confidence interval methods for habitat use-availability studies. Journal Potential impacts of other uses of the washes, of Wildlife Management 60:653–658. such as livestock grazing and wood cutting, CUTLER, T.L. 1996. Wildlife use of two artificial water should also be examined. developments on the Cabeza Prieta National Wildlife While certain attributes of woody perennial Refuge, southwestern Arizona. Master’s thesis, Uni- vegetation exhibit little interannual variation versity of Arizona, Tucson. CUTLER, T.L., AND M.L. MORRISON. 1998. Habitat use by (e.g., density of plants), other attributes of small vertebrates at two water developments in woody vegetation (e.g., fruit production, flow- southwestern Arizona. Southwestern Naturalist 43: ering, and leaf cover) can vary considerably 155–162. over time. These ephemeral aspects of woody EBERHARDT, L.L., AND J.M. THOMAS. 1991. Designing environmental field studies. Ecological Monographs vegetation could be monitored over time, 61:53–73. especially as they relate to food availability. GAUTHREAUX, S.A., JR.1992. Preliminary lists of migrants Herbaceous vegetation is extremely ephemeral for Partners in Flight neotropical migratory bird in the Sonoran Desert (Wiens 1991) and could conservation program. Partners in Flight Newsletter be sampled each year that bird surveys are 2:30. GORI, D. 1992. Know your element: cottonwood-willow conducted to account for this spatial and tem- riparian forests. The Nature Conservancy Arizona poral variation. Due to the importance of Chapter Newsletter 14:1–12. desert mistletoe (Phorodendron californica) to HENSLEY, M.M. 1954. Ecological relations of the breeding many neotropical migrants (personal observa- bird population of the desert biome in Arizona. Eco- tion), at the very least, we recommend moni- logical Monographs 24:185–207. JOHNSON, R.R., L.T. HAIGHT, AND J.M. SIMPSON. 1977. toring this plant species’ numbers, health, and Endangered species versus endangered habitats: a fruit production over time. Future vegetation concept. USDA Forest Service, General Technical sampling need not be done with the rigor of the Report RM-43:52–58, Fort Collins, CO. current protocol, but should be standardized. KARR, J.R. 1968. Habitat and avian diversity on strip-mined land in eastern Illinois. Condor 70:348–357. KRUEPER, D.J. 1993. Effects of land use practices on west- ACKNOWLEDGMENTS ern riparian ecosystems. Pages 321–330 in D.M. Finch and P.W. Stangel, editors, Status and management of We thank the Range Management Office at neotropical migratory birds. USDA Forest Service, Luke Air Force Base for funding and the General Technical Report RM-229, Fort Collins, CO. School of Renewable Natural Resources at the LEVINS, R. 1968. Evolution in changing environments. Princeton University Press, Princeton, NJ. University of Arizona for financial and logisti- LUCKENBACH, R.A. 1977. An analysis of off-road vehicle cal support. We are indebted to T. Abeloe, G. use on desert avifaunas. Transactions of the North 2004] PASSAGE NEOTROPICAL MIGRANTS 71

American Wildlife and Natural Resources Confer- Simpson, editor, US/IBP Synthesis Series 4. Strouds- ence 43:157–162. burg, PA. MOORE, F.R., S.A. GAUTHREAUX, P. KERLINGER, AND T.R. SKAGEN, S.K., C.P. MELCHER, W.H. HOWE, AND F.L. KNOPF. SIMMONS. 1993. Stopover habitat: management impli- 1998. Comparative use of riparian corridors by mi- cations and guidelines. Pages 58–69 in D.M. Finch grating birds in southeast Arizona. Conservation Biol- and P.W. Stangel, editors, Status and management of ogy 12:896–909. neotropical migratory birds. USDA Forest Service, SPSS, Inc. 1992. SPSS/PC+ advanced statistics, version General Technical Report RM-229, Fort Collins, CO. 5. SPSS Inc., Chicago. ______. 1995. Habitat requirements during migration: STEVENS, L.E., B.T. BROWN, J.M. SIMPSON, AND R.R. important link in conservation. Pages 121–144 in JOHNSON. 1977. The importance of riparian habitat T.E. Martin and D.M. Finch, editors, Ecology and to migrating birds. Pages 156–164 in R.R. Johnson management of neotropical migratory birds. Oxford and D.A. Jones, editors, Importance, preservation and University Press, New York. management of riparian habitat. USDA Forest Ser- MORRISON, M.L., R.W. MANNAN, AND B.G. MARCOT. 1998. vice, General Technical Report RM-43, Fort Collins, Wildlife-habitat relationships: concepts and applica- CO. tions. 2nd edition. University of Wisconsin Press, TURNER, R.M., AND D.E. BROWN. 1994. Sonoran desert- Madison. scrub. Pages 181–221 in D.E. Brown, editor, Biotic OHMART, R.D., AND B.W. ANDERSON. 1982. North American communities: southwestern United States and desert riparian ecosystems. Pages 433–474 in G.L. northwestern Mexico. University of Utah Press, Salt Bender, editor, Reference handbook on the deserts Lake City. of North America. Greenwood Press, Westport, CT. VANDER WALL, S.B. 1980. The structure of Sonoran Desert NORUSIS, M.J. 1990. SPSS/PC+, advanced statistics. SPSS bird communities: effects of vegetation structure and Inc., Chicago. precipitation. Doctoral dissertation, Utah State Uni- PARKER, K.C. 1986. Partitioning of foraging space and nest versity, Logan. sites in a desert shrubland bird community. Ameri- VANDER WALL, S.B., AND J.A. MACMAHON. 1984. Avian can Midland Naturalist 115:255–267. distribution patterns along a Sonoran Desert bajada. PHILLIPS, A.R. 1975. The migrations of Allen’s and other Journal of Arid Environments 7:59–74. hummingbirds. Condor 77:196–205. VERNER, J. 1985. Assessment of counting techniques. Pages PHILLIPS, A.R., J. MARSHALL, AND G. MONSON. 1964. The 247–301 in R.F. Johnston, editor, Current ornithology, birds of Arizona. University of Arizona Press, Tucson. volume 2. Plenum Press, New York. RAITT, R.J., AND R.L. MAZE. 1968. Densities and species WAUER, R.H. 1977. Significance of Rio Grande riparian composition of breeding birds of a creosote commu- systems upon the avifauna. Pages 165–174 in R.R. nity in southern New Mexico. Condor 70:193–205. Johnson and D.A. Jones, editors, Importance, preser- RALPH, C.J., G.R. GUEPEL, P. PYLE, T.E. MARTIN, AND D.F. vation and management of riparian habitat. USDA DESANTE. 1993. Handbook of field methods for Forest Service, General Technical Report RM-43, monitoring landbirds. USDA Forest Service, Gen- Fort Collins, CO. eral Technical Report PSW-GTR-144, Albany, CA. WIENS, J.A. 1991. The ecology of desert birds. Pages 278– REA, A.M. 1983. Once a river. University of Arizona, Tucson. 310 in G.A. Polis, editor, The ecology of desert com- ROSENBERG, K.V., R.D. OHMART, W.C. HUNTER, AND B.W. munities. University of Arizona Press, Tucson. ANDERSON. 1991. Birds of the Lower Colorado River ZAR, J.H. 1996. Biostatistical analysis. 3rd edition. Pren- Valley. University of Arizona Press, Tucson. tice-Hall, Inc., Upper Saddle River, NJ. SELLERS, W.D., R.H. HILL, AND M. SANDERSON-RAE. 1985. Arizona climate: the first hundred years (1885–1985). Received 1 July 2002 University of Arizona Press, Tucson. Accepted 28 March 2003 SIMPSON, B.B., J.L. NEFF, AND A.R. MOLDENKE. 1977. Prosopis flowers as a resource. Pages 84–107 in B.B. Western North American Naturalist 64(1), ©2004, pp. 72–77

SPECIES DIVERSITY AND HABITAT OF GRASSLAND PASSERINES DURING GRAZING OF A PRESCRIBE-BURNED, MIXED-GRASS PRAIRIE

Robert F. Danley1, Robert K. Murphy2, and Elizabeth M. Madden3

ABSTRACT.—No published data exist on responses of grassland passerines and their habitat to combined grazing and burning treatments in northern mixed-grass prairie. At Lostwood National Wildlife Refuge (LNWR) in northwestern North Dakota, we monitored breeding bird occurrence, abundance, and habitat during successive annual grazing treatments (1998–2000) on 5 prescribe-burned, mixed-grass prairie management units (range = 50–534 ha, each burned 3–6 times in the previous 10–20 years). All breeding passerine species characteristic of upland, northern mixed-grass prairie were common (>10% occurrence) during at least 1 of 3 years on burned and grazed units, except Chestnut-collared Longspur (Calcarius ornatus), which was uncommon. Vegetation was generally shorter and sparser than that found on 4 nearby units treated by fire only (1999; density, visual obstruction, and height, all P < 0.01). Regardless, occurrences of individual bird species resembled those previously documented on prairie units at LNWR with similar fire histories but no grazing; however, Brown-headed Cowbird (Molothrus ater) occurred 2.4 times more frequently on burned and grazed units studied. Our data suggest that species diversity of breeding grassland passerines changes little during initial years of rotation grazing at moderate stocking rates in fire-managed, northern mixed-grass prairie at LNWR.

Key words: prescribed fire, rotation grazing, habitat management, grassland passerine, mixed-grass prairie, northern Great Plains, species diversity.

The evolution of Great Plains grasslands was Our primary objective was to measure shaped by interacting fire and grazing distur- species diversity and habitat of breeding bances (Higgins 1986) along with climatic grassland passerines at LNWR on prairie units variability (Bragg 1994). To conserve these managed by a combination of prescribed fire grasslands and associated wildlife communi- and livestock grazing. Specifically, we sought ties, land managers often use prescribed fire to document breeding bird occurrence and or livestock grazing to mimic historic distur- abundance, species diversity, and habitat (i.e., bances. Diversity and abundance of grassland vegetation) conditions during rotation grazing birds can increase after fire is reintroduced to of prairie 1–4 years after the last of several northern mixed-grass prairie (Johnson 1997, prescribed burns. Our secondary objective Madden et al. 1999). Depending on the graz- was to compare breeding bird diversity with ing system employed, livestock grazing can that of prairie at LNWR with similar fire histories but no grazing (Madden et al. 1999). reduce abundances of some breeding bird species, such as Baird’s Sparrow (scientific STUDY AREA AND METHODS names of bird species are listed in Table 2; Kantrud 1981). Surprisingly, no published data LNWR covers 109 km2 of rolling to hilly exist on breeding passerine responses to com- moraine in Burke and Mountrail Counties, bined fire and grazing treatment regimes in northwestern North Dakota (48°37′N, northern mixed-grass prairie. At Lostwood 102°27′W). The refuge is 55% native prairie, National Wildlife Refuge (LNWR) in north- 21% previously cropped fields that were western North Dakota, cattle grazing is used revegetated with native and introduced plants, mainly to reduce exotic plants, especially 20% wetlands, 2% trees, and 2% tall shrubs smooth brome (Bromus inermis), on northern (Murphy 1993). Native prairie is a needle- mixed-grass prairie renovated by fire (U.S. grass-wheatgrass (Stipa-Agropyron) associa- Fish and Wildlife Service 1998). tion heavily invaded by western snowberry

1U.S. Fish and Wildlife Service, Lostwood National Wildlife Refuge, 8315 Highway 8, Kenmare, ND 58746. 2U.S. Fish and Wildlife Service, Des Lacs National Wildlife Refuge Complex, Kenmare, ND 58746. 3U.S. Fish and Wildlife Service, Medicine Lake National Wildlife Refuge, Medicine Lake, MT 59247.

72 2004] PASSERINES ON BURNED AND GRAZED PRAIRIE 73

(Symphoricarpos occidentalis) and 2 exotic randomly established 75-m-radius survey plots grasses, smooth brome and Kentucky blue- with centers at least 250 m apart. Roads, trees, grass (Poa pratensis). Soils are mainly gravelly unit boundaries, and most wetlands were clay-loams that occur as Silty and Thin Upland avoided (Madden et al. 1999). Survey plots range sites (Soil Conservation Service 1984). remained fixed during the course of our study Climate is semiarid with a mean annual pre- (3–15 plots per unit; units D through H in cipitation of 42 cm (Madden et al. 1999). Table 1). In June each year we conducted The floral integrity of LNWR’s mixed-grass three 10-minute surveys of singing males prairie deteriorated during decades of long- within each plot (Hutto et al. 1986), during the term rest or light grazing but began to recover period from one-half hour before sunrise until after 3–5 prescribed burns were applied over 0900 CST. Surveys were not conducted during 10–20 years (U.S. Fish and Wildlife Service rain, fog, or winds >16 km ⋅ h–1. Abundance of 1998, Madden et al. 1999). The burning treat- each species was the mean number of singing ment phase was followed by a mixed grazing males per plot (i.e., means derived from 3 sur- and burning phase. Five management units at vey visits to each plot were averaged). Brown- LNWR entered this mixed treatment phase in headed Cowbirds were tallied regardless of 1998 (4 units) and 1999 (1 unit), 1–2 years after sex, and their total number on a survey of a last having been burned (Table 1). On each plot was divided by 2 for abundance (Madden unit we employed a rotation grazing system et al. 1999). with 1 cattle herd. Each unit was divided into Each year from late June through mid-July, 3 equal-sized cells, and each cell was grazed we recorded vegetation structure and general by cattle for 14 days. Every year 2 of 3 cells in composition for each plot along 2 perpendicu- each unit received a 2nd grazing treatment for lar, 150-m transects that bisected the plot cen- another 14 days, with 28 days of rest between ter on a randomly selected bearing (Grant et grazing treatments. The grazing season was late al. 2004). We identified the dominant plant May through mid-August (about 2.5 months). group (woody or herbaceous [i.e., grasses, Stocking rates were 0.6–1.2 ha ⋅ AUM–1 (Table sedges, forbs], native or exotic vegetation) at 1), which we considered moderate compared 0.5-m intervals for a total of 300 readings per with roughly 0.8 ha ⋅ AUM–1 recommended transect (n = 600 readings per plot). We ran- based on area soil types (Soil Conservation domly omitted some transect data to avoid over- Service 1984). representing plot centers (Grant et al. 2004). During 1998–2000 we sampled vegetation At 15-, 35-, 55-, and 75-m stops from the cen- on all of these prescribe-burned and grazed ter along each transect (n = 16 measures per management units and birds on all or a subset plot), we used a 7-mm-diameter rod to mea- of them (Table 1). In 1999 we also sampled sure vegetation density (i.e., total contacts on vegetation on 4 other management units at the rod; Rotenberry and Wiens 1980), maxi- LNWR (36, 45, 61, 755 ha) that had burn his- mum height (highest dm contacted), and litter tories similar to other burn-graze units stud- depth (to nearest cm). Visual obstruction read- ied, but they lacked grazing treatments. We ings (VORs; Robel et al. 1970) were recorded sampled birds on 4 of the burn-graze units in at each stop in 1999 and 2000 to estimate veg- 1998. Two of these were relatively small for etation height-density. VORs were recorded bird sampling (only 3–4 survey plots each; from 4 m away, in 4 cardinal directions, at a Table 1). Therefore, during 1999 and 2000 we height of 1 m. Structural measures were aver- substituted a larger unit that entered the aged for each plot. We used 2-sample t tests to mixed treatment phase in 1999, while continu- assess differences in vegetation structure ing to sample birds on the other 2 large burn- between treatment (burn-graze) units and graze units. We did not sample birds on con- control (burn-only) units in 1999. trol (burn-only) units because 3 of the 4 units We compared frequencies of occurrence of either had more wetlands or more trees than common breeding bird species during the the 5 burn-graze units or were only about 100 post-fire grazing treatments of prairie with those m wide. documented in 1994 at LNWR on prairie with We followed methods outlined in Madden similar fire histories but no grazing (based on (1996) and Madden et al. (1999) for measuring a subset of data from Madden et al. 1999). To bird occurrence and vegetation attributes. We minimize differential observer bias in this 74 WESTERN NORTH AMERICAN NATURALIST [Volume 64

TABLE 1. Prescribed burning and grazing treatment history on management units where breeding passerine species were surveyed, Lostwood National Wildlife Refuge, northwestern North Dakota. Precipitation Years since Fire Stocking rate Year (cm)a Unitsb No. burnsc last firec index (x–)d (AUM ⋅ ha–1)c,e 1994 71.3 A, B, C 4, 4, 4 8, 2, 2 1.5 (no grazing) 1998 51.7 D, E, F, G 5, 6, 3, 4 1, 2, 2, 2 2.9 1.2, 0.8, 1.2, 1.2 1999 71.9 F, G, H 3, 4, 5 3, 3, 2 1.6 0.8, 0.8, 0.6 2000 49.4 F, G, H 3, 4, 5 4, 4, 3 1.1 0.8., 0.8, 0.6 aTotal rainfall and snowmelt from previous June through May of respective study year (U.S. Fish and Wildlife Service unpublished refuge climatological data). bArea (ha) of and number of point count plots sampled for each unit: A, 89 and 9; B, 372 and 20; C, 494 and 20; D, 102 and 4; E, 50 and 3; F, 271 and 10; G, 534 and 15; H, 313 and 10. cListed respectively for units for each year. dFire index is a gauge of fire experience based on total number of prescribed burns divided by number of years since the last burn, with 0 indicating no burns and 6.0 indicating many burns, the last recently (Madden et al. 1999). eAUM = Animal Unit Month.

TABLE 2. Frequency of occurrence (%)a and abundanceb of breeding passerine species during 3 years of rotation grazing following prescribed fire treatments on Lostwood National Wildlife Refuge.

d ______1998 (n = 32) ______1999 (n = 35) ______2000 (n = 35) ______Abundance ______Abundance ______Abundance c – – – Species % x sx– % x sx– % x sx– Eastern Kingbird (Tyrannus tyrannus) 37.5 0.16 0.04 25.7 0.12 0.04 17.1 0.07 0.03 Horned Lark (Eremophila alpestris)*e 15.6 0.05 0.02 0 2.9 0.02 0.02 Sprague’s Pipit (Anthus spragueii)* 15.6 0.05 0.02 34.3 0.14 0.04 31.4 0.16 0.05 Common Yellowthroat (Geothlypis trichas)* 0 11.4 0.05 0.02 8.6 0.03 0.02 Clay-colored Sparrow (Spizella pallida)* 87.5 0.92 0.13 80.0 0.57 0.08 85.7 0.78 0.07 Vesper Sparrow (Pooecetes gramineus)* 28.1 0.15 0.05 11.4 0.06 0.03 11.4 0.05 0.02 Savannah Sparrow (Passerculus sandwichensis)* 100 0.98 0.10 97.1 0.97 0.08 100 1.50 0.09 Baird’s Sparrow (Ammodramus bairdii)* 18.8 0.08 0.04 60.0 0.41 0.07 51.4 0.30 0.06 Grasshopper Sparrow (Ammodramus savannarum)* 34.4 0.20 0.06 48.6 0.29 0.06 60.0 0.59 0.11 LeConte’s Sparrow (Ammodramus leconteii) 18.8 0.07 0.03 25.7 0.12 0.04 25.7 0.09 0.02 Bobolink (Dolichonyx oryzivorus)* 46.9 0.29 0.06 57.1 0.45 0.08 57.1 0.27 0.04 Western Meadowlark (Sturnella neglecta)* 40.6 0.18 0.04 22.9 0.10 0.03 22.9 0.10 0.03 Brewer’s Blackbird (Euphagus cyanocephalus) 50.0 0.28 0.08 17.1 0.07 0.03 0 Brown-headed Cowbird (Molothrus ater)* 68.8 0.69 0.10 60.0 0.69 0.14 77.1 1.07 0.18 aPercentage of 75-m-radius plots at which a species was detected. bMean number of singing males detected per 75-m-radius plot. cOnly common species, i.e., those detected at >10% of all plots, are included. Species less frequently detected included Western Kingbird (Tyrannus verticalis), Sedge Wren (Cistothorus platensis), Gray Catbird (Dumetella carolinensis), Yellow Warbler (Dendroica petechia), Nelson’s Sharp-tailed Sparrow (Ammodra- mus nelsoni), Song Sparrow (Melospiza melodia), and Chestnut-collared Longspur (Calcarius ornatus). dNumber of 75-m-radius plots surveyed. eAsterisks denote breeding bird species characteristic of northern mixed-grass prairie (Stewart 1975:25). 2004] PASSERINES ON BURNED AND GRAZED PRAIRIE 75 comparison, RFD conducted bird surveys in burned prairie that was being grazed by live- 1998–2000 with training and instruction from stock during our study and prairie with a simi- RKM and EMM, who conducted 1994 sur- lar fire history but no grazing, studied earlier veys in Madden et al. (1999). We used fire at LNWR (Table 4). Among common species, indices and weather records to identify years however, Brown-headed Cowbird occurred of comparable fire and precipitation history 2.4 times more frequently on burned and (Table 1; 1994 versus 1999). Units in the com- grazed prairie than on burn-only prairie. parison were reasonably similar in range site and general vegetation makeup (i.e., woody DISCUSSION and exotic vegetation). Makeup of breeding grassland bird species RESULTS at LNWR seemed to change little during the first years of rotation grazing treatment at We detected 21 bird species between 1998 moderate stocking rates in a fire-managed, and 2000: 17 species in both 1998 and 1999 northern mixed-grass prairie. However, Brown- and 14 species in 2000. Thirteen species were headed Cowbirds occurred much more fre- common (detected at >10% of plots) in 1998 quently on burned and grazed prairie than on and 1999 and 11 were common in 2000 (Table burn-only prairie. In the Great Plains the cow- 2). In all years Savannah Sparrow and Clay- bird once associated with American bison colored Sparrow were nearly ubiquitous (de- (Bison bison) herds but now associates with tected at ≥80% of plots) and Brown-headed cattle herds (Lowther 1993). Nest parasitism Cowbird was almost as widespread. Several by Brown-headed Cowbirds can reduce pro- other species were fairly common, although ductivity of northern prairie birds such as occurrence of some varied among years. For ex- Baird’s Sparrow (Davis and Sealy 1998), but ample, Baird’s Sparrow, Grasshopper Sparrow, implications of increased cowbird abundance and Sprague’s Pipit occurred about twice as for host species on burned and grazed mixed- frequently in 1999 and 2000 as in 1998. Vesper grass prairie are currently unmeasured. Sparrow and Western Meadowlark occurred Bird species we documented as common about one-half as frequently in the 2nd and during post-fire grazing treatments of prairie 3rd study years. Horned Lark was rarely de- at LNWR included nearly all species charac- tected after 1998. Abundances of species teristic of upland, northern mixed-grass prairie roughly paralleled their respective frequen- (Stewart 1975:25). Chestnut-collared Longspur cies of occurrence (Table 2). was not common in our study, however, proba- Vegetation density and litter depth on the bly because it favors areas with heavier graz- burn-graze units averaged only about 8–9 con- ing pressure (Kantrud 1981, Hill and Gould tacts and 1 cm in all years (Table 3). All units 1997). Madden et al. (1999) also rarely noted were dominated by a mix of native herbaceous longspurs on ungrazed prairie at LNWR that and native woody vegetation. Vegetation den- had fire histories similar to those of our man- sity, VOR, and maximum height were greater agement units. Two other species listed by on control (burn-only) units than on burn- Stewart (1975)—Lark Bunting (Calamospiza graze units measured the same year (1999; melanocorys) and Red-winged Blackbird (Age- df = 7; t = 4.17, 4.57, and 7.15, respectively; laius phoeniceus)—were not observed on sur- P = 0.004, 0.003, and <0.001), but we de- vey plots in this study. Lark Bunting is no- tected no difference in litter depth (Aspin- madic and occurs sporadically in the LNWR Welch unequal variance test, df = 7, t = 1.52, area (U.S. Fish and Wildlife Service 1998), and P = 0.22). Three of these burn-graze units Red-winged Blackbird is mainly associated were sampled for birds for comparison with an with wetland habitats (Stewart 1975), which earlier study; the mean and variation of struc- were excluded in our sampling. tural characteristics of this subsample were During consecutive years of grazing, the identical to those of respective characteristics bird community closely followed a post-fire of all 5 units, except for total hits (x– = 8.5, s– pattern. Abundances of Baird’s Sparrow, Bobo- – x = 0.5 in subsample, versus x = 8.3, sx– = 0.3 link, Grasshopper Sparrow, and Sprague’s in Table 3). Pipit were lowest in 1998 (x– = 1.8 years post- Frequencies of occurrence for grassland fire), a result similar to a 1st-year, post-fire birds were comparable between prescribe- response in mixed-grass prairie (Pylypec 1991, 76 WESTERN NORTH AMERICAN NATURALIST [Volume 64

TABLE 3. Structure and general composition of vegetation on prescribe-burned, mixed-grass prairie at Lostwood National Wildlife Refuge during rotation grazing treatments 1–4 years post-fire, and on prairie treated by prescribed burning only.

______Burn-graze ______Burn-only a ______1998 (n = 4) ______1999 (n = 5) ______2000 (n = 5) ______1999 (n = 4) – – – – Variable x sx– x sx– x sx– x sx– Density (total hits) 7.6 0.4 8.3 0.3 9.1 0.5 10.9 0.6 Litter depth (cm) 1.2 0.1 1.3 0.1 1.2 0.1 2.4 0.7 VORb (dm) —c —c 1.3 0.1 1.5 0.1 2.0 0.2 Maximum height (dm) 2.0 0.1 2.6 0.1 2.8 0.1 3.5 0.1 Herbaceous % 51.1 2.4 42.8 6.3 44.2 4.4 42.1 3.1 Native % 75.3 4.6 84.6 1.3 88.3 2.1 82.5 2.8 aNumber of management units. bVisual obstruction reading (Robel et al. 1970). cNo data.

TABLE 4. Frequency of occurrence (%)a of singing male passerines on prescribe-burned management units (1994; n = 3) and on burned and grazed units (1999; n = 3) with similar precipitation and fire indices at Lostwood National Wildlife Refuge.

b ______Burn-only ______Burn-graze – – Species x sx– x sx– Sprague’s Pipit 16.7 8.8 30.0 17.3 Clay-colored Sparrow 88.3 7.3 82.2 9.6 Vesper Sparrow 10.3 2.9 12.2 9.0 Savannah Sparrow 93.3 3.3 96.6 3.3 Baird’s Sparrow 60.7 22.5 57.7 18.9 Grasshopper Sparrow 50.0 25.0 43.3 20.2 Bobolink 57.3 19.7 54.4 17.2 Western Meadowlark 25.0 13.2 23.3 8.8 Brown-headed Cowbird 25.0 13.2 60.0 17.3 aPercentage of 75-m-radius plots at which a species was detected. bData for burn-only units are a subset from Madden et al. (1999).

Johnson 1997, Madden et al. 1999). Abun- on burn-only prairie at LNWR with similar dances of these species increased in 1999 and fire histories as reported in Madden (1996: 2000 (x– = 2.7 and 3.7 years post-fire), except 136). When vegetation data were collected for Bobolink, which increased and then decreased both treatments in our study (1999), the in abundance. This pattern of change also burned and grazed prairie also was shorter characterized a fire-only regime studied previ- and sparser than burn-only prairie with simi- ously at LNWR (Madden et al. 1999). Horned lar fire histories. Larks, however, deviated from the typical Models in Madden et al. (2000) predicted post-fire pattern. This species was never com- bird species occurrence based on absolute mon in burn-only prairie studied by Madden measures of vegetation on prescribe-burned, et al. (1999) but was common in the 1st year of mixed-grass prairie at LNWR. For example, our study, possibly because the relatively short, VOR best predicted occurrence of Baird’s post-fire vegetation height and density likely Sparrow, Grasshopper Sparrow, and Sprague’s were further reduced by livestock grazing, Pipit. Baird’s Sparrow occurrence is predicted creating relatively barren habitat preferred by to be 45%–75% when the VOR is 1.3 dm (Mad- the lark (Beason 1995). Plant density, VOR, den et al. 2000). Our data on burn-graze units and litter depth on the burned and grazed (1999) support this prediction; the mean VOR prairie in our study were slightly lower than and associated frequency of occurrence of 2004] PASSERINES ON BURNED AND GRAZED PRAIRIE 77

Baird’s Sparrow were 1.3 dm and 58%, respec- for rapidly assessing composition of northern prairie tively. VOR models in Madden et al. (2000) for vegetation. Ecological Restoration 22: in press. HIGGINS, K.F. 1986. Interpretation and compendium of Grasshopper Sparrow and Sprague’s Pipit historical fire accounts in the northern Great Plains. were similarly supported by our data. These U.S. Fish and Wildlife Service, Resource Publication comparisons with models derived from fire- 161. managed prairie also suggest that alternate HILL, D.P., AND L.K. GOULD. 1997. Chesnut-collared Long- defoliation tools or combinations of tools may spur (Calcarius ornatus). In: A. Poole and F.B. Gill, editors, Birds of North America 288. Academy of be used to reach the same habitat targets for Natural Sciences, Philadelphia, PA, and American grassland bird occurrence at LNWR. Ornithologists’ Union, Washington, DC. A contrasting point between our study and HUTTO, R.L., S.M. PLETSCHET, AND P. H ENDRICKS. 1986. A earlier work by Madden et al. (1999) at LNWR fixed-radius point count method for nonbreeding and breeding season use. Auk 103:593–602. is that we measured bird occurrence and veg- JOHNSON, D.H. 1997. Effects of fire on bird populations in etation during the habitat treatment phase mixed-grass prairie. Pages 181–206 in F. B. Knopf rather than afterward. Clearly, study of effects and F.B. Samson, editors, Ecology and conservation of a combined fire and grazing regime in of Great Plains vertebrates. Springer, New York. northern mixed-grass prairie should include a KANTRUD, H.A. 1981. Grazing intensity effects on the breeding avifauna of North Dakota native grasslands. post-grazing component, in addition to more Canadian Field-Naturalist 95:404–417. extensive replication and random treatment LOWTHER, P.E. 1993. Brown-headed Cowbird (Molothrus and control assignment. Regardless, 2–3 con- ater). In: A. Poole and F.B. Gill, editors, Birds of North secutive summers of rotation livestock grazing America 47. Academy of Natural Sciences Philadel- phia, PA, and American Ornithologists’ Union, Wash- at moderate stocking rates during our study ington, DC. appeared to maintain the relatively high MADDEN, E.M. 1996. Passerine communities and bird- breeding bird diversity noted on fire-treated, habitat relationships on prescribe-burned, mixed grass northern mixed-grass prairie at LNWR by prairie in North Dakota. Master’s thesis, Montana Madden et al. (1999). State University, Bozeman. MADDEN, E.M., A.J. HANSEN, AND R.K. MURPHY. 1999. Influence of prescribed fire history on habitat and ACKNOWLEDGMENTS abundance of passerine birds in northern mixed-grass prairie. Canadian Field-Naturalist 113:627–640. We dedicate this paper to the memory of MADDEN, E.M., R.K. MURPHY, A.J. HANSEN, AND L. MUR- Josh Cole. Josh enthusiastically aided our veg- RAY. 2000. Models for guiding management of prairie bird habitat in northwestern North Dakota. Ameri- etation sampling as did Chandra Ramsdell, can Midland Naturalist 144:377–392. both as Youth Conservation Corps employees. MURPHY, R.K. 1993. History, nesting biology, and preda- Range technicians of the U.S. Fish and Wild- tion ecology of raptors on the Missouri Coteau of life Service’s fire program also assisted. Karen northwestern North Dakota. Doctoral dissertation, Montana State University, Bozeman. Smith provided background information and PYLYPEC, B. 1991. Impacts of fire on bird populations in a helpful perspectives. Comments from Larry Igl, fescue prairie. Canadian Field-Naturalist 105:346–349. Dan Reinking, and an anonymous reviewer ROBEL, R.J., J.N. BRIGGS, A.D. DAYTON, AND L.C. HUL- greatly improved the manuscript. BERT. 1970. Relationship between visual obstruction measurements and weight of grassland vegetation. Journal of Range Management 23:295–298. LITERATURE CITED ROTENBERRY, J.T., AND J.T. WIENS. 1980. Habitat structure, patchiness, and avian communities in North Ameri- BEASON, R.C. 1995. Horned Lark (Eremophila alpestris). can steppe vegetation: a multivariate analysis. Ecol- In: A. Poole and F.B. Gill, editors, Birds of North ogy 61:1228–1250. America 195. Academy of Natural Sciences, Philadel- SOIL CONSERVATION SERVICE. 1984. Technical guide notice phia, PA, and American Ornithologists’ Union, Wash- ND-35. U.S. Department of Agriculture, Bismarck, ington, DC. ND. BRAGG, T.B. 1994. The physical environment of Great Plains STEWART, R.E. 1975. Breeding birds of North Dakota. Tri- grasslands. Pages 49–81 in K.A. Keeler and A. Joern, college Center for Environmental Studies, Fargo, editors, The changing prairie: North American ND. grasslands. Oxford University Press, New York. U.S. FISH AND WILDLIFE SERVICE. 1998. Lostwood National DAVIS, S.K., AND S.G. SEALY. 1998. Nesting biology of Wildlife Refuge comprehensive conservation plan. Baird’s Sparrow in southwestern Manitoba. Wilson U.S. Fish and Wildlife Service, Kenmare, ND. Bulletin 110:262–270. GRANT, T.R., E.M. MADDEN, R.K. MURPHY, M.P. NENNEMAN, Received 22 April 2002 AND K.A. SMITH. 2004. Management-based approach Accepted 16 January 2003 Western North American Naturalist 64(1), ©2004, pp. 78–85

SPAWNING ECOLOGY OF FINESPOTTED SNAKE RIVER CUTTHROAT TROUT IN SPRING STREAMS OF THE SALT RIVER VALLEY, WYOMING

Michael P. Joyce1 and Wayne A. Hubert1

ABSTRACT.—We studied spawning ecology of cutthroat trout (Oncorhynchus clarki) in streams that originate as springs along the Salt River, a Snake River tributary in western Wyoming. We assessed (1) relative numbers of upstream-migrant and resident adults present during the spawning period in spring streams, (2) influence of habitat modification on use of spring streams for spawning, and (3) habitat features used for spawning in spring streams. Four spring streams were studied, 2 with substantial modification to enhance trout habitat and 2 with little or no modification. Modifications consisted primarily of constructing alternating pools and gravel-cobble riffles. Only a small portion of adult fish in spring streams during the spawning period had migrated upstream from the Salt River between March and the middle of June. Larger numbers of adult fish and more redds were observed in the 2 modified streams compared with the 2 streams with little or no modification. Most spawning occurred on constructed riffles with small gravel and over a narrow range of depths and velocities. Cutthroat trout, rainbow trout (Oncorhynchus mykiss), and their hybrids were observed in 1 stream with habitat modifications, indicating that measures to halt invasion by rainbow trout, as well as habitat improvement, are needed to preserve this native trout within the Salt River valley.

Key words: cutthroat trout, Oncorhynchus clarki bouvieri, Snake River, spawning, migration, habitat improvement, redd.

Cutthroat trout (Oncorhynchus clarki) were described fluvial-adfluvial migratory patterns widespread across western North America with of finespotted Snake River cutthroat trout in several described subspecies isolated in vari- the Snake River. It is believed that finespotted ous watersheds (Behnke 1992). Distribution of Snake River cutthroat trout evolved fluvial- finespotted Snake River cutthroat trout over- adfluvial migration because high spring flows lapped Yellowstone cutthroat trout (O. c. bou- and sediment movement in the Snake River vieri), so the debate exists whether the fine- limit spawning success (Kiefling 1978). In con- spotted Snake River form is a unique sub- trast, spring streams provide relatively stable species. The historical range of finespotted flows with little sediment movement during Snake River cutthroat trout begins at the the spring. Snake River below Jackson Lake and contin- The finespotted Snake River cutthroat trout ues downstream to Palisades Reservoir. It also is the only native trout in the Salt River water- includes the downstream portions of tribu- shed (Isaak 2001), the most downstream Snake taries from the Gros Ventre River to the Salt River tributary where the fish occurred natu- River (Behnke 1992). Large-spotted Yellow- rally. Cutthroat trout are declining throughout stone cutthroat trout occur naturally in the their natural range for several reasons (Behnke headwaters of several Snake River tributaries 1992, Duff 1996, Kruse et al. 2000), many of in this area (Behnke 1992). which may be affecting finespotted Snake Finespotted Snake River cutthroat trout ex- River cutthroat trout in the Salt River water- hibit both fluvial and fluvial-adfluvial migratory shed. Rainbow trout (Oncorhynchus mykiss), patterns (Varley and Gresswell 1988, Northcote which can hybridize with cutthroat trout, and 1997). Fluvial fish reside and move within a brown trout (Salmo trutta) and brook trout single stream or river segment throughout life, (Salvelinus fontinalis; both are competitors of whereas fluvial-adfluvial fish reside in a main- cutthroat trout) have become naturalized in stream river and migrate seasonally into tribu- the valley (Hudelson 1995). Habitat has been taries. Hayden (1968) and Kiefling (1978, 1997) affected by numerous anthropogenic activities,

1U.S. Geological Survey, Wyoming Cooperative Fish and Wildlife Research Unit, University of Wyoming, Laramie, WY 82071-3166.

78 2004] SPAWNING ECOLOGY OF CUTTHROAT TROUT 79 particularly (1) by water diversion, causing per- reach; Bee Creek had no constructed pool-riffle iodic dewatering of long segments of tribu- pairs. Near the mouth of the spring streams, taries and the Salt River headwater (Coving- mean daily discharge from March through July ton and Hubert 2003), and (2) by bank erosion, 2000 was 0.51 m3 ⋅ s–1 (range, 0.51–0.62 m3 ⋅ s–1) contributing sediment to streams throughout in Christensen Creek, 0.33 m3 ⋅ s–1 (range, the watershed (Gelwicks et al. 2003). 0.28–0.38 m3 ⋅ s–1) in Perk Creek, 1.11 m3 ⋅ s–1 A need exists for better understanding of (range, 0.89–1.88 m3 ⋅ s–1) in Anderson Creek, the ecology of finespotted Snake River cut- and 0.15 m3 ⋅ s–1 (no measurable variation) in throat trout in order to preserve and manage Bee Creek (Joyce 2001). Mean daily water this native fish for future benefit. We studied temperatures in the 4 spring streams were the spawning ecology of finespotted Snake similar, ranging from 5°C in early March to river cutthroat trout in spring streams because 15°C in late July 2000 (Joyce 2001). of the hypothesized value of spring streams for fluvial-adfluvial fish. Our objectives were to METHODS assess (1) relative numbers of upstream-migrant and resident adults during the spawning period A weir was placed at the downstream end of in spring streams, (2) influence of habitat mod- the study reaches in Christensen and Ander- ification on use of spring streams for spawn- son Creeks to capture fish moving upstream ing, and (3) habitat features used for spawning (Fig. 2). Weirs were aluminum rods placed 13 in spring streams. mm apart and fitted into racks. The Christensen Creek weir was 1800 m upstream from the STUDY AREA Salt River, and the Anderson Creek weir was 1600 m upstream. No potential spawning habi- The Salt River watershed encompasses 2150 tat was observed downstream of the weirs. km2 in western Wyoming and eastern Idaho Cutthroat trout caught in the weirs were mea- (Fig. 1). A 6th-order stream at its mouth, it has sured for total length (TL, mm) and observed a mean annual discharge of 22.5 m3 ⋅ s–1. River for fungal growth. Adults (≥30 cm TL) free of channel elevations range from 1750 m to 2150 fungus were tagged (Floy T-bar FD-94) behind m. The lower 72 km of the Salt River is peren- the dorsal fin, their adipose fin was clipped, nial due to spring stream input, but upstream and these fish were released upstream of the the river is dewatered annually to support irri- weir. Two trap nets (20-mm bar mesh) were gated agriculture. Finespotted Snake River cut- placed upstream of each weir to capture fish throat trout occur throughout the perennial moving downstream. Cutthroat trout caught in reach, but little spawning habitat is available these traps were checked for tags and fin clips, due to low channel slope and extensive silt de- measured, and released downstream of the position (Isaak 2001, Gelwicks et al. 2002). weir. We installed weirs and trap nets 7 March Spring streams adjacent to the Salt River 2000 and removed them 14 June 2000. flow short distances (Fig. 1; Isaak 2001), and We snorkeled to assess adult cutthroat trout most have wide, shallow channels dominated abundance. Riffle, pool, glide, and culvert loca- by sand and silt substrates with few patches of tions were recorded for each stream. Streams clean gravels suitable as cutthroat trout spawn- were stratified into reaches between culverts ing substrate. To enhance trout habitat, private because these conduits could impede upstream landowners have modified segments of a small movement (Fig. 2). Reaches where water was number of streams by constructing pools for dammed and ponded were omitted, along with adult habitat and cobble-gravel riffles that pro- reaches with no pool or glide habitat greater vide potential spawning sites (Kiefling 1997). than 30 cm deep; 20% of the pools and glides We studied 4 spring streams in the Salt in each reach were randomly selected for sam- River valley (Fig. 1). Preliminary surveys iden- pling (14 in Anderson Creek, 13 in Christensen tified that Christensen Creek and Perk Creek Creek, 9 in Perk Creek, and 6 in Bee Creek). had, respectively, 45 and 32 constructed pool- Pools and glides were snorkeled 5 times in riffle pairs over the length of study reaches; Christensen and Anderson Creeks, 3 times in Anderson Creek had 6 contructed pool-riffle Perk Creek, and twice in Bee Creek between pairs all in the upstream portion of the study 30 March and 14 June 2000. 80 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Fig. 1. Location of study streams in the Salt River valley, Wyoming.

Snorkel counts were expanded to estimate SRCE (A/a). We calculated abundance of adult cutthroat trout abundance using the unmarked fish by summing abundance esti- equation, mates for each reach. Number of immigrants in each stream at the time of snorkeling was SRCE = 1.04 + 1.07 SRCS, estimated by totaling the number of fish captured in the weir and released upstream and where SRCS = number of fish observed and then subtracting the number captured in the SRCE = estimated abundance from depletion trap nets and released downstream to that date. electrofishing. We developed the regression Habitat in the spring streams was measured 2 equation (r = 0.95) using estimates from 24 in each pool, riffle, and glide habitat unit (Bis- pools and glides sampled by snorkeling and 3- son et al. 1981). To estimate water surface area, or 4-pass depletion electrofishing during spring we measured length across 2 representative 2000 in Christensen and Anderson Creeks (see transects and wetted width following the thal- Joyce 2001, Joyce and Hubert 2003). Snorkel weg. Water depth was measured at several counts of fish observed without tags or fin points, and proportions (nearest 5%) of 5 sub- clips were used to estimate SRCE. SRCE esti- strate classes were visually estimated: (1) clay, mates for each pool or glide in a reach were silt, or sand, <2 mm in diameter; (2) small Σ summed ( SRCE), total area of all pools and gravel, 2–20 mm; (3) large gravel, 21–64 mm; glides (A) and total area snorkeled (a) in the (4) cobble, 65–256 mm; or (5) boulder, >256 reach were determined, and fish abundance mm (Bain et al. 1985). In riffles, an upstream Σ (Ni) in the reach was estimated as Ni = transect was established near the riffle edge, 2004] SPAWNING ECOLOGY OF CUTTHROAT TROUT 81

Fig. 2. Maps of the Christensen, Anderson, Perk, and Bee Creeks study areas showing the locations of culverts and reaches in each stream, and locations of weirs in 2 sampled streams. and a 2nd transect was located at half the riffle not be distinguished (usually the tailspill), or length. Water depth and velocity (0.6 of depth) when the redd was under woody riparian veg- were measured at 1/4, 1/2, and 3/4 of the etation or overhanging bank. We determined width across both transects. that fish were continuing to spawn if gravels We surveyed each stream for redds at least were clean because periphyton quickly devel- 6 times from March through mid-July 2000 by oped on undisturbed gravel. Depth, velocity walking along the bank and observing redds in (0.6 of depth), and substrate at redds were riffles. Redds were identified by a patch of measured at the pit front, pit bottom, tailspill gravel that was clean of periphyton and by the front, tailspill crest, and tailspill end (Grost et presence of a pit and tailspill (Crisp and Car- al. 1991). We visually identified dominant (cov- ling 1989). Locations of new redds were re- ering the most surface area) and subdominant corded during each survey. Redd features were (covering the 2nd most surface area) substrates measured except when the configuration could of the pit and tailspill. 82 WESTERN NORTH AMERICAN NATURALIST [Volume 64

TABLE 1. Total surface area of pools and glides in the study area, total surface area of pools and glides in which snorkeling counts were made, estimated abundance of unmarked adult (≥30 cm total length) finespotted Snake River cutthroat trout in the study areas on each sampling date, and number of marked fish present in the stream on each sampling date in Christensen and Anderson Creeks during spring 2000. Surface area Surface area (m2) Estimated unmarked fish and number of Location (m2) in snorkeling counts marked fish (in parentheses) by date Christensen Creek 27,806 16,350 16 April 27 April 5 May 23 May 10 June 499 (36) 710 (46) 687 (65) 624 (83) 535 (117) Anderson Creek 38,775 35,120 17 April 2 May 10 May 24 May 6 June 184 (1) 170 (7) 216 (8) 154 (11) 207 (12)

RESULTS ing counts. Estimated abundance of adult trout ranged from 110 to 203 fish during 3 sampling Movements dates from 20 April to 11 June 2000. However, More adult cutthroat trout were captured adult cutthroat trout, rainbow trout, and cut- moving upstream into Christensen Creek than throat trout × rainbow trout hybrids were ob- into Anderson Creek. In Christensen Creek, served while snorkeling and electrofishing. 151 fish were captured by the weir, but in The snorkeler could not accurately determine Anderson Creek only 15 fish were captured by if observed fish were cutthroat trout, rainbow the weir. Capture of fish by weirs began in trout, or cutthroat trout × rainbow trout hybrids, early April and peaked in late May and early so abundance estimates of cutthroat trout could June. Temporal pattern was similar in both not be made in Perk Creek. streams. Six reaches were identified in Bee Creek More adult cutthroat trout were captured (Fig. 2) with 19,000 m2 of pool and glide habi- moving downstream in Christensen Creek tat. Snorkeling counts were made over the than in Anderson Creek. Of 131 fish captured entire study area twice during the spawning moving downstream in Christensen Creek, 33 period, and no adult trout were observed in were marked. Peak downstream movement in Bee Creek. Christensen Creek occurred during the week Proportions of Immigrants of 17–23 May 2000. In Anderson Creek, 5 fish were captured moving downstream and 3 Cutthroat trout that had moved upstream were marked. and were captured by the weirs comprised a small proportion of adult fish in both Chris- Abundance tensen and Anderson Creeks (Table 1). Fish We identified 7 reaches in Christensen Creek captured, tagged, and released upstream from (Fig. 2) with total pool and glide habitat com- the weir comprised 6%–18% of the estimated prising 27,806 m2 of the study area, and we total number of adult fish in Christensen Creek made snorkeling counts in 59% of this habitat. and 1%–7% in Anderson Creek during the 5 Estimated abundances of unmarked adult cut- snorkeling dates. throat trout ranged from 499 to 710 fish during the 5 sampling dates (Table 1). Marked fish Distributions of Redds were observed over the entire study area. Large numbers of redds were observed over Five reaches were identified in Anderson the length of the study areas in Christensen Creek (Fig. 2) with pool and glide habitat com- and Perk Creeks where substantial habitat prising 38,775 m2 of the study area, and snor- modifications had occurred. There were 49 keling counts were made in 91% of this habi- riffles in Christensen Creek: 22 in reach 1, 5 in tat. Estimated abundances of unmarked adult reach 2, 0 in reach 3, 5 in reach 4, 0 in reach 5, fish ranged from 154 to 216 fish during 5 sam- 2 in reach 6, and 15 in reach 7 (Fig. 2); 64 redds pling dates (Table 1). Marked fish were observed were found on 24 of 49 riffles. Reach 7 had the over the entire study area. greatest number of redds (39) and reach 1 had Two reaches were identified in Perk Creek the 2nd highest number (14). Most redds were (Fig. 2) with 12,117 m2 of pool and glide habi- on constructed riffles. Clean gravels indicating tat, all of which were included in the snorkel- active spawning were observed in Christensen 2004] SPAWNING ECOLOGY OF CUTTHROAT TROUT 83

Creek from early April through late June; peak TABLE 2. Mean water depths (cm) and water velocities spawning activity occurred around 1 May 2000. (cm per second) at 0.6 water depth with 95% confidence intervals (in parentheses) at 5 locations in redds measured In Perk Creek, 36 riffles were identified, 22 among 57 redds in Christensen Creek during spring 2000. in reach 1 and 9 in reach 2 (Fig. 2); 72 redds were found on 27 riffles. Again, most redds Location Depth Velocity were on constructed riffles. However, we could Pit front 23 (22–25) 47 (43–51) not determine if redds were the result of spawn- Pit bottom 30 (29–32) 43 (39–47) ing by cutthroat trout, rainbow trout, or their Tailspill front 23 (22–25) 47 (22–72) hybrids. Tailspill crest 15 (14–16) 65 (61–69) Four redds were observed in Anderson Creek Tailspill end 26 (24–30) 41 (35–47) and none in Bee Creek where there was little or no habitat modification. In Anderson Creek we identified 52 riffles: 8 in reach 1, 22 in reach 2, 3 in reach 3, 15 in reach 4, and 4 in reach 5. that migrated into 2 spring streams during the One redd was found on each of 4 riffles in spawning season was a small portion of the Anderson Creek. In Bee Creek, only 2 riffles total number of adult fish in each stream, sug- were identified and no redds were observed. gesting that most spawning in spring streams was by fish with a fluvial life history. Varley Features of Redds and Gresswell (1988) found fluvial, fluvial- We measured features of 57 redds. In adfluvial, lacustrine-adfluvial (i.e., adults spend Christensen Creek, where the cutthroat trout most of their life in lakes and ascend tribu- was the only Oncorhynchus observed during taries to spawn), and allacustrine (i.e., adults spring 2000 (Table 2), relatively narrow confi- spend most of their life in lakes and move into dence intervals suggested substantial similar- lake outlets to spawn) migratory patterns among ity among redds. Small gravel most frequently Yellowstone cutthroat trout in the Yellowstone dominanted substrate, and large gravel was Lake drainage. It is likely that a wide range of the most frequent subdominant substrate in movement patterns commonly exists among the pit of redds. Small gravel was the most fre- Yellowstone cutthroat trout populations across quent dominant substrate and sand the most their natural range. frequent subdominant substrate in the tailspill Christensen and Perk Creeks had more sub- of redds. stantial pool and riffle habitat enhancement, Cutthroat trout constructed redds in portions more adult fish, and more redds than Anderson of riffles with smaller gravel, shallower water and Bee Creeks, which had little or no habitat depths, and greater water velocities than that modification, few pools, few or no adult fish, which was generally available among riffles in and few or no redds. Only cutthroat trout were Christensen Creek. Median water depth in rif- observed in Christensen and Anderson Creeks, fles was 28 cm, whereas median depth at the so numbers of adult fish and redds were in- front of redd pits was 23 cm. Median water dicative of use by this species. However, we velocity in riffles was 0.40 m ⋅ s–1, but median could not determine numbers of adult cut- water velocity at the front of redd pits was throat trout or redds constructed by them in 0.44 m ⋅ s–1. Riffles in Christensen Creek aver- Perk Creek because of the occurrence of both aged 56% small gravel, but small gravel was rainbow trout and cutthroat trout × rainbow the dominant substrate in 74% of the redd pits trout hybrids in the stream. Habitat modifica- and in 84% of the redd tailspills. tions in Christensen and Perk Creeks converted long, shallow glides to pool and riffle DISCUSSION complexes. Spawning habitat was improved by placement of gravel in riffles between pools, and We captured cutthroat trout moving up- habitat for adults was created in constructed stream and downstream in 2 spring streams pools adjacent to riffles. A substantial increase during the spawning season, suggesting that a in immigrant spawners and an estimated spawn- fluvial-adfluvial migratory pattern may occur ing population increase of almost sixfold among at least some adults in the Salt River occurred where spawning habitat was improved (Varley and Gresswell 1988, Northcote 1997). in spring streams tributary to the Snake River However, the number of adult cutthroat trout (Kiefling 1981). Similarly, cutthroat trout redd 84 WESTERN NORTH AMERICAN NATURALIST [Volume 64 densities were related to abundance of spawn- hybrids were also found in Perk Creek in 2000 ing gravels in Montana streams, but no other (Evans and Shiozawa 2001, Joyce 2001). These measured habitat features appeared to affect observations suggest that habitat modifications redd densities (Magee et al. 1996). to enhance trout habitat may be beneficial to Timing of upstream movements of adult cutthroat trout, but their preservation in the cutthroat trout into Christensen and Anderson Salt River system must also involve efforts to Creeks was similar, with peak movements in halt (1) invasion by rainbow trout, (2) hybridi- late May and early June. In upstream Snake zation of native cutthroat trout with rainbow River tributaries, spawning migrations by cut- trout, and (3) development of hybrid swarms, throat trout have been observed as early as as has been observed in other watershed late February and as late as early August, but (Kruse et al. 2000). peak spawning occurs in April (Hayden 1968). Although only narrow ranges of variability in ACKNOWLEDGMENTS features of cutthroat trout redds were observed in Christensen Creek, features of redds We thank C. Mattix, R. Gipson, and D. Zafft vary among streams. For example, measure- for help in the field, numerous landowners in ment of Yellowstone cutthroat trout redds in Salt River valley for allowing us to sample on Idaho indicated that mean water depth at the their property, and T. Johnson for all of his front of the pit (Thurow and King 1994) was assistance. Funding was provided by the Wyo- similar to that observed in Christensen Creek, ming Game and Fish Department. The Wyo- whereas mean water depth used by Westslope ming Cooperative Fish and Wildlife Research cutthroat trout in the Blackfoot River, Montana, Unit is jointly funded by the U.S. Geological was less (Schmetterling 2000). Mean water Survey, University of Wyoming, Wyoming Game velocity at the front of redds in Christensen and Fish Department, and Wildlife Manage- Creek was slower than for Westslope cutthroat ment Institute. trout in Blackfoot River, Montana tributaries (Schmetterling 2000). It is likely that the size LITERATURE CITED of spawning adults and the amount of avail- BAIN, M.B., J.T. FINN, AND H.E. BOOKE. 1985. Manage- able habitat contribute to variation in redd ment briefs qualifying stream substrate for habitat features of cutthroat trout among stream sys- analysis studies. North American Journal of Fish- tems. eries Management 5:499–506. In Christensen Creek small gravel (2–20 BEHNKE, R.J. 1992. Native trout of western North Amer- mm) was the most frequent substrate observed ica. American Fisheries Society Monograph 6. Amer- ican Fisheries Society, Bethesda, MD. in both riffles and redds. Hayden (1968) re- BISSON, P.A., J.L. NIELSON, R.A. PALMASON, AND L.E. ported that cutthroat trout in Snake River GROVE. 1981. A system of naming habitat types in spring streams preferred gravels of 25–64 mm small streams, with examples of habitat utilization by but did not comment on availability of differ- salmonids during low stream flow. Pages 62–73 in N.B. Armantrout, editor, Acquisition and utilization ent substrate sizes. Spawning substrates used of aquatic habitat inventory information. Western by cutthroat trout in Christensen Creek were Division American Fisheries Society, Portland, OR. similar to those for Yellowstone cutthroat trout COVINGTON, J.S., AND W. A . H UBERT. 2003. Trout popula- (Thurow and King 1994) and Westslope cut- tion responses to restoration of stream flows. Envi- throat trout (Magee et al. 1996, Schmetterling ronmental Management 31:135–146. CRISP, D.T., AND P.A. CARLING. 1989. Observations on sit- 2000) in other systems. ing, dimensions and structure of salmonid redds. Our observations suggest that the construc- Journal of Fish Biology 34:119–134. tion of pools and cobble-gravel riffles in spring DUFF, D.A., EDITOR. 1996. Conservation assessment for streams is likely to benefit both fluvial-adflu- inland cutthroat trout status and distribution. U.S. vial and fluvial finespotted Snake River cut- Department of Agriculture, Forest Service, Inter- mountain Region, Ogden, UT. throat trout in the Salt River valley. However, EVANS, R.P., AND D.K. SHIOZAWA. 2001. The genetic status we observed adult cutthroat trout, rainbow of trout from Perk Creek. Final report. Brigham Young trout, and cutthroat trout × rainbow trout hy- University, Provo, UT. brids during the spawning season in Perk Creek, GELWICKS, K.R., D.J. ZAFFT, AND R.G. GIPSON. 2002. Com- 1 of the 2 study streams with substantial habi- prehensive study of the Salt River fishery between Afton and Palisades Reservoir from 1995 to 1999 with tat modification. Age-0 cutthroat trout, rain- historical review: fur trade to 1998. Wyoming Game bow trout, and cutthroat trout × rainbow trout and Fish Department, Fish Division, Cheyenne. 2004] SPAWNING ECOLOGY OF CUTTHROAT TROUT 85

GROST, R.T., W.A. HUBERT, AND T.A. WESCHE. 1991. ______. 1997. A history of the Snake River spring creek Description of brown trout redds in a mountain spawning tributaries. Wyoming Game and Fish stream. Transactions of the American Fisheries Soci- Department, Fish Division, Administrative Report, ety 120:582–588. Cheyenne. HAYDEN, P.S. 1968. The reproductive behavior of Snake KRUSE, C.G., W.A. HUBERT, AND F. J . R AHEL. 2000. Status River cutthroat trout in three tributary streams in of Yellowstone cutthroat trout in Wyoming waters. Wyoming. Master’s thesis, University of Wyoming, North American Journal of Fisheries Management Laramie. 20:692–704. HUDELSON, R.A. 1995. An inventory survey of waters in MAGEE, J.P., T.E. MCMAHON, AND R.F. THUROW. 1996. the Salt River drainage, Lincoln County, Wyoming. Spatial variation in spawning habitat of cutthroat Wyoming Game and Fish Department Administra- trout in a sediment-rich stream basin. Transactions tive Report, Cheyenne. of the American Fisheries Society 125:768–779. ISAAK, D.J. 2001. A landscape ecological view of trout pop- NORTHCOTE, T.G. 1997. Potamodromy in salmonidae: liv- ulations across a Rocky Mountain watershed. Doc- ing and moving in the fast lane. North American toral dissertation, University of Wyoming, Laramie. Journal of Fisheries Management 17:1029–1045. JOYCE, M.P. 2001. Reproduction of Snake River cutthroat SCHMETTERLING, D.A. 2000. Redd characteristics of flu- trout in spring streams tributary to the Salt River, vial westslope cutthroat trout in four tributaries to Wyoming. Master’s thesis, University of Wyoming, the Blackfoot River, Montana. North American Jour- Laramie. nal of Fisheries Management 20:776–783. JOYCE, M.P., AND W.A. HUBERT. 2003. Snorkeling as an THUROW, R.F., AND J.G. KING. 1994. Attributes of Yellow- alternative to depletion electrofishing for assessing stone cutthroat trout redds in tributary of the Snake cutthroat trout and brown trout in stream pools. River, Idaho. Transactions of the American Fisheries Journal of Freshwater Ecology 18:215–222. Society 123:37–50. KIEFLING, J.W. 1978. Studies of the ecology of the Snake VARLEY, J.D., AND R.E. GRESSWELL.1988. Ecology, status, River cutthroat trout. Wyoming Game and Fish De- and management of the Yellowstone cutthroat trout. partment, Fisheries Technical Bulletin 3, Cheyenne. American Fisheries Society Symposium 4:13–24. ______. 1981. Snake River investigations. Federal Aid Project Completion Report F-37-R. Wyoming Game Received 18 March 2002 and Fish Department, Cheyenne. Accepted 4 February 2003 Western North American Naturalist 64(1), ©2004, pp. 86–92

HABITAT OF THREE RARE SPECIES OF SMALL MAMMALS IN JUNIPER WOODLANDS OF SOUTHWESTERN WYOMING

Kevin M. Rompola1,2 and Stanley H. Anderson1

ABSTRACT.—Southwestern Wyoming constitutes the northern limit of the ranges of the cliff chipmunk (Tamias dorsalis), pinyon mouse (Peromyscus truei), and canyon mouse (P. crinitus). In addition to trying to determine their presence in the region, we wanted to identify habitat characteristics commonly used by each of these species. We used Sherman live-traps to sample 14 sites representing 2 distinct habitat types in 1998 and 1999: juniper-rocky slopes and juniper cliffs. Seventeen habitat characteristics were measured at capture locations for each species and compared with randomly located points. Best subsets multiple logistic regression was used to construct models that distinguish between used and available habitat for each species. The cliff chipmunk occurred in both rocky slopes and cliffs. The pinyon mouse was also captured in rocky slopes and cliffs and was most often captured in locations in the interior of the juniper woodland with high tree canopy cover, high forb cover, and low density of rock outcrops. The canyon mouse was captured only in cliffs at sites consisting of high forb cover, high rock cover, and high tree density.

Key words: juniper, cliff chipmunk, Tamias dorsalis, pinyon mouse, Peromyscus truei, canyon mouse, Peromyscus crinitus, habitat, logistic regression, information theory.

Effective wildlife management and conser- ern Wyoming (Clark and Stromberg 1987), and vation rely on biologists’ understanding factors they may have experienced some habitat loss that influence the distribution of species in- as a result of construction of Flaming Gorge cluding those at the periphery of their range. Dam and the subsequent creation of Flaming Research focusing on the ecology of game Gorge Reservoir in the early 1960s. species has given biologists considerable Our objectives were to determine occurrence knowledge of the factors affecting their num- of the cliff chipmunk, pinyon mouse, and canyon bers and distribution, thus enabling managers mouse in the juniper ( Juniperus osteosperma) to make informed decisions on management. woodlands of southwestern Wyoming and pro- However, this information is largely unavail- vide data on habitat association. To do this we able for many nongame species. For instance, measured variables representing microhabitat small mammals are an important prey compo- characteristics at capture locations and com- nent of most ecosystems in which they occur pared them with randomly located sites. The (Vaughan 1986), but we often do not have ade- results provided information on their distribu- quate information on such species to make tion and factors influencing the distribution of informed management decisions (Gibson 1988). these 3 small mammal species at the edge of The cliff chipmunk (Tamias dorsalis), pinyon their range. mouse (Peromyscus truei), and canyon mouse (P. crinitus) are found throughout the Great MATERIALS AND METHODS Basin (Burt and Grossenheider 1980). While populations of these species are considered The study area is in southwestern Wyoming stable throughout their geographical distribu- in south central Sweetwater County. Trapping tion, they are considered rare in Wyoming, the took place south of Rock Springs, Wyoming, to northern extent of their range (Fertig 1997, the east of Flaming Gorge Reservoir and north Luce and Oakleaf 1998), which constitutes the of the Utah and Wyoming border. A “naturally northern limit of their geographical distribu- patchy” juniper woodland and sagebrush-grass- tion (Burt and Grossenheider 1980, Clark and land mosaic characterize the landscape. In Stromberg 1987). The species are known to general, big sagebrush (Artemisia tridentata) occur only in Sweetwater County in southwest- dominates the lower elevations (1860 m, near

1Wyoming Cooperative Research Unit, Department of Zoology and Physiology, University of Wyoming, Laramie, WY 82071-3166. 2Corresponding author.

86 2004] RARE MAMMALS AT RANGE LIMIT 87

Flaming Gorge Reservoir), with juniper wood- Traps that we found closed but without a cap- lands occupying ridges and slopes. Pinyon pine ture were considered half a trap night. Traps (Pinus edulis) occurs at very low densities in remained closed during the daylight to pre- the southern portion of the study area. In vent small mammals from being captured dur- addition to sagebrush, true mountain mahogany ing periods of high temperatures. (Cercocarpus montanus) is another common We identified captured animals to species; shrub species found throughout the study area. each was sexed, examined for reproductive The primary land use is cattle grazing and rec- status, weighed, and marked with a uniquely reation activities such as hunting and camping. numbered ear tag (Monel size 1, National Band Juniper-dominated rocky slopes and cliff and Tag). All captured individuals were re- areas were identified as 2 dominant habitat leased at the point of capture. components of the probable landscape for these We measured 17 habitat variables (Table 1) species. Low-gradient slopes characterize the at 21 randomly located points within each trap- rocky slopes with their moderate to high juniper ping grid and at the 1st trapping location of tree canopy cover and variable amounts of her- each individual cliff chipmunk, pinyon mouse, baceous understory ground cover. Isolated and canyon mouse. Locations of 3 random rock outcrops are found throughout the rocky points were determined on each row of traps slope habitat. Cliffs occur in areas characterized within the grid. A 3-digit random number was by high-gradient slopes, with juniper as the used to determine the location of the random dominant vegetation component, and shrubs, point along the length of the row of traps, and grasses, and forbs common in the understory. a 2-digit random number indicated direction After an initial review of the area, we selected and distance of the point from the row. Even the sites so that they represented the habitat numbers placed the random points to the right groups and the area. of the row, and odd numbers to the left. This From May through August 1998 and 1999, point, marked using a fluorescent flag, indicated we conducted small mammal surveys. Mammal the center of a circular sampling plot with an trapping was conducted at 7 rocky slopes and 8-m radius encompassing 0.02 ha. 7 cliff sites using 7-cm × 9-cm × 23-cm Sher- We used the criteria of Dueser and Shugart man-live traps, arranged in grids consisting of (1978) to select habitat variables that were 49 traps with 15-m spacing between traps. measured: (1) each variable should provide a The exact configuration of the trapping grid measure of the structure of the environment often depended on size and shape of the habi- which is either known or reasonably suspected tat patch being sampled. In general, we estab- to influence the distribution and local abun- lished 7 × 7-m grids; however, cliff sites often dance of small mammals; (2) each variable were too narrow for such configurations. In should be quickly and precisely measurable these areas, to maintain a more or less equal with nondestructive sampling procedures; effort at all sites, our grids consisted of 3 rows (3) each variable should have intraseason vari- of 12 traps and 1 row of 13 traps. Trap grids ation that is small relative to interseason varia- ranged from 0.74 ha to 0.81 ha because of this tion; and (4) each variable should describe the variation in trap grid configuration. environment in the immediate vicinity of the Traps were baited with a combination of capture. The variables selected represent 3 rolled oats and peanut butter, and to each trap strata: tree overstory, understory, and ground we added polyester bedding for thermal insu- cover. lation to decrease the mortality rate of cap- From the center of each habitat sampling tured individuals exposed to low overnight plot, we measured distance to the nearest log, temperatures. Each trapping session consisted diameter of that log, distance to nearest barren of 4 consecutive nights. Traps were opened in expanse of rock, and distance to the nearest the evening at approximately 1900 hours and edge of juniper vegetation type. Tree and shrub checked and closed beginning at 0700 hours. density as well as average tree diameter for Trapping was performed at 4 sites (2 of each the random point and capture location was habitat type) simultaneously with 1 session of determined using the point-quarter method 2 grids only. Two trapping sessions were con- described by Cottam and Curtis (1956). Height ducted at each site: 1 between 18 May and 30 of the nearest shrub in each quarter was also June, and again between 7 July and 12 August. measured to estimate average shrub height. 88 WESTERN NORTH AMERICAN NATURALIST [Volume 64

TABLE 1. Mean values for each habitat variable measured at random points in rocky slope and cliff habitat types and at cliff chipmunk, pinyon mouse, and canyon mouse capture locations in southwestern Wyoming in 1999. Values with the same superscsript letter are significantly different (P < 0.05). Random Cliff Pinyon Canyon Variable plots chipmunk mouse mouse % tree canopy cover 17.0b 17.0 25.9 13.1 % grass cover 5.6a,b 3.5a 3.2b 4.1 % forb cover 2.2a,b 1.7a 1.0b 3.7 % shrub cover 4.6 4.8 2.9 5.3 % litter cover 19.1 20.5 21.4 13.6 % bare ground 56.3a,c 50.8a 58.0 42.2c % rock cover 13.1a,c 18.2a 13.9 35.6c Shrub density (no. ha–1) 2879 1981 2563 2052 Average shrub height (cm) 48.3a 58.9a 46.9 62.4 Tree density (no. ha–1) 251 310 386 232 Distance to nearest tree (m) 4.6a,b 3.6a 2.7b 3.8 Distance to nearest shrub (m) 2.4b 2.2 1.8b 2.1 Distance to nearest log (m) 7.8a,b,c 4.6a 3.0b 3.5c Diameter of nearest log (cm) 17.0a 17.9a 19.2 21.8 Distance to nearest rock outcrop (m) 10.9a,b 6.6a 15.8 2.1c Distance to juniper woodland edge (m) 47.8a,b 68.7a 237.7b 77.4

Two transects, 1 going north–south and the the number of variables to be used in multi- other east–west, were established through variate analysis. Variables with a univariate each plot center point. Along each transect we significance of P < 0.25 were considered sig- established Daubenmire quadrats at the cen- nificant and were included in the full model ter point and at 4 m and 8 m from the center while the others were discarded. Hosmer and point for a total of 9 quadrats (Daubenmire Lemeshow (1989) suggested using this level of 1959). On each quadrat we estimated the pro- significance because lower levels of significance portion of ground cover by grass, forbs, shrubs, may eliminate variables that may increase the litter, bare ground, and rock. Because of the goodness-of-fit of a reduced model to the data. small stature of juniper trees, we were unable Best subsets multiple logistic regression to use an ocular tube to measure tree canopy was used to determine which combinations of cover. Instead, we used a 5-m pole placed per- the variables identified in the univariate analy- pendicular to the ground at 1 m apart along ses provided the best model for predicting the both transects. Tree canopy cover was measured occurrence of each rodent species. Best sub- in 3 strata (Gilbert and Allwine 1991; 0–2, 2–4 sets is an effective model-building technique and 4+ m) using the pole marked in 2-m seg- that identifies collections of variables, all of ments. Tree canopy cover in any of the 3 strata which could possibly be weakly associated was recorded when a tree branch touched the with the response variable but are important predictors when taken together (Hosmer and pole within the respective 2-m intervals. Lemeshow 1989). We used the Akaike’s Infor- Habitat data collected at random points mation Criterion (AIC) to determine the best were combined across grids for each habitat model from the larger set. The AIC scored type. In addition, data collected at all capture each model according to number of parame- locations were combined across grids for each ters and goodness-of-fit of the model, the model rodent species. To eliminate multicollinearity with the lowest AIC value being considered the in multivariate analyses, we used Pearson pro- most efficient. However, final model selection duct moment correlation analysis to test for was also based on biological interpretability. linear correlation between all possible pairs of variables measured. In cases where the corre- RESULTS lation coefficient (CC) was >0.625, the variable perceived as having the least biological In 5397.5 trap nights, we captured 113 in- significance was eliminated subjectively from dividual cliff chipmunks, 19 pinyon mice, and further analyses. Univariate binary logistic re- 13 canyon mice. The cliff chipmunk was cap- gression analysis was used to further reduce tured in 13 of 14 sites sampled, the pinyon 2004] RARE MAMMALS AT RANGE LIMIT 89

TABLE 2. Univariate significance (P-values) of each habitat variable measured in predicting cliff chipmunk, pinyon mouse, and canyon mouse occurrence in rocky slope and cliff habitat types in southwestern Wyoming in 1999. Signifi- cant at P ≤ 0.05.

______Species Cliff chipmunk Pinyon mouse Canyon mouse Variable (n) (119) (19) (13)

DISCRETE Shrub density Low Medium 0.98 0.84 0.43 High 0.57 0.57 0.36 Tree density Low Medium 0.02* 0.64 0.30 High 0.09 0.63 0.23 Average tree size Small Medium 0.95 0.47 0.47 Large 0.59 0.89 0.89

CONTINUOUS Canopy cover 0.25 0.00* 0.58 % grass cover 0.00* 0.10 0.60 % forb cover 0.08 0.02* 0.00* % shrub cover 0.77 0.37 0.77 % litter cover 0.44 0.41 0.22 % bare ground cover 0.02* 0.66 0.02* % rock cover 0.01* 0.51 0.00* Average shrub height 0.00* 0.50 0.32 Distance to nearest tree 0.04* 0.05* 0.33 Distance to nearest shrub 0.71 0.66 0.66 Distance to nearest log 0.08 0.05* 0.30 Diameter of nearest log 0.63 0.21 0.23 Distance to nearest rock outcrop 0.03* 0.04* 0.07 Distance to juniper woodland edge 0.00* 0.00* 0.16

*Significant values at 0.05 level.

mouse in 5 sites in both habitats, and the can- significant (univariate significance < 0.25) pre- yon mouse in only 2 of 7 cliff habitat sites. dictors of the occurrence of cliff chipmunks, Habitat data were collected at 294 random plots, pinyon mice, and canyon mice. This was accom- and 113, 19, and 13 centered plots for cliff plished by comparing used sites (capture loca- chipmunks, pinyon mice, and canyon mice, tions) with available habitat (random plots). respectively (Table 1). These analyses indicated that 12 variables were Pearson product moment correlation analy- significant predictors of the occurrence of the sis showed that 4 variables exhibited linear cliff chipmunk, 8 for the pinyon mouse, and 7 correlation: tree canopy cover in low strata, tree for the canyon mouse (Table 2). These vari- canopy cover in middle strata, and tree canopy ables were then used to construct full regres- cover in upper strata were highly correlated sion models for each species. (CC > 0.725) to overall tree canopy cover. Best subsets logistic regression constructed Therefore, the 3 strata of canopy cover were 14 reduced models for the cliff chipmunk, 9 removed and only overall canopy cover was for the pinyon mouse, and 8 for the canyon included in further analysis. Height of the mouse. For the cliff chipmunk there were 3 nearest shrub was also eliminated, which was models within 2.5 AIC units. The lowest-AIC correlated (CC = 0.644) to average shrub model (296.7) was a 4-variable model. The 2nd height. best model included 5 parameters, and the From the remaining variables, univariate 3rd model consisted of 3 parameters. The 1st binary logistic regression analysis was used to model was discarded because the data did not determine which microhabitat variables were fit the logistic regression model (P < 0.05). 90 WESTERN NORTH AMERICAN NATURALIST [Volume 64

TABLE 3. Parameter estimates (β), standard error, and odds ratios for the best multivariate model for distinguishing between cliff chipmunk, pinyon mouse, and canyon mouse capture locations and random sites in the rocky slope and cliff habitat types in southwestern Wyoming in 1999. β β Species sx– ( )Odds ratios Cliff chipmunk % grass cover –0.133 0.017 0.876 % bare ground cover –0.023 0.002 0.977 Average shrub height 0.013 0.060 1.013 Pinyon mouse % canopy cover 0.029 0.014 1.030 % forb cover –0.482 0.167 0.618 Distance to nearest rock outcrop 0.050 0.031 1.051 Distance to juniper woodland edge 0.016 0.004 1.016 Canyon mouse % forb cover 0.362 0.178 1.436 % rock cover 0.076 0.024 1.078 High tree density 1.186 0.415 3.275

Ultimately, the model consisting of 3 parame- distances to rock outcrops and the woodland ters was chosen (Table 3) because it was the edge best distinguished pinyon mouse capture simplest with regard to the fewest parameters, locations from random sites. and the data fit the logistic regression model The model that best distinguished canyon (P > 0.05). Once this model was selected, tests mouse capture locations from random sites were conducted for interaction effects between was a 3-parameter reduced model with an variables included in the reduced model. All AIC value of 80.63 (Table 3). According to the potential 2-way interactions (%Grass cover * AIC, the next best model had a value of 82.32 %Bare ground cover, %Ground cover * Aver- and 4 parameters. While this 2nd model had a age shrub height, and %Bare ground cover * slightly better goodness-of-fit (P = 0.32 com- Average shrub height) were tested, and none pared with 0.29), the predictability between were found to be significant (Ramsey and the 2 was virtually identical (88.5% for the 3- Schafer 1997). The final model indicates that parameter model compared with 88.4% for the cliff chipmunk capture locations consisted of 4-parameter model). It appears that the addi- lower grass cover, lower bare ground cover, and tional variable in the 2nd model does not in- taller shrubs than random sampling plots. crease its predictability over the 1st model. Of the 9 models constructed by best sub- The 3-parameter model was selected as the sets logistic regression for the pinyon mouse, better model for distinguishing between can- 3 combinations of variables had AIC values yon mouse capture locations and random plots. within 1 unit of each other. The model with Tests were conducted for all possible 2-way the lowest AIC value (109.85) was reduced to interactions (%Forb cover * %Rock cover, %Forb a 4-parameter model. The other models con- cover * High tree density, %Rock cover * High sisted of 3 parameters and 5 parameters with tree density) between the variables included AIC values of 110.65 and 110.70, respectively. in the model (Ramsey and Schafer 1997). None Once this model was selected (Table 3), of these interactions proved to be significant we tested for all possible 2-way interactions (P < 0.05). The final model indicates canyon that were considered biologically important mouse capture locations are characterized by (%Canopy cover * %Forb, %Canopy cover * greater forb cover, rock cover, and bare ground Distance to juniper woodland edge, and than random plots in the cliff habitat type. %Forb cover * Distance to juniper woodland edge) between the variables included in the DISCUSSION model (Ramsey and Schafer 1997). None of these interactions were found to be significant The cliff chipmunk, pinyon mouse, and can- (P > 0.05). The final model indicates that greater yon mouse were considered rare in Wyoming canopy cover, lower forb cover, and greater because there were fewer than 5 documented 2004] RARE MAMMALS AT RANGE LIMIT 91 occurrences of each species in the state (Fertig have greater amounts of forb cover and a 1997) prior to our study. Clary (1917) first de- shorter distance to the woodland edge, both of scribed the cliff chipmunk and canyon mouse which the pinyon mouse selected against. in Wyoming as “among the characteristic Thus, neither rocky slopes nor cliffs provide Upper Sonoran mammals of the Green River an optimal combination of microhabitat char- Valley.” The canyon mouse was known to occur acteristics for the pinyon mouse. This is con- only in northeastern Arizona, southeastern sistent with Brown’s (1984) theory that the Utah, and adjacent parts of Colorado and New limit of a species distribution occurs where Mexico (Osgood 1909). In 1929, Svihla and environmental requirements for that species Svihla (1929) collected 1 cliff chipmunk in are not met. Wyoming near the Utah border. At the same During our investigation the canyon mouse time they collected 3 canyon mice near the was found only in the cliff habitat type, which Utah border (Svihla and Svihla 1929, 1931). is consistent with the common description of The earliest published account of the pinyon canyon mouse habitat (Hardy 1945, Hall and mouse in Wyoming appears to be 1942 (Hoff- Hoffmeister 1946, Baker 1968, Clark and Strom- meister 1951). During an expedition through berg 1987, Johnson and Armstrong 1987). Our the area in 1959, Durrant and Dean (1960) results contradict Egoscue’s (1964) idea of the collected 3 cliff chipmunks and 5 pinyon mice importance of vegetation on the distribution of in Utah but found none in Wyoming. canyon mice. On average, canyon mice were Surveys have also been conducted in more captured at sites in the cliff habitat type with recent years. During an intensive sampling higher tree canopy cover and very dense trees, effort in 1979, cliff chipmunks were captured along with higher rock cover. While cliff sites at 1 site located approximately in the center of in the study area consisted of high rock cover, our study area (Belitsky 1981). Canyon mice the rocky slope type had higher canopy cover also were captured north of our study area and tree density. Similar to the results of the (Belitsky 1981). pinyon mouse, this may indicate that available We found that the cliff chipmunk is distrib- cliffs provide only marginal canyon mouse uted throughout the juniper woodland in our habitat. However, because it was found to study area. Of 14 juniper woodland, rocky occur only in the cliff habitat type, and assum- slope, and cliff sites, this species occurs in 13 ing Johnson’s (1986) suggestion regarding inter- of them. The cliff chipmunk is commonly asso- specific effects applies to this part of its range, ciated with cliffs and rocky outcrops in juniper it is likely that a combination of competition woodlands throughout its distribution (Hart and habitat quality is important in determin- 1971, Belitsky 1981). Apparently, it uses cliff ing the canyon mouse distribution in this area. structures for den sites and some foraging, pri- It is also likely that cliffs provide some other marily in the early spring. requirement or that slopes lack the factor. The cliff chipmunk’s diet comprises almost Hedderson (1992) suggested that “periph- exclusively vegetation, primarily forbs and eral populations may include genotypes which grasses. We found grasses and forbs to be are unique for any given species. Protection of more abundant in the rocky slope habitat type. such populations is thus thought to deserve Juniper tree density was also higher in the priority equaling that granted other types of rocky slope habitat type than in cliffs; thus, rare species.” These comparisons may provide foraging may require less effort to harvest insight into which habitat or environmental more juniper berries. factors are influencing the distribution and Our analysis suggests that habitat in south- abundance of these species, particularly as western Wyoming may be suboptimal for the they approach the limit of their range. pinyon mouse. Pinyon mice commonly occurred in sites of higher canopy cover and lower forb ACKNOWLEDGMENTS cover than was available throughout the rocky slope and cliff habitat types. Pinyon mice se- We thank the Wyoming Game and Fish lected against proximity of rock outcrops and Department for financially supporting this pro- avoided the edge of the juniper woodland. ject, the United States Department of the Rocky slope habitat type may be better suited Interior Bureau of Land Management for logis- for the pinyon mouse. However, rocky slopes tical support, R. Olson and N. Stanton for their 92 WESTERN NORTH AMERICAN NATURALIST [Volume 64 support throughout this project and their com- and small mammals. United States Department of ments that dramatically improved this manu- Agriculture, Forest Service, General Technical Report RM-166. script, L. Neasloney for Geographic Information GILBERT, F.F., AND R. ALLWINE. 1991. Small mammal com- System data of the study area, and M. Neigh- munities in the Oregon Cascade Range. Pages 257– bors for providing data regarding historical 267 in L.F. Ruggiero, K.B. Aubry, A.B. Carey, and locations of the species of interest. Finally, we M.H. Huff, technical coordinators, Wildlife and veg- thank D. Pavlacky for his support in the field. etation of unmanaged Douglas-fir forests. United States Forest Service, General Technical Report PNW-GTR-285. LITERATURE CITED HALL, E.R., AND D.F. HOFFMEISTER. 1946. Mammals of Nevada. University of California Press, Berkeley. BAKER, R.H. 1968. Habitats and distribution. Pages 98–126 HARDY, R. 1945. Influence of types of soil upon local dis- in J.A. King, editor, Biology of Peromyscus (Roden- tribution of some mammals in southwestern Utah. tia). American Society of Mammalogists, Provo, UT. Ecological Monographs 15:71–108. BELITSKY, D.W. 1981. Small mammals of the Salt Wells– HART, E.G. 1971. Food preferences of the cliff chipmunk, Pilot Butte Planning Unit. United States Department Eutamias dorsalis, in northern Utah. Great Basin of the Interior, Bureau of Land Management, Rock Naturalist 31:182–188. Springs, WY. HEDDERSON, T.A. 1992. Rarity at range limits: dispersal BROWN, J.H. 1984. On the relationship between abun- capacity and habitat relationships of extraneous moss dance and distribution of species. American Natural- species in a boreal Canadian National Park. Biologi- ist 124:255–279. cal Conservation 59:113–120. BURT, W.H., AND R.P. GROSSENHEIDER. 1980. A field guide HOFFMEISTER, D.F. 1951. A taxonomic and evolutionary to the mammals of North America north of Mexico. study of the pinyon mouse, Peromyscus truei. Illinois Houghton Mifflin Company, New York. Biological Monographs 21:1–104. CLARY, M. 1917. Life zone investigations in Wyoming. HOSMER, D.W., AND S. LEMESHOW. 1989. Applied logistic United States Department of Agriculture, Biological regression. John Wiley and Sons, New York. Survey, North American Fauna 42. JOHNSON, D.W. 1986. Desert buttes: natural experiments CLARK, T.W., AND M.S. STROMBERG. 1987. Mammals of for testing theories of island biogeography. National Wyoming. University of Kansas, Museum of Natural Geographic Research 2:152–166. History, Lawrence. JOHNSON, D.W., AND D.M. ARMSTRONG. 1987. Peromyscus COTTAM, G., AND J.T. CURTIS. 1956. The use of distance crinitus. Mammalian Species 287:1–8. measures in phytosociological sampling. Ecology 37: LUCE, B., AND B. OAKLEAF. 1998. Nongame bird and 451–460. mammal plan, mammalian species of special con- DAUBENMIRE, R. 1959. A canopy-coverage method of veg- cern: a plan for inventories and management on non- etational analysis. Northwest Science 33:43–64. game birds and mammals in Wyoming. Wyoming DUESER, R.D., AND H.H. SHUGART. 1978. Microhabitats in Game and Fish Department, Cheyenne. a forest floor small-mammal fauna. Ecology 59:89–98. OSGOOD, W.H. 1909. Revision of the mice of the American DURRANT, S.D., AND N.K. DEAN. 1960. Mammals of Flam- genus Peromyscus. United States Deptartment of Agri- ing Gorge Reservoir Basin. Pages 209–242 in C.E. culture, Biological Survey, North American Fauna 28. Dibble and C.C. Stouts, editors, Ecological studies RAMSEY, F.L., AND D.W. SCHAFER. 1997. The statistical of the flora and fauna of Flaming Gorge Reservoir. sleuth: a course in methods of data analysis. Duxbury Anthropological Papers, University of Utah, Salt Lake Press, Belmont, CA. City. SVIHLA, R.D., AND A. SVIHLA. 1929. Occurrence of the EGOSCUE, H.J. 1964. Ecological notes and laboratory life golden-breasted canyon mouse in Utah and Wyoming. history of the canyon mouse. Journal of Mammalogy Journal of Mammalogy 12:315. 45:387–396. ______. 1931. Mammals of the Uinta Mountain Region. FERTIG, W. 1997. Wyoming plant and animal species of Journal of Mammalogy 12:256–266. special concern. Wyoming Natural Diversity Data- VAUGHAN, T.A. 1986. Mammalogy. 3rd edition. Harcourt base, Laramie. Brace Jovanovich, Orlando, FL. GIBSON, J.W. 1988. The management of amphibians, reptiles, and small mammals in North America: the Received 2 January 2002 need for an environmental attitude adjustment. Accepted 30 December 2002 Pages 4–10 in Management of amphibians, reptiles, Western North American Naturalist 64(1), ©2004, pp. 93–100

BEAVERS INDIRECTLY ENHANCE THE GROWTH OF RUSSIAN OLIVE AND TAMARISK ALONG EASTERN MONTANA RIVERS

Peter Lesica1 and Scott Miles1

ABSTRACT.—Russian olive and tamarisk are introduced woody plants invading western North American riparian communities. Beavers can play an important role in structuring these communities by removing the dominant cottonwood trees. Our study explored the way in which beavers interact with cottonwood, Russian olive, and tamarisk along 4 rivers on the Great Plains of eastern Montana. We sampled cottonwood stands that supported populations of 1 or both exotic species, recording beaver damage and density in addition to size and age of cottonwood, Russian olive, and tamarisk. In stands where beaver had been present, they felled an average of 80% of cottonwood trees while rarely using Russian olive or tamarisk. Beaver foraging was apparent in nearly 90% of stands within 50 m of the river channel but only 21% of stands farther away, creating a sunny corridor along the river channel that may increase the invasive potential of Russian olive and tamarisk. Growth rates of both Russian olive and tamarisk were substantially higher where beavers had reduced the cottonwood canopy cover. Managers wishing to reintroduce beavers should consider the potential effect on invasive exotic plants.

Key words: exotic invasion, beavers, riparian, cottonwood, Russian olive, tamarisk, Elaeagnus angustifolia, Tamarix, corridors, natural enemies hypothesis.

Exotic species may become invasive in their preferred trees and shrubs (Johnston and native communities because they lack natural Naiman 1990). However, little is known about enemies while the native community domi- the effects of beavers on the growth and dis- nants do not (Harris 1988, Keane and Crawley persal of exotic plants. 2002). Herbivory of plant community domi- Russian olive (Elaeagnus angustifolia) and nants benefits invading plants indirectly by tamarisk (Tamarix spp.) are large shrubs or reducing their competitive superiority (Keane small trees introduced from Eurasia. Both and Crawley 2002). Herbivore-caused distur- species invade wetland and riparian habitats of bances have been implicated in the invasion of western North America (Robinson 1965, Olson several nonnative species (Mack 1981, Braith- and Knopf 1986). Mature Russian olive trees waite et al. 1989, McClaran and Anable 1992). bear numerous clusters of small, edible, berry- North American beavers (Castor canaden- like fruits in late summer. Fruits are consumed sis) are an important force structuring riparian and dispersed by birds and mammals such as systems throughout temperate and boreal North starlings (Sturnus vulgaris) and racoons (Pro- America (Naiman et al. 1988, Donkor and cyon lotor; Kindschy 1998, Lesica and Miles Fryxell 1999). On smaller streams they construct personal observation). In addition, ripe fruits dams and build lodges in the resulting impound- will float for up to 48 hours (Lesica and Miles ments. On larger rivers beavers usually dig unpublished data), allowing dispersal by water. dens in banks near the water line and never Seeds germinate under a wide variety of mois- construct dams. Beavers consume many species ture conditions at different times of the grow- of plants, often preferring willows, aspen, ing season (Shafroth et al. 1995). Tamarisk pro- poplars, and cottonwoods (Salix spp. and Popu- duces large numbers of small, wind-borne seeds lus spp.; Hall 1960, McGinley and Whitham throughout the growing season (Brock 1994). 1985, Johnston and Naiman 1990). They sever Seedlings establish on bare, fresh alluvial stems of trees and shrubs at the base and eat deposits or in other moist, disturbed soil (Strom- the bark and cambial tissue. Beavers can alter berg 1997, Taylor et al. 1999). Both species structure and composition of riparian plant have the potential to greatly alter the composi- communities by reducing the dominance of tion, structure, and functioning of riparian

1Conservation Biology Research, 929 Locust, Missoula, MT 59802.

93 94 WESTERN NORTH AMERICAN NATURALIST [Volume 64 communities in the western U.S. (Everitt 1980, range from 939 m to 662 m. The entire Mon- Bush and Smith 1995, Lesica and Miles 2001a). tana reach of the Bighorn River below Yellow- We previously found that beavers facilitate tail Dam was included in our study. The river replacement of cottonwood by Russian olive flows with relatively few meanders and little along the Marias River in north central Mon- braiding against bluffs on 1 side or the other of tana (Lesica and Miles 1999). Here we expand a wide valley. Surface elevations at Yellowtail our study of exotic woody plants to 3 rivers in Dam and Custer are 916 m and 860 m, respec- southeastern Montana and explore how beavers tively. Our study sites on the Powder River were may be affecting both Russian olive and tama- between the Wyoming border and Powderville. risk invasions. In particular, we seek to deter- The river meanders through a wide valley with mine (1) how commonly beavers use tamarisk surface elevations of 975 m to 870 m. and Russian olive and (2) how beaver use of Field Methods cottonwood affects the performance of these co-occurring exotics. We conducted our study over a 3-year period: Marias River in 1997, Yellowstone River in METHODS 1998, Bighorn and Powder Rivers in 1999. We mapped potential study sites from a canoe. We Study Areas located 34 sites on the Marias and Yellowstone We conducted our study on portions of 4 Rivers with large (≥20 plants ⋅ ha–1) stands of major rivers on the Great Plains of eastern Russian olive and 93 sites on the Bighorn and Montana (Fig. 1). Climate of the region is semi- Powder Rivers with large (≥1 ha with ≥2 dis- arid and continental. Mean annual precipita- tinct size classes) stands of tamarisk. We tion was ca. 32 cm in 1950–1980 with 70% to blindly selected a subset from this preliminary 80% falling in April through September. Mean inventory for study (see Lesica and Miles January minimum and July maximum temper- 2001a, 2001b for more detail on site selection). atures ranged from –15°C to –17°C and 30°C At each site we subjectively located 1 or 2 to 32°C, respectively (NOAA 1982). Natural sample plots representing stands with sapling vegetation of highest riparian terraces is domi- or larger cottonwood (>2.5-cm diameter at nated by silver sagebrush (Artemisia cana), ground level) and representative of distinct western wheatgrass (Agropyron smithii), prairie river terrace habitats supporting Russian olive sandreed (Calamovilfa longifolia), and green or tamarisk: 9 plots on the Bighorn River, 21 on needlegrass (Stipa viridula); however, exten- the Marias River, 8 on the Powder River, and sive areas of upper terrace have been con- 26 on the Yellowstone River. Sample plots were verted for agricultural crops. Terraces closer to 500 m2 and circular or rectangular, depending the river channel support riparian vegetation on the shape of the stands being sampled. dominated by plains cottonwood (Populus del- For each sample plot we estimated mean toides, hereafter referred to as cottonwood), distance from plot center to edge of the near- sandbar willow (Salix exigua), buffaloberry est river channel. Estimates of this variable for (Shepherdia argentea), and hydrophytic grasses 6 plots on the Yellowstone River were inadver- and sedges. Cottonwood forests may be hun- tently lost during fieldwork. In each sample dreds of meters wide in meandering reaches plot we estimated tall-cottonwood (>10 m of the rivers. high) canopy cover with a spherical densiome- We sampled stands on the lower Marias ter at plot center in circular plots or at centers River between Tiber Dam and Loma. The of the 2 halves of rectangular plots. We recorded lower river valley is a few hundred meters to number of cottonwood trees >2.5 cm basal over 1 km wide and frequently bounded by diameter (bd) into 3 size classes: sapling (2.5– steep breaks eroded from soft sedimentary 13 cm bd), pole (13–23 cm bd), and mature formations. Surface elevations range from 861 (>23 cm bd). Russian olive trees >90 cm tall m to 779 m. We sampled stands on the lower were divided into 3 size classes: sapling (<8 Yellowstone River between Billings and Terry. cm bd), pole (8–13 cm bd), and mature (>13 The upper portion of the study reach gener- cm bd). For tamarisk plants we recorded ally has a braided channel, while the lower plants >100 cm tall, measured the tallest stem portion has a single channel confined between with a gauging pole, and counted number of high terraces or low bluffs. Surface elevations live stems. 2004] BEAVERS AND WOODY PLANT INVADERS 95

Fig. 1. Location of study reaches (thickened lines) along 4 rivers on the Great Plains of eastern Montana.

We obtained age estimates for 3 represen- damage by tapered severing with tooth marks tative Russian olive and tamarisk plants in on limbs or boles. dominant size classes and at least 1 plant in Vascular plant nomenclature follows the subordinate classes. In eastern Montana, tama- Great Plains Flora Association (1986). We follow risk has a shrub growth form. New branches Welsh et al. (1987) in referring to our tamarisk and roots arise from older branches that have plants as T. ramosissima. These plants are dif- been buried by sediment (Everitt 1980). We ficult to distinguish from T. chinensis (Brock excavated tamarisk plants and attempted to 1994) and have also been incorrectly called T. obtain a cross section from the point just be- pentandra (Baum 1967). low the union of the lowest stems. However, Data Analysis some tamarisk age estimates may be inaccu- rate because it was not always possible to Mean annual growth rate was estimated as determine exactly the level of initial establish- basal diameter divided by age for Russian olive ment. We obtained age estimates for Russian and as size index divided by age for tamarisk. olive from cross sections or increment cores Tamarisk size index is height of the tallest live taken just above ground level. Many Russian stem multiplied by the number of live stems, olive branch at ground level, and root flare integrating the influence of height and stem occurs just below, so we often took increment number on shrubby tamarisk plants. We believe cores or cross sections from the base of the this is a reasonable index of biomass because largest leader. The largest leader was 1 year linear regression models of stem length effec- younger than the plant in 13 of 16 juveniles tively accounted for variation in weight of where both were measured; thus, we added 1 tamarisk stems (R2 = 0.87, P < 0.001, n = 24). year to the estimated age of Russian olive when Only Russian olive and tamarisk plants with cores were taken from the largest leader. We unambiguous age determinations were used measured basal diameter of sample cottonwood for growth-rate estimates. We used analysis of and Russian olive to the nearest 1 cm with a variance (ANOVA) to assess the effect of a tall- diameter tape. Annual rings were counted using cottonwood canopy on the untransformed values a 10X microscope. of growth rates for Russian olive and tamarisk. In each plot we recorded the number of Growth rates for Russian olive were higher for cottonwood, Russian olive, and tamarisk plants older trees, so we analyzed juvenile (5–10 damaged by beavers. We recognized beaver years old) and mature (>10 years old) trees 96 WESTERN NORTH AMERICAN NATURALIST [Volume 64 separately. Russian olive growth did not vary among rivers (F2,169 = 1.4, P = 0.25), but because it had a tendency to decrease farther from the river (R2 = 0.07, P < 0.05), we used distance to channel as a covariate in the ANOVA models to account for this correlation. Tamarisk growth rates did not vary with distance from the river (R2 = 0.001, P = 0.77) but were different among rivers (F2,117 = 6.6, P = 0.002), so river was included as a factor in the ANOVA model. We used the non-para- metric Spearman’s rank correlation coefficient (ρ) to assess the association between proportion of cottonwood damaged by beaver and plot distance from the river channel, because a large number of plots had either 0% or 100% beaver damage.

RESULTS

Cottonwood occurred with either Russian olive or tamarisk or both in 64 study plots on the 4 rivers. Russian olive did not occur in study plots on the Powder River, and tamarisk did not occur on the Marias River. Density of sapling and larger cottonwood varied from 20 to 2420 plants ⋅ ha–1 with a mean of 511 plants ⋅ –1 ⋅ –1 ha (sx– = 64 plants ha ). Density of Russ- ian olive varied from 20 to 1220 plants ⋅ ha–1 ⋅ –1 with a mean of 153 plants ha (sx– = 28 plants ⋅ ha–1). Russian olive trees varied from 4 to 36 years old with a mean of 11 years. Density of tamarisk varied from 40 to 2200 plants ⋅ ha–1 ⋅ –1 with a mean of 815 plants ha (sx– = 163 plants ⋅ ha–1). Tamarisk plants were 3 to 37 years old with a mean of 17 years. Density did not vary significantly among the rivers for cottonwood (F3,60 = 1.1, P = 0.34), Russian olive (F2,49 = Fig. 2. Basal diameter growth rate of (A) juvenile (5–10 1.9, P = 0.17), or tamarisk (F2,16 = 1.6, P = 0.23). We failed to find a difference in density years) and (B) mature (>10 years) Russian olive, and (C) size (maximum stem height × stem number) growth rate of Russian olive (t = 0.69, P = 0.50) or tama- of tamarisk with and without a cottonwood canopy >10 m risk (t = 0.16, P = 0.87) in open areas com- high. Sample sizes given above the bars. pared with those beneath a cottonwood canopy, probably due to large variation in density among stands. The presence of a cottonwood canopy was tamarisk plants under a cottonwood canopy associated with a lower growth rate of Russian was reduced by 62% compared with those in olive and tamarisk (Fig. 2). Mean annual growth the open (F1,116 = 7.2, P = 0.009). rate (diameter/age) of juvenile Russian olive Beaver damage to cottonwood was common was reduced by 36% under a cottonwood can- and severe. In stands along the Marias, Powder, opy compared with the open (F1,86 = 2.5, P = or Yellowstone Rivers where beaver use was 0.11), and growth rate of mature plants was recorded, beavers damaged more than 80% of reduced by 41% under cottonwood (F1,56 = cottonwood trees on average (Fig. 4). Beaver 21.1, P < 0.001). Mean annual growth rate of damage to cottonwood was less severe along 2004] BEAVERS AND WOODY PLANT INVADERS 97

Fig. 3. Percent of study plots with beaver-damaged cottonwood at different distances from river channels. Sam- ± ple sizes given inside bars. Fig. 4. Mean proportion ( sx–) of cottonwood, Russian olive, and tamarisk plants damaged by beaver in study plots where they had been present. Sample sizes given above bars. the Bighorn River. Overall beaver damage was present in nearly 90% of study plots within 50 m of river channels, but only 21% of plots farther away from channels showed damage (Fig. 3; may be the case with Russian olive and tama- χ2 = 8.54, P = 0.003). There was strong nega- risk in North America. Cottonwood is poten- tive correlation between plot distance from the tially a stronger competitor than either Russ- river and proportion of cottonwood trees dam- ian olive or tamarisk because it grows taller aged by beaver (ρ = 0.66, P < 0.001). (Sher et al. 2000, Lesica and Miles 2001a, Beaver had little or no direct impact on 2001b), out-competing them for light, presum- Russian olive and tamarisk. There were no ably the limiting resource in these mesic envi- beaver-damaged tamarisk in any of the 10 ronments. Cottonwood and other trees in the plots where beaver use was recorded on the genus Populus are highly preferred food of Bighorn, Powder, and Yellowstone Rivers (Fig. beavers (Hall 1960, McGinley and Whitham 4). Tamarisk did not occur in Marias River 1985, Johnston and Naiman 1990). Presence of study plots. Russian olive was not damaged by beavers strongly alters the competitive hierar- beavers in Bighorn and Yellowstone River study chy between cottonwood and the introduced plots (Fig. 4). A mean of 15% of Russian olive species. At our study sites with evidence of showed beaver damage compared to 89% for beaver damage, an average of 80% of cotton- cottonwood in plots where both occurred on woods had been felled. In contrast, Russian the Marias River (paired-sample t = 8.81, P < 0.001). Most beaver-damaged cottonwoods olives were rarely damaged by beavers, and were cut off at the base, while damage to Russ- we never observed beaver-damaged tamarisk. ian olive was usually confined to 1 or 2 basal Beavers create areas of lower competitive limbs. stress by felling the dominant cottonwoods. Most beaver damage to cottonwood occurred DISCUSSION within 50 m of river channels, and most cottonwood stands within this distance had been Exotics may become invasive because they damaged. Beavers forage near river banks for lack a full complement of pests, predators, and 2 possible reasons: they are more vulnerable pathogens that negatively affect native species to coyotes and other predators farther from the (Harris 1988, Keane and Crawley 2002). This safety of water, and transporting food material 98 WESTERN NORTH AMERICAN NATURALIST [Volume 64 greater distances requires more energy (Jenk- (Salix; Currier 1982, Lesica and Miles 2001a). ins 1980, Belovsky 1984). Beaver activity helps Decline of these dominant native riparian create and maintain corridors of open, sunny woody plants could cause loss of habitat for habitat along river channels that would other- species such as cavity-nesting and insectivo- wise be dominated by cottonwood. rous birds (Knopf and Olson 1984, Olson and Absence of a cottonwood canopy in low-ter- Knopf 1986). Tamarisk infestations may alter race, streamside habitats allowed higher growth riparian function in a number of ways. Dense rates for both Russian olive and tamarisk. stands of tamarisk are reported to transpire Although a tall-cottonwood canopy signifi- large quantities of water, perhaps leading to cantly affects growth of the exotic species, it lowered stream flows (Brotherson and Field may not always reduce recruitment (Lesica 1987). Tamarisk invasion may have a detrimen- and Miles 2001a) or density. Nonetheless, plants tal effect on small mammal populations (Ellis of both introduced species under a cotton- et al. 1997). In addition, tamarisk stands have wood canopy are smaller and presumably pro- altered avian communities, with insectivores duce fewer fruits than plants of the same age and frugivores being more common in native growing in full sun. Slower growth and associ- vegetation (Cohan et al. 1978). ated reduced fecundity could retard population Although both species of beavers have been growth rates and invasion potential (William- extirpated from much of their original range in son 1989, Rejmanek 1996). historic times, this trend is now being reversed, Rivers act as dispersal corridors for many as they are reintroduced in parts of North plants (Malanson 1993, Jansson et al. 2000), America (Albert and Trimble 2000) and north- including exotics (Thebaud and Debussche ern Europe (Castor fiber; Danilov and Kan’shiev 1991, de Waal et al. 1994, Parendes and Jones 1983, Hartman 1994, Macdonald et al. 2000). 2000). Non-forested corridors created by In general, these reintroductions enhance nat- beavers along rivers may facilitate dispersal of ural functioning of riparian areas (Naiman et Russian olive and tamarisk. Russian olive fruits al. 1988, Naiman and Rogers 1997) and may have spongy flesh surrounding the large seed. even help control exotic plants (Albert and They float for at least 48 hours once the seed Trimble 2000). Many native plant species benefit from streamside beaver activity (Naiman is mature and the flesh has dried (Lesica and et al. 1988). For example, herbaceous species, Miles unpublished data). Russian olive fruits especially grasses and sedges, are more abun- are readily eaten by racoons that use the ripar- dant in the zone of reduced shade created by ian zone as foraging corridors (Jones et al. 1983) beaver (Lesica and Miles unpublished data). and disperse the seeds in their feces (Lesica We certainly do not recommend removing bea- and Miles personal observation). Unfortunately, vers from natural systems. However, if tamarisk, viability of these seeds is not known. Frugivo- Russian olive, or other invasive riparian exotics rous birds also disperse Russian olive seeds occur in drainages, the presence of beavers (Olson 1974, Kindschy 1998), and many of these may exacerbate the problem and make control species preferentially use riparian corridors more difficult. It may be best to first control for foraging (Finch and Ruggiero 1993). By undesirable plants before proceeding with removing cottonwoods near the river, beavers beaver reintroduction. could promote earlier and greater Russian olive fruiting and allow more fruits to reach ACKNOWLEDGMENTS the water or be moved along the riparian corridor by dispersing animals. Beavers may also We thank Joe Frazier and Janet Henderson enhance the dispersal of tamarisk wind-borne for help in the field. Rick Blaskovich, Larry Rau, seeds along the river corridor by removing Jody Peters and other employees of the Bur- trees and widening the zone of increased wind eau of Land Management (BLM), the Bureau turbulence. of Reclamation, and the Crow Tribe helped Continued spread of both Russian olive and with logistical support. We thank numerous tamarisk is considered a threat to the integrity private landowners and public land lessees for of native communities. Researchers have spec- access to their land. Funding was provided by ulated that Russian olive can hinder recruitment the Montana State Office of the BLM, the of native cottonwoods (Populus) and willows Montana Department of Fish, Wildlife and 2004] BEAVERS AND WOODY PLANT INVADERS 99

Parks, and the U.S. Bureau of Reclamation. HARTMAN, G. 1994. Long-term population development of Dean Pearson and Calvin Porter made several a reintroduced beaver (Castor fiber) population in Sweden. Conservation Biology 8:713–717. helpful comments on the manuscript. JANSSON, R., C. NILSSON, AND B. RENOFALT. 2000. Frag- mentation of riparian floras in rivers with multiple LITERATURE CITED dams. Ecology 81:899–903. JENKINS, S.H. 1980. A size-distance relation in food selec- ALBERT, S., AND T. T RIMBLE. 2000. Beavers are partners in tion by beavers. Ecology 61:740–746. riparian restoration on the Zuni Indian Reservation. JOHNSTON, C.A., AND R.J. NAIMAN. 1990. Browse selection Ecological Restoration 18:87–92. by beaver: effects on riparian forest composition. BAUM, B.R. 1967. Introduced and naturalized tamarisks in Canadian Journal of Forest Research 20:1036–1043. the United States and Canada (Tamaricaceae). Bai- JONES, J.K., D.M. ARMSTRONG, R.S. HOFFMAN, AND C. leya 15:19–25. JONES. 1983. Mammals of the northern Great Plains. BELOVSKY, G.E. 1984. Summer diet optimization by University of Nebraska Press, Lincoln. beaver. American Midland Naturalist 111:209–221. KEANE, R.M., AND M.J. CRAWLEY. 2002. Exotic plant inva- BRAITHWAITE, R.W., W.M. LONSDALE, AND J.A. ESTBERGS. sions and the enemy release hypothesis. Trends in 1989. Alien vegetation and native biota in tropical Ecology and Evolution 17:164–170. Australia: the impact of Mimosa pigra. Biological KINDSCHY, R.R. 1998. European starlings disseminate viable Conservation 48:189–210. Russian olive seeds. Northwestern Naturalist 79: BROCK,J.H. 1994. Tamarix spp. (salt cedar), an invasive 119–120. exotic woody plant in arid and semi-arid riparian KNOPF, F.L., AND T.E. OLSON. 1984. Naturalization of habitats of western USA. Pages 27–44 in L.C. de Waal Russian olive: implications to Rocky Mountain wild- et al., editors, Ecology and management of invasive life. Wildlife Society Bulletin 12:289–298. riverside plants. John Wiley and Sons, New York. LESICA, P., AND S. MILES. 1999. Russian olive invasion BROTHERSON, J.D., AND D. FIELD. 1987. Tamarix: impacts into cottonwood forests along a regulated river in of a successful weed. Rangelands 9:110–112. north-central Montana. Canadian Journal of Botany BUSH, D.E., AND S.D. SMITH. 1995. Mechanisms associ- 77:1077–1083. ated with decline of woody species in riparian eco- ______. 2001a. Natural history and invasion of Russian systems of the southwestern U.S. Ecological Mono- olive along eastern Montana rivers. Western North graphs 65:347–370. American Naturalist 61:1–10. COHAN, D.R., W. ANDERSON, AND R.D. OHMART. 1978. ______. 2001b. Growth limits on tamarisk at the northern Avian population responses to saltcedar along the margin of its range in Montana. Wetlands 21:240–246. lower Colorado River. USDA Forest Service, Gen- MACDONALD, D.W., F.H. TATTERSALL, S. RUSHTON, A.B. eral Technical Report WO-12, Washington, DC. SOUTH, S. RAO, P. MAITLAND, AND R. STRACHEN. 2000. CURRIER, P.J. 1982. The floodplain vegetation of the Platte Reintroducing the beaver to Scotland: a protocol for River: phytosociology, forest development, and seed- identifying and assessing suitable release sites. Ani- ling establishment. Doctoral dissertation, Iowa State mal Conservation 3:125–133. University, Ames. MACK, R.N. 1981. Invasion of Bromus tectorum L. into west- DANILOV, P.I., AND V. Y. K AN’SHIEV. 1983. The state of populations and ecological characteristics of European ern North America: an ecological chronicle. Agro- (Castor fiber L.) and Canadian (Castor canadensis ecosystems 7:145–165. Kuhl.) beavers in northwestern USSR. Acta Zoolog- MALANSON, G.P. 1993. Riparian landscapes. Cambridge ica Fennica 174:95–97. University Press, Cambridge. MCCLARAN, M.P., AND M.E. ANABLE. 1992. Spread of intro- DE WAAL, L.C., L.E. CHILD, P.M. WADE, AND J.H. BROCK, EDITORS. 1994. Ecology and management of invasive duced Lehmann lovegrass along a grazing intensity riverside plants. John Wiley and Sons, New York. gradient. Journal of Applied Ecology 29:92–98. DONKOR, N.T., AND J.M. FRYXELL. 1999. Impact of beaver MCGINLEY, M.A., AND T.G. W HITHAM. 1985. Central place foraging on structure of lowland boreal forests of foraging by beavers (Castor canadensis): a test of for- Algonquin Provincial Park, Ontario. Forest Ecology aging predictions and the impact of selective feeding and Management 118:83–92. on the growth form of cottonwoods (Populus fremon- ELLIS, L.M., C.S. CRAWFORD, AND M.C. MOLLES. 1997. tii). Oecologia 66:558–562. Rodent communities in native and exotic riparian NAIMAN, R.J., AND K.H. ROGERS. 1997. Large animals and vegetation in the middle Rio Grande valley of cen- system-level characteristics in river corridors. Bio- tral New Mexico. Southwestern Naturalist 42:13–19. Science 47:521–529. EVERITT, B.L. 1980. Ecology of saltcedar—a plea for re- NAIMAN, R.J., C.A. JOHNSON, AND J.C. KELLEY. 1988. Alter- search. Environmental Geology 3:77–84. ation of North American streams by beaver. Bio- FINCH, D.M., AND L.F. RUGGIERO. 1993. Wildlife habitats Science 38:753–762. and biological diversity in the Rocky Mountains and NATIONAL OCEANIC AND ATMOSPHERIC ASSOCIATION northern Great Plains. Natural Areas Journal 13: (NOAA). 1982. Monthly normals of temperature, 191–203. precipitation and heating and cooling degree days. GREAT PLAINS FLORA ASSOCIATION. 1986. Flora of the Montana 1951–1980. National Climate Center, Great Plains. University Press of Kansas, Lawrence. Asheville, NC. HALL, J.G. 1960. Willow and aspen in the ecology of beaver OLSON, D.F.1974. Elaeagnus. Pages 376–379 in C.E. Schop- on Sagehen Creek, California. Ecology 41:484–494. meyer, editor, Seeds of woody plants in the United HARRIS, P. 1988. Environmental impacts of weed-control States. USDA Forest Service Agriculture Handbook insects. BioScience 38:542–548. 450, Washington DC. 100 WESTERN NORTH AMERICAN NATURALIST [Volume 64

OLSON, T.E., AND F.L. KNOPF. 1986. Naturalization of STROMBERG, J.C. 1997. Growth and survivorship of Fre- Russian olive in the western United States. Western mont cottonwood, Gooding willow and salt cedar Journal of Applied Forestry 1:65–69. seedlings after large floods in central Arizona. Great PARENDES, L.A., AND J.A. JONES. 2000. Role of light avail- Basin Naturalist 57:198–208. ability and dispersal in exotic plant invasion along TAYLOR, J.P., D.B. WEBSTER, AND L.M. SMITH. 1999. Soil roads and streams in the H.J. Andrews Experimental disturbance, flood management, and riparian woody Forest, Oregon. Conservation Biology 14:64–75. plant establishment in the Rio Grande floodplain. REJMANEK, M. 1996. A theory of seed plant invasiveness: Wetlands 19:372–382. the first sketch. Biological Conservation 78:171–181. THEBAUD, C., AND M. DEBUSSCHE. 1991. Rapid invasion ROBINSON, T.W. 1965. Introduction, spread and areal of Fraxinus ornus L. along the Herault River system extent of saltcedar (Tamarix) in the western states. in southern France: the importance of seed dispersal U.S. Geological Survey Professional Paper 491-A. by water. Journal of Biogeography 18:7–12. SHAFROTH, P.B., G.T. AUBLE, AND M.L. SCOTT. 1995. Ger- WELSH, S.L., N.D. ATWOOD, S. GOODRICH, AND L.C. HIG- mination and establishment of the native plains cot- GINS. 1987. A Utah flora. Great Basin Naturalist Mem- tonwood (Populus deltoides Marshall subsp. monilif- oirs 9:1–894. era) and the exotic Russian-olive (Elaeagnus angusti- WILLIAMSON, M. 1989. Mathematical models of invasion. folia L.). Conservation Biology 9:1169–1175. Pages 329–350 in J.A. Drake et al., editors, Biological SHER, A.A., D.L. MARSHALL, AND S.A. GILBERT. 2000. invasions: a global perspective. John Wiley and Sons, Competition between native Populus deltoides and Chichester. invasive Tamarix ramosissima and the implications for reestablishing flooding disturbance. Conserva- Received 13 May 2002 tion Biology 14:1744–1754. Accepted 23 January 2003 Western North American Naturalist 64(1), ©2004, pp. 101–108

CONTRIBUTION TO THE MORPHOLOGY AND DESCRIPTIVE BIOLOGY OF CAURINELLA IDAHOENSIS (EPHEMEROPTERA: EPHEMERELLIDAE)

Luke M. Jacobus1 and W.P. McCafferty1

ABSTRACT.—Reared specimens from Bridge Creek, Idaho County, Idaho, provide the bases for the first descriptions of Caurinella idahoensis egg and alate stages and redescription of the larval stage. Larvae are distinguished from other Nearctic Ephemerellinae species by the distinctive posterolateral projections on abdominal segment 9. Male adults have a unique combination of characters associated with their genitalia. Larvae were associated with the colonial blue- green alga Nostoc parmelioides in a clear, cold headwater stream. Several other Diptera, Ephemeroptera, Plecoptera, and Trichoptera also were found in cohabitation with C. idahoensis. Amorphous detritus appears to be a major component of the diet of C. idahoensis, and larvae may defend small territories on rock surfaces. Larvae exhibited prolonged preemergence behavior in the laboratory rearing apparatus.

Key words: Caurinella idahoensis, Ephemeroptera, Ephemerellidae, stage descriptions, Nostoc parmelioides, behavior, Bitterroot Mountain Range.

Caurinella Allen is a monospecific genus in the Bitterroot Mountain Range to find, study, (Allen 1984) of the subfamily Ephemerellinae and rear C. idahoensis. of the Ephemerellidae (McCafferty and Wang 2000). Caurinella idahoensis Allen has been METHODS known from the larval stage only, and it has been reported from only 2 locales in Idaho On 29 July 2002 we located a population of (Bushy Creek, tributary of Lochsa River, Idaho late instar larvae in Bridge Creek, Idaho County, County [Allen 1984], and Eggers Creek, tribu- Idaho, at 1708 m above sea level. We placed a tary of Silver Creek, Valley County [Edmunds series of these larvae in resealable food con- and Murvosh 1995]) and 1 locale in far west- tainer cups with water and stones from their habitat. Aeration was provided by portable ern Montana, near the Idaho border (McCaf- aquarium aerators, and each cup was sus- ferty 2001a). Recently, additional C. idahoensis pended above loose ice in a large, insulated larvae were collected from central Idaho (G.T. cooler. Temperature in the cooler varied from Lester, Moscow, ID, personal communication). 7°C to 13°C during the 3 days of our return to Barbour et al. (1999) inferred that C. idahoen- Indiana from Idaho. Once in the laboratory, sis larvae are primary collector-gatherers and we opened the cooler during the day in an air- that the species is extremely sensitive to pollu- conditioned room (20°–23°C, 46%–53% relation, being found only in minimally disturbed tive humidity) and exposed its contents to streams of high water quality (for discussion of ambient light from a south-facing window. Ice biological indicators and water quality para- changes continued twice a day, and we added meters, see Hilsenhoff 1982, 1998, Cummins small amounts of bottled mineral water to and Merritt 1996). Edmunds and Waltz (1996) each cup each day to maintain water levels. included Caurinella in their larval key to North Subimagos were removed from the rearing American Ephemeroptera genera. apparatus and placed in small plastic cups that As part of our studies of poorly known North contained crumpled tissue paper. We fixed American Ephemeroptera (e.g., McCafferty adults in a 70% ethanol solution about 12 hours 2001a, 2001b, Jacobus and McCafferty 2002) after emergence from the subimago. and our revisionary studies of Ephemerelli- At the study site we measured water tem- nae, we traveled to the Lochsa River drainage perature and photographed the larval habitat.

1Department of Entomology, Purdue University, West Lafayette, IN 47907.

101 102 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Samples of stone substrate from where the General dorsal color light gray to dark khaki, study organisms were found were collected for with varied gray to dark brown markings; pale further study, including the identification of ventrally, without markings. Head: brown spot associated epilithic periphyton. Any macroin- between compound eye and antennal base; vertebrates collected with the study organism other variable brown markings present; vertex were fixed in a 90% ethanol solution in the smooth with scattered, hairlike setae. Anten- field and later transferred to a 70% ethanol nal scape khaki, pedicel brown, flagellar seg- solution for storage. A few specimens of the ments brown. Genae and frontal ridge slightly study organism were fixed in ethanol at the produced to form rounded shelf, with numer- study site for later examination and study. ous, long, hairlike setae. Clypeus produced Using stereo and compound light microscopy, slightly. Labrum dark brown; margins with we studied in detail the exoskeletal morpholo- dense fringe of short, fine, fimbriate setae. gies of mature larvae, larval exuviae, reared Maxillary palpi reduced, 3-segmented, with male and female adults, and subimagos. A few apical segment very small. Thorax: brown, specimens were dissected for microanatomical microscopic excrescences present on dorsal analysis and extraction of eggs. We examined surface; no prominent dorsal tubercles present. and documented eggs from a female adult us- Pronotum pale, with smoky tinge at anterome- ing scanning electron microscopy and described dial margin, 2 pairs prominent dark brown them following the terminology of Studemann medial spots, and additional variable brown and Landolt (1997). Foregut contents were ana- markings; lateral margins and posterior margin lyzed following the methodology of McShaffrey dark brown. Mesothorax with variable brown (1988). markings; posterior tips of forewing pads dark brown. Legs covered with long, hairlike setae. MORPHOLOGY AND TAXONOMY Femora brown with pale foremargin and pale medial spot; scattered, short, stout, branched Descriptions setae present on dorsal surface; hind margin EGG (FIG. 1).—One polar cap. Live color with long, stout setae in slightly elevated sock- yellow; alcohol specimens purple with tan cap. ets. Tibiae khaki; inner margin with long, stout Chorion with geometric macrorelief of juxta- setae. Tarsi brown; claws with single row of posed hexagonal structures. 3–5 denticles. Abdomen: brown, with large, MATURE LARVA (FIG. 2).—Length, in milli- paired, pale spots on terga 1–7; terga 8–9 with meters: body 6.8–7.3; caudal filaments 3.0–4.2. large, pale medial areas; small, sublateral pale

Fig. 1. Scanning electron micrograph of Caurinella idahoensis egg (scale bar = 20 µm). 2004] CAURINELLA MORPHOLOGY AND BIOLOGY 103

Fig. 2. Caurinella idahoensis larval habitus, dorsal view. 104 WESTERN NORTH AMERICAN NATURALIST [Volume 64 spots on terga 1–9; terga pale laterally; tergum clouded with white. All legs pale, each tro- 10 mostly brown with anteromedial pale spot; chanter with white spot. Length of segments stout setae and hairlike setae present on pos- of foreleg, in millimeters: trochanter = 0.3, terior margins of terga; scattered spicules pre- femur = 1.8, tibia = 3.7, tarsus I = 0.1, tarsus sent on terga sublaterally; pair of posterior, II = 1.0, tarsus III = 0.9, tarsus IV = 0.9, tar- transverse, rounded ridges with dorsally pro- sus V = 0.4, claws = 0.1. Abdomen: ivory white, jecting, long, hairlike setae present on terga 8 with segments 1–7 translucent, segments 8–10 and 9, more prominent on tergum 8; postero- opaque; dorsal median areas opaque white on lateral projections present on segments 3 or segments 2–10, except for thin, translucent, 4–9; projections pale, with row of short, robust median stripe; ventral median areas opaque setae laterally and scattered, hairlike setae; white on segments 6–10; hind margins of terga projections progressively larger posteriorly; and sterna opaque white; segments 6–10 with segment 8 projections slightly upturned later- at least part of pleural margins opaque white. ally and distally; segment 9 projections promi- Genitalia as in Figure 4; penes with shallow nent, at least subequal to length of segment at apical notch; forceps segment 2 somewhat midline, upturned at apices. Gill lamellae on dorsoventrally compressed; forceps segment 3 segments 3–7; lamellae 3–6 with median, mit- ellipsoidal. Caudal filaments white, relatively ten-shaped, brown region and pale central spot; densely covered with short, fine setae. lamellae 7 reduced, brown. Lower fimbriate FEMALE ADULT.—Length, in millimeters: portion of gills lamelliform; bifurcate on gills body 8.1, forewings 9.8, caudal filaments 9.5, 3–5, integral on gills 6 and 7. Caudal filaments forelegs 6.0. Thorax coloration sometimes much brown, with curved, hairlike setae at apex of lighter than in male. Wings similar to those in each segment and sparse, fine, intrasegmental male, except longitudinal veins of forewing setae. variably olive brown. Abdominal segments MALE ADULT.—Length, in millimeters: body similar to those of male, except with faint, gray 8.5, forewings 9.0, caudal filaments 11.2. Head: tracheation marks. Subanal plate gently round- tan. Antennae tan. Ocelli white with reddish ed, with very shallow apical notch (Fig. 5). brown base. Upper portions of dioptic com- SUBIMAGOS.—Coloration similar to that of pound eyes nearly contiguous. Thorax: drab adults, except with dusky, pale blue-gray wings olive green in life, light tan in alcohol; protho- and dull body coloration. rax with irregular gray markings and strong Diagnoses median keel; meso- and metathorax (Fig. 3) with sparse, irregular yellow and white mark- Larvae may be distinguished from other ings. Wings hyaline, with all veins, intercalar- Nearctic Ephemerellinae species by the distinc- ies, and crossveins pale; stigmatic area lightly tive posterolateral projections on abdominal

3 4 5

Figs. 3–5. Caurinella idahoensis: 3, male adult meso- and metanotum, dorsal view; 4, male genitalia, ventral view; 5, female adult posterior abdominal segments, ventral view. 2004] CAURINELLA MORPHOLOGY AND BIOLOGY 105 segment 9 (Fig. 2). Male adults may be distin- ing (Nostocales: Nostocaceae; Fig. 6), which guished from other Nearctic Ephemerellinae occurred on cobble and rock surfaces exposed species by the following combination of char- to the stream current. We collected only 1 acters: penes shaped as in Figure 4 and lack specimen of C. idahoensis that was not associ- spines, genital forceps segment 3 ellipsoidal, ated with N. parmelioides, and it may have and genital forceps segment 2 somewhat dorso- been displaced by our activity in the stream. ventrally compressed and not bowed, twisted, Association of other ephemerellid species with or constricted (Fig. 4). stream macroalgae has been documented previously (Percival and Whitehead 1929, Jones Remarks 1950, Hynes 1961, McShaffrey and McCaf- Allen (1984) indicated that the maxillary ferty 1991). Just as filamentous algae were palpi of C. idahoensis are 2-segmented. Mature found to function as a filter that makes fine and immature specimens we examined have detritus available to Ephemerella needhami 3-segmented maxillary palpi. However, imma- McDunnough (McShaffrey and McCafferty ture specimens have maxillary palpi with a very 1990), the dense fields of auriform macro- small apical segment that could be overlooked colonies of N. parmelioides may provide a sim- easily. ilar benefit to C. idahoensis. Nostoc parmelioides is a relatively well- Material Examined known and widespread epilithic colonial blue- IDAHO: Idaho Co., Bridge Cr. at Hoodoo green alga (Sheath and Cole 1992, Dodds et Lake Rd. (FR360), 46°21′53″N, 114°38′11″W al. 1995, Potts 2000) that “forms small, olive- (WGS84), 1708 m elev., 29-VII-2002, W.P. green, shelving thalli” on exposed rock surfaces, McCafferty, L.M. Jacobus, 3 male adults, 2 especially in headwater streams (Prescott female adults, 1 male subimago, 2 female sub- 1970). Nearly all N. parmelioides colonies that imagos, associated exuviae (alates emerged 9- we found in Bridge Creek were the auriform VIII through 17-VIII), 4 larvae (Purdue Uni- macrocolonies described by Brock (1960), versity Entomological Research Collection, which are known to take on such form when West Lafayette, IN, USA [PERC]); same data, inhabited by a host-specific Cricotopus Wulp 1 male adult, 1 set larval exuviae (C.P. Gillette larva (Diptera: Chironomidae) (Wirth 1957, Museum of Arthropod Diversity, Colorado McCafferty 1981). For additional discussion of State University, Fort Collins, CO, USA). the Nostoc-Cricotopus association, see Ward et MONTANA: Missoula Co., 0.6 mi below al. (1985), Kleinhaus and Keiser (1988), Dodds Lolo Pass, 6-VI-1994, D.L. Gustafson, 2 larvae (1989), Dodds and Marra (1989), and Sabater (PERC). and Muñoz (2000). Ward et al. (1985) observed that auriform N. parmelioides macrocolonies HABITAT OBSERVATIONS are found most typically in open-canopied 1st- or 2nd-order streams that are subject to a Bridge Creek is a perennial 2nd-order stream high-intensity light regime. This is consistent at our study site in the Clearwater National with our study site at Bridge Creek. Forest, Idaho County, Idaho. The stream, 15–50 Other macroinvertebrates found on larger cm deep and 4–5 m wide at our study site, has rocks having patches of auriform N. parme- clear, cold (10°C) water and a moderately swift lioides macrocolonies and C. idahoensis indi- current. The stream substrate is composed viduals included Diptera larvae of Micropsec- mostly of pale, rough cobble and rocks, 8–30 tra sp. (Chironomidae); Ephemeroptera larvae cm in diameter, and some bedrock. Direct sun- of Ameletus similior McDunnough (Ameleti- light shone on the stream reach we studied. dae), Baetis bicaudatus Dodds (Baetidae), Our use of common benthic sampling de- Cinygmula sp. (Heptageniidae), and Paralep- vices, such as a D-frame dipnet and a kick- tophlebia sp. A (Leptophlebiidae); Plecoptera screen, yielded only 2 larvae of C. idahoensis. larvae of Megarcys sp. (Perlodidae), Sweltsa However, when we handpicked and examined sp. (Chloroperlidae), Visoka cataractae (Neave) the cobble and rocks from stream runs, we (Nemouridae), Yoraperla sp. (Peltoperlidae), readily located C. idahoensis larvae at the bases and Zapada sp. (Nemouridae); and Trichop- of small, earlike or auriform macrocolonies of tera larvae of Arctopsyche grandis (Banks) the blue-green alga Nostoc parmelioides Kütz- (Hydropsychidae), Rhyacophila arcopedes Banks 106 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Fig. 6. Nostoc parmelioides, auriform macrocolonies.

(Rhyacophilidae), R. tucula Ross, and R. vac- available for dissection and foregut analysis. cua Milne. Most of these taxa have been asso- However, we were able to examine the foregut ciated with cold, high-gradient, 1st- or 2nd- contents of 1 larva fixed with alcohol at the order streams (Smith 1968, Morihara and study site. Amorphous detritus was the largest McCafferty 1979, McCafferty 1981, Stewart component of the foregut contents, and this is and Stark 1988, Stark and Nelson 1994, Wig- usually suggestive of a highly specialized gins 1996, Zloty and Pritchard 1997). Cooper feeder (Hawkins 1985). Fragments of 1 chi- et al. (1986) and Dudley et al. (1986) previ- ronomid midge larva also were found in the ously described the invertebrate fauna of a foregut contents examined. It is probable that southern California stream that contained C. idahoensis is merely an opportunistic carni- Nostoc. vore, such as has been observed for E. needhami (McShaffrey and McCafferty 1990). The BEHAVIOR OBSERVATIONS structure of C. idahoensis mouthparts is typical of species in the mayfly superfamily Ephe- Because our efforts at the study site were merelloidea (McCafferty and Benstead 2002) focused on collecting specimens for rearing, and also appears well suited for scraping and we had very few field-collected specimens biting or shredding. 2004] CAURINELLA MORPHOLOGY AND BIOLOGY 107

At the study site we observed that small dering illustrations and discussing rearing cobble (up to 9 cm in diameter) seldom con- techniques. Chris Oseto and Al York (West tained more than a single larva. Our labora- Lafayette, IN) discussed rearing techniques tory observations revealed that larvae of C. and provided some materials for the rearing idahoensis possibly defend small territories on apparatus. Lu Sun (West Lafayette, IN) assisted rock surfaces. Individual larvae occupying the with scanning electron microscopy and imag- same rock in the rearing apparatus would ing, which was conducted at the Life Science approach each other until physical contact Microscopy Facility, Purdue University. This with the head and forelegs was made. The 2 study was funded in part by USEPA fellow- remained pressed together, head to head, mov- ship 91601701-0 to LMJ and NSF grant DEB- ing slightly, until 1 individual backed away. 9901577 to WPM. Such interaction usually lasted approximately 15 seconds. Further investigation of this be- LITERATURE CITED havior is needed. Subimagos emerged 11–19 days after we ALLEN, R.K. 1984. A new classification of the subfamily Ephemerellinae and the description of a new genus. collected the larvae from Bridge Creek. Most Pan-Pacific Entomologist 60:245–247. subimagos emerged during afternoon daylight BARBOUR, M.T., J. GERRITSEN, B.D. SNYDER, AND J.B. hours. Emergence behavior was observed in STRIBLING. 1999. Rapid bioassessment protocols for the rearing apparatus as follows. Final instar use in wadeable streams and rivers: periphyton, benthic macroinvertebrates, and fish. 2nd edition. EPA larvae floated to the water surface, at which 841-B-99-002, United States Environmental Protec- point the thoracic terga split along the ecdy- tion Agency, Office of Water, Washington, DC. sial line. Larvae moved to the edge of the rear- BROCK, E.M. 1960. Mutualism between the midge Crico- ing cup and grasped a partially submerged topus and the alga Nostoc. Ecology 41:474–483. strip of paper that we placed along the inside COOPER, S.D., T.L. DUDLEY, AND N. HEMPHILL. 1986. The biology of chaparral streams in southern California. of the cup. Larvae remained on the paper strip Pages 139–152 in J. DeVries, editor, Proceedings of for a few minutes just under the water surface the chaparral ecosystem research conference. Report and then climbed up the paper so that the 62, California Water Resources Center, Davis, CA. head and thorax were exposed to the air and CUMMINS, K.W., AND R.W. MERRITT. 1996. Ecology and distribution of aquatic insects. Pages 74–86 in R.W. the abdomen was in the water. A film of water Merritt and K.W. Cummins, editors, An introduction remained between the exposed body and the to the aquatic insects of North America. 3rd edition. paper. The larvae remained in this position for Kendall-Hunt, Dubuque, IA. a few more minutes, and the gills continued DODDS,W.K. 1989. Photosynthesis of two morphologies beating in the water until immediately before of Nostoc parmelioides (Cyanobacteria) as related to current velocities and diffusion patterns. Journal of emergence of the subimago. The subimagos Phycology 25:258–262. then emerged and perched on the exposed DODDS, W.K., AND J.L. MARRA. 1989. Behaviors of the portion of the larval exuviae. After less than 1 midge, Cricotopus (Diptera: Chironomidae) related minute, the subimago wings unfurled com- to mutualism with Nostoc parmelioides (Cyanobacte- pletely and the subimagos flew away from the ria). Aquatic Insects 11:201–208. DODDS, W.K., D.A. GUDDER, AND D. MOLLENHAUER. 1995. larval exuviae. Elapsed time from when the The ecology of Nostoc. Journal of Phycology 31:2–18. larvae floated until the subimagos flew from DUDLEY, S.D., T.L. COOPER, AND N. HEMPHILL. 1986. the rearing cup was approximately 40 min- Effects of macroalgae on a stream invertebrate com- utes. Subimagos molted to the adult stage munity. Journal of the North American Benthologi- cal Society 5:93–106. approximately 30 hours after emergence from EDMUNDS, G.F., JR., AND C.M. MURVOSH. 1995. System- the larval exuviae. atic changes in certain Ephemeroptera studied by R.K. Allen. Pan-Pacific Entomologist 71:157–160. ACKNOWLEDGMENTS EDMUNDS, G.F., JR., AND R.D. WALTZ. 1996. Ephemer- optera. Pages 126–163 in R.W. Merritt and K.W. Cummins, editors, An introduction to the aquatic Boris Kondratieff (Fort Collins, CO), Brian insects of North America. 3rd edition. Kendall-Hunt, Krestian (Moscow, ID), David Krogmann Dubuque, IA. (West Lafayette, IN), Carole Lembi (West HAWKINS, C.P. 1985. Food habits of species of ephemerel- Lafayette, IN), Gary Lester, John Pfeiffer, and lid mayflies (Ephemeroptera: Insecta) in streams of Mike Walters (Moscow, ID) provided valuable Oregon. American Midland Naturalist 113:343–352. HILSENHOFF,W.L. 1982. Using a biotic index to evaluate discussion and technical assistance. We thank water quality in streams. Wisconsin Department of Arwin Provonsha (West Lafayette, IN) for ren- Natural Resources Technical Bulletin 132:1–22. 108 WESTERN NORTH AMERICAN NATURALIST [Volume 64

______. 1998. A modification of the biotic index of organic PERCIVAL, E., AND H. WHITEHEAD. 1929. A quantitative stream pollution to remedy problems and permit its study of some types of stream-bed. Journal of Ecol- use throughout the year. Great Lakes Entomologist ogy 17:282–314. 31:1–12. POTTS, M. 2000. Nostoc. Pages 465–504 in B.A. Whitton HYNES, H.B.N. 1961. The invertebrate fauna of a Welsh and M. Potts, editors, The ecology of cyanobacteria. mountain stream. Archiv fuer Hydrobiolgie 57: Kluwer Academic Publishers, Dordrecht, Nether- 344–348. lands. JACOBUS, L.M., AND W. P. M CCAFFERTY. 2002. Analysis of PRESCOTT, G.W. 1970. How to know the freshwater algae. some historically unfamiliar Canadian mayflies (Ephe- 2nd edition. William C. Brown, Dubuque, IA. meroptera). Canadian Entomologist 134:141–155. SABATER, S., AND I. MUÑOZ. 2000. Nostoc verrucosum JONES, J.R.E. 1950. A further ecological study of the River (Cyanobacteria) colonized by a chironomid larva in a Rheidol: the food of the common insects of the main Mediterranean stream. Journal of Phycology 36: stream. Journal of Animal Ecology 19:159–174. 59–61. KLEINHAUS, S., AND A.D. KEISER. 1988. Ecology and bio- SHEATH, R.G., AND K.M. COLE. 1992. Biogeography of mechanical consequences of living together. Induced stream macroalgae in North America. Journal of morphological change in a Nostoc and Cricotopus Phycology 28:448–460. symbiosis. American Zoologist 28:34A. SMITH, S.D. 1968. The Rhyacophila of the Salmon River MCCAFFERTY,W.P. 1981. Aquatic entomology. Jones and drainage of Idaho with special reference to larvae. Bartlett, Sudbury, MA. Annals of the Entomological Society of America 61: ______. 2001a. Notes on distribution and orthography 655–674. associated with some poorly known North American STARK, B.P., AND C.R. NELSON. 1994. Systematics, phy- mayflies (Ephemeroptera). Entomological News 112: logeny and zoogeography of genus Yoraperla (Ple- 121–122. coptera: Peltoperlidae). Entomologica Scandinavica ______. 2001b. Status of some historically unfamiliar Amer- 25:241–273. ican mayflies (Ephemeroptera). Pan-Pacific Ento- STEWART, K.W., AND B.P. STARK. 1988. Nymphs of North mologist 77:210–218. American stonefly genera (Plecoptera). Thomas Say MCCAFFERTY, W.P., AND J.P. BENSTEAD. 2002. Cladistic Foundation, Entomological Society of America 12: resolution and ecology of the Madagascar genus i–xiii + 1–460. Manohyphella Allen (Ephemeroptera: Teloganodi- STUDEMANN, D., AND P. L ANDOLT. 1997. Eggs of Ephe- dae). Annales de Limnologie 38:41–52. merellidae (Ephemeroptera). Pages 362–371 in P. MCCAFFERTY, W.P., AND T.- Q. W ANG. 2000. Phylogenetic Landolt and M. Sartori, editors, Ephemeroptera and systematics of the major lineages of pannote mayflies Plecoptera: biology—ecology—systematics. Mauron (Ephemeroptera: Pannota). Transactions of the Amer- + Tinguely & Lachat SA, Fribourg. ican Entomological Society 126:9–101. WARD, A.K., C.N. DAHM, AND K.W. CUMMINS. 1985. Nostoc MCSHAFFREY, D.G. 1988. Behavior, functional morphol- (Cyanophyta) productivity in Oregon stream ecosys- ogy, and ecology related to feeding in aquatic insects tems: invertebrate influences and differences be- with particular reference to Stenacron interpuncta- tween morphological types. Journal of Phycology 21: tum, Rhithrogena pellucida (Ephemeroptera: Hepta- 223–227. geniidae), and Ephemerella needhami (Ephemer- WIGGINS, G.B. 1996. Larvae of the North American caddis- optera: Ephemerellidae). Doctoral dissertation, Pur- fly genera (Trichoptera). 2nd edition. University of due University, West Lafayette, IN. Toronto Press, Toronto. MCSHAFFREY, D., AND W. P. M CCAFFERTY. 1990. Feeding WIRTH, W.W. 1957. The species of Cricotopus midges liv- behavior and related functional morphology of the ing in the blue-green alga Nostoc in California mayfly Ephemerella needhami (Ephemeroptera: Ephe- (Diptera: Tendipedidae). Pan-Pacific Entomologist merellidae). Journal of Insect Behavior 3:673–688. 33:121–126. ______. 1991. Ecological association of the mayfly Ephe- ZLOTY, J., AND G. PRITCHARD. 1997. Larvae and adults of merella needhami (Ephemeroptera: Ephemerellidae) Ameletus mayflies (Ephemeroptera: Ameletidae) and the green alga Cladophora (Chlorophyta: Clado- from Alberta. Canadian Entomologist 129:251–289. phoraceae). Journal of Freshwater Ecology 6:383–394. MORIHARA, D.K., AND W. P. M CCAFFERTY. 1979. The Baetis Received 26 November 2002 larvae of North America (Ephemeroptera: Baetidae). Accepted 10 April 2003 Transactions of the American Entomological Society 105:139–221. Western North American Naturalist 64(1), ©2004, pp. 109–124

A PUTATIVE HYBRID SWARM WITHIN OÖNOPSIS FOLIOSA (ASTERACEAE: ASTEREAE)

Jonathan F. Hughes1 and Gregory K. Brown2

ABSTRACT.—Oönopsis foliosa var. foliosa and var. monocephala are endemic to short-grass steppe of southeastern Colorado and until recently were considered geographically disjunct. The only known qualitative feature separating these 2 varieties is floral head type; var. foliosa has radiate heads, whereas var. monocephala heads are discoid. Sympatry between these varieties is restricted to a small area in which a range of parental types and intermediate head morphologies is observed. We used distribution mapping, morphometric analyses, chromosome cytology, and pollen stainability to characterize the sympatric zone. Morphometrics confirms that the only discrete difference between var. foliosa and var. monocephala is radiate versus discoid heads, respectively. The outer florets of putative hybrid individuals ranged from conspicuously elongated yet radially symmetric disc-floret corollas, to elongated radially asymmetric bilabiate- or deeply cleft corollas, to stunted ray florets with appendages remnant of corolla lobes. Chromosome cytology of pollen mother cells from both putative parental varieties and a series of intermediate morphological types collected at the sympatric zone reveal evidence of translocation heterozygosity. Pollen stainability shows no significant differences in viability between the parental varieties and putative hybrids. The restricted distribution of putative hybrids to a narrow zone of sympatry between the parental types and the presence of meiotic chromosome-pairing anomalies in these intermediate plants are consistent with a hybrid origin. The high stainability of putative-hybrid pollen adds to a growing body of evidence that hybrids are not universally unfit.

Key words: plant hybridization, hybrid zone, Asteraceae, Astereae, sympatry, floral traits, chromosome cytology, pollen stainability.

Populations that diverge because of isola- 1985, Stebbins 1985, Hewitt 1988, Arnold 1997). tion and natural selection may reunite and Clinal variation in morphology and gene fre- hybridize, followed by stabilization and con- quency characterizes tension zones and, with tinuation of hybrid genotypes (Stebbins 1969). cytonuclear disequilibrium, hybrids are less fit Hybridization is a frequent and important than the parents. This hybrid disadvantage process in plant speciation, and hybrid zones exists independent of the environment (Barton offer experimental material for studies of char- and Hewitt 1985, Hewitt 1988, Arnold 1997). acters and processes involved in divergence As with tension zones, the mosaic hybrid-zone and speciation (Stebbins 1969, 1985, Hewitt model assumes that hybrids are unfit compared 1988, Abbott 1992, Harrison 1993, Rieseberg to parental taxa and that selection against hy- and Ellstrand 1993, Ellstrand et al. 1996, Arnold brids is endogenous. Conversely, the mosaic 1997). Possible outcomes of hybridization in- hybrid-zone model assumes that parental taxa clude breakdown of isolating barriers, intro- have adapted to different environments and gression, increased genetic diversity, and the that patterns of morphologic or genetic varia- origin of adaptations, ecotypes, and species tion across the zone reflect patchy environ- (Hewitt 1988, Nason et al. 1992, Arnold 1997, ments (Harrison 1986, Wang et al. 1999). With Whitham et al. 1999). bounded hybrid superiority, hybrids are more Conceptual frameworks of hybrid zone evo- fit than the parents in the sympatric zone be- lution include tension zones, mosaic hybrid cause of natural selection in environments that zones, and bounded hybrid superiority (Arnold are novel compared with parental environments 1997). Tension zones exist when endogenous (Moore 1977, Harrison 1986, Freeman et al. selection against hybrids and dispersal of par- 1995, Graham et al. 1995, Arnold 1997). Exam- ents regulate and maintain the genetic struc- ples of bounded hybrid superiority include ture of the hybrid zone (Barton and Hewitt hybrids of Argyranthemum (Brochmann 1987),

1U.S. Geological Survey at the Department of Earth and Space Sciences, University of Washington, Box 351310, Seattle, WA 98195-1310. 2Department of Botany, Rocky Mountain Herbarium, University of Wyoming, Laramie, WY 82071.

109 110 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Iris (Arnold et al. 1990a, 1990b, Arnold 1997), Oönopsis foliosa is a perennial herb, 15–30 and Artemisia subspecies (Freeman et al. 1991, cm high, usually with several leafy, erect stems 1995, Graham et al. 1995, Wang et al. 1997). extending from a woody caudex. Hall (1928) Hybridization is relatively frequent in Aster- distinguished var. foliosa (then Haplopappus aceae (Jackson and Dimas 1981, Brochmann fremontii var. fremontii) from var. monocephala 1987, Freeman et al. 1991, Rieseberg 1991, (then H. fremontii var. monocephalus) by ray Abbott 1992, Nesom 1994, Rieseberg et al. 1995, floret occurrence and number of heads per Ellstrand et al. 1996). Adding a new example, stem. Hall described heads of var. foliosa as Schulz and Shaw (1992) reported a small area radiate with well-developed ray florets, these of sympatry between Oönopsis foliosa var. being several to solitary per stem. In contrast, foliosa and var. monocephala on the U.S. Army discoid solitary heads characterized var. mono- Piñon Canyon Maneuver Site (PCMS). PCMS cephala. Hall also noted a tendency for smaller is located approximately 40 miles northeast of heads in var. monocephala. Schulz and Shaw Trinidad, Colorado, in the semiarid, short-grass (1992) confirmed Hall’s distinction of these steppe of the Piñon Canyon region of Las Ani- varieties based upon the presence/absence of mas County (Fig. 1). Schulz and Shaw (1992) ray flowers, but not the number of heads per estimated the sympatric zone to be a few hun- stem. Both varieties are polymorphic with dred hectares in size and noted that it con- respect to number of heads per stem, varying tained plants with head morphologies inter- from one to several. mediate between var. foliosa (radiate) and var. The varieties of Oönopsis foliosa have chro- monocephala (discoid). They hypothesized hy- mosome numbers of 2n = 20. The base chro- bridization to be the most likely explanation for mosome number for the genus is x = 5, plac- the observed intermediate plants. ing O. foliosa at the tetraploid level with O. Our objectives in this study were to verify engelmannii. All remaining taxa of Oönopsis and map the extent of sympatry between the 2 are at or derived from the diploid level (2n = varieties of Oönopsis foliosa, to document the 10; Brown and Evans in preparation). range of intermediate head morphologies, and Oönopsis foliosa is restricted to the plains to elucidate the origin of the Oönopsis hybrid of southeastern Colorado (Brown and Evans in zone. preparation). Variety foliosa is especially common in the Arkansas River valley in Bent, TAXONOMIC HISTORY Crowley, and Otero Counties, while var. monocephala appears to be restricted to Las Animas Oönopsis is a genus nearly endemic to the County. For many years var. monocephala was high plains of Colorado and Wyoming. For rarely collected and, consequently, considered most of its taxonomic history, Oönopsis has rare. However, fieldwork (Schulz and Shaw been treated as a subgenus within Haplopappus 1992, Brown and Evans in preparation, this (Hall 1928). However, recent studies (Nesom study) has documented large populations of and Morgan 1990, Morgan and Simpson 1992, var. monocephala, especially on extensive clay Lane and Hartman 1996, Lane et al. 1996) flats east and northeast of Trinidad, Colorado. support the recognition of Oönopsis as a dis- The U.S. Department of the Interior (1993) tinct genus. Biologically, the group is perhaps changed the conservation status of O. foliosa most interesting because members are consid- var. monocephala from category 2, which means ered to be nutritionally obligate for, and hyper- there is some evidence of vulnerability, to cate- accumulators of, selenium and thus are reported gory 3C, which means the taxon is secure to be selenium indicator species (Trelease and throughout most of its range. The Colorado Trelease 1938, Rosenfeld and Beath 1964). A Natural Heritage Program (1997) lists var. pending taxonomic revision of Oönopsis (Brown monocephala as globally rare. and Evans in preparation) recognizes 6 taxa [O. engelmannii (A. Gray) E. Greene, O. foliosa METHODS (A. Gray) E. Greene var. foliosa, O. foliosa var. monocephala (A. Nelson) Kartesz and Gandhi, Distribution mapping and sample collect- O. multicaulis (Nutt.) E. Greene, O. wardii (A. ing, conducted during June 1992 and 1993, Gray) E. Greene, and a new species], and that coincided with the onset of anthesis for both nomenclature is followed here. varieties. We mapped the sympatric zone on 2004] HYBRID SWARM WITHIN OÖNOPSIS FOLIOSA 111

Colorado

Delhi

O. foliosa var. foliosa Bear Spring Thatcher Hills Big Arroyo N Hills

104°01'00" * 05 Dillingham Burson miles Camp atoire River Point 0 5 Tyrone Purg kilometers Brown * Sheep Oönopsis foliosa var. foliosa Camp O. foliosa monocephala Biernacki var. * Ranch Model * Putative hybrids * Outlying putative hybrids The Hogback PCMS Boundary U.S. Highway 350 Atchinson Topeka and O. foliosa var. monocephala 37°29'28" Santa Fe Railroad

Fig. 1. Location of U.S. Army Piñon Canyon Maneuver Site (PCMS) in Las Animas Co., Colorado, and the mapped zone of sympatry for Oönopsis foliosa var. foliosa and O. foliosa var. monocephala.

foot using U.S. Geological Survey topographic ental varieties and putative hybrids (Hughes maps (1:24,000 scale). Specimens used for veg- 1995). We measured corolla length, depth of etative measurements and all other vouchers corolla cleft on the adaxial/posterior side, and are on file at the Rocky Mountain Herbarium. androecium length on flowers preserved from See Appendixes A and B for information on 116 individuals (32 radiate, 55 intermediate, specimen locality and collection use. and 29 discoid). Radiate and discoid categories Floral heads with at least 2 rings of outer include individuals from, and disjunct from, florets in anthesis were fixed in FAA (formal- the sympatric zone. dehyde:glacial acetic acid:absolute ethanol; To evaluate a subjective classification of 1:1:18 v:v:v) for a minimum of 24 hours and head morphology, we divided the 116 samples transferred to 70% ethanol for storage. Mor- using principal components analysis (PCA) phological samples were collected from var. and WARD’S cluster analysis (Casgrain and foliosa and var. monocephala populations adja- Legendre 2001). Both analyses reveal the cent to and disjunct from the sympatric zone amount of variation explained by the morpho- (Appendix A) and from the sympatric zone logical traits, but each relies on different (Appendix B). We collected healthy specimens assumptions. When used together they pro- of goldenweed from isolated patches, or popu- vide a clear illustration of variation explained lations, as we mapped the geographic distribu- by the floral features. tion of sympatry. Examples of all known radi- We used chromosome cytology and pollen ate-discoid intermediates are included. staining to estimate cytonuclear disequilib- After an intensive study of more than 30 floral rium and viability, respectively. Differences in and vegetative traits, we identified 3 outermost chromosome structure that exist between the floret features useful for 3 differentiating par- varieties, such as translocations and inversions, 112 WESTERN NORTH AMERICAN NATURALIST [Volume 64 are expected during meiosis in hybrid individ- normalities that lead to genetically unbalanced uals (Stebbins 1971). Abnormal chromosome microgametophytes could result in pollen with pairing during meiosis in individuals with disrupted callose metabolism, and these pollen intermediate head morphology compared with samples will not stain. Thus, unstained pollen parental exemplars that are disjunct from the is considered reproductively sterile. The cor- sympatric zone suggests hybridization. Struc- relation between stainability and fertility is tural differences between chromosomes of the approximate because other factors affect fertil- hybrid parents typically have a negative ity after callose deposition. impact on the fertility of hybrids due to segre- We used florets from herbarium vouchers gation problems during meiosis (Stebbins 1971). for 119 individuals (48 var. foliosa, 34 from Gametes produced by the hybrid can suffer sympatric zone, 37 var. monocephala), repre- severe duplications and deletions and thus re- senting 53 populations, for pollen stainability duce the number of viable gametes. Low pollen studies. Anthers were dissected from disc flo- stainability in individuals with intermediate rets judged to be at anthesis, placed in a drop head morphology compared with parents sup- of aniline blue in lactophenol (Hauser and ports a hybrid tension zone (Barton and Hewitt Morrison 1964) on a microscope slide, and 1985, Arnold 1997). ruptured to release the pollen. We applied a To evaluate meiotic chromosome behavior, coverslip and allowed pollen grains to react we analyzed immature floral heads from 55 with the aniline blue for approximately 24 individual plants (5 var. foliosa, 45 from sym- hours before scoring. We prepared at least 2 patric zone, 5 var. monocephala). We fixed slides, each from a different floret, for each samples in Carnoy’s solution (Cohn 1964) for a collection. Using a compound microscope with minimum of 24 hours and then transferred bright-field illumination, we systematically them to 70% ethanol for storage at –20°C. We scanned slides to ensure that pollen grains were selected chromosome samples from individual scored only once. We scored pollen as “stained” plants bearing at least one head with mature when it appeared uniformly dark blue or black, outermost florets to ensure that we sampled and as “nonstained” when unevenly stained the range of intermediate head morphologies. and/or light blue. Additionally, we recorded Materials for chromosomal study of var. foliosa the number of pollen grains that were less than and var. monocephala were collected from 1/3 the diameter of normal pollen (micro- populations disjunct from the sympatric zone pollen). Total pollen counts for each collection (Appendix A). Using a dissecting microscope, were approximately 500 grains. Percent pollen we removed anthers from an individual floret stainability and percent micropollen were and transferred them into a drop of 1% acetic determined for each population. All popula- carmine on a microscope slide. While the tions of var. foliosa and var. monocephala used samples were in the stain, we removed micro- for pollen analysis are disjunct from the sym- sporocyte masses from the anther, discarded patric zone. All sympatric zone individuals the anther wall tissue, applied a coverslip, and checked for pollen stainability have intermedi- squashed microsporocyte masses with a com- ate head morphology. bination of heating over an alcohol flame followed by finger pressure on the coverslip. RESULTS Pollen stainability is an indirect measure of Geographic Limits of pollen fertility (Hauser and Morrison 1964). Sympatric Zone Aniline blue lactophenol is specific for callose, a carbohydrate produced by the immature Sympatry between var. foliosa and var. microgametophyte that forms a layer just in- monocephala is restricted to the western half side the pollen wall. The rationale for using of PCMS (Fig. 1). Field observation and voucher analine blue lactophenol stems from the notion collections reveal a morphological gradient that genetically balanced and physiologically from north to south through the sympatric healthy plants will produce a high percentage zone. The northern end of the gradient has of genetically balanced, normal immature micro- monomorphic populations of var. foliosa inter- gametophytes that deposit callose. Because nor- spersed with intermediates, while the south- mal pollen stains, it is therefore inferred to be ern end includes populations of var. mono- reproductively viable. Conversely, meiotic ab- cephala interspersed with intermediates. Near 2004] HYBRID SWARM WITHIN OÖNOPSIS FOLIOSA 113 the center of the sympatric zone at Burson 3I, 4) that is conspicuously smaller than the Camp (Fig. 1), monomorphic populations of type-5 ray-floret blade. Individuals with type- either parental type are rare and intermedi- 4 florets often have anomalous appendages or ates are most abundant. The morphological lobes but, regardless of corolla size or appear- gradient results in a spectrum of populations ance, produce a morphologically normal androe- that differ in the ratio of individuals with radi- cium (Fig. 4). Individuals with intra-head or ate (var. foliosa), discoid (var. monocephala), inter-head polymorphisms are uncommon but, and intermediate heads (putative hybrids; Fig. when present, usually involve the co-occurrence 2). Oönopsis populations have a patchy distrib- of type-3 and type-4 outer florets. We observed ution across the landscape. no polymorphisms in individuals with type-1 Some populations within the sympatric zone or type-5 outermost florets. contain no known individuals with intermedi- PCA and WARD’S cluster analysis distin- ate head morphology and are monomorphic guish corolla morphology of the parental taxa for either radiate or discoid heads. Popu- and show that putative hybrids are intermedi- lations with intermediates typically include only ate (Fig. 5). With an eigenvalue of 2.7, corolla 1 of the parental types. When considering the length explains 90% of the variation. This entire sympatric zone, the range of variation measure of variance exceeds that predicted by in outermost floret morphology appears to be a “broken-stick” random distribution (61%) and continuous (Fig. 3). Although we did not find thus is significant (Legendre and Legendre the varieties growing together west of PCMS, 1998, Casgrain and Legendre 2001). Eigenval- populations with intermediates are present ues of 0.2 and 0.1 for corolla cleft depth and along U.S. Highway 350 and west of Model. androecium length, respectively, do not explain enough variation, 8% and 2%, to exceed that Classification of Hybrid expected under a “broken-stick” random dis- Morphology tribution, 28% and 11%. Using corolla and androecial features, we Chromosomes and Inferences recognize 5 morphological categories of outer- of Viability most florets. Type-1 florets are disc florets, i.e., the normal outermost floret in var. mono- Meiotic chromosome behavior in discoid, cephala heads (Figs. 2A, 3A). In contrast, the radiate, and intermediate individuals associ- type-5 floret is the normal ray floret charac- ated with the sympatric zone (Appendix B) is teristic for var. foliosa (Figs. 2F, 3J). Type-5 flo- typically problematic (Fig. 6). Chain and ring rets are pollen sterile, lacking an androecium multivalents, some quite large, are common in (Fig. 4). prophase I. Other indications of irregular We classified the outermost florets of puta- meiosis in sympatric zone individuals include tive hybrids into 3 groups intermediate to bridge and fragment formation at anaphase parental morphology. Type-2 outermost florets and telophase I, chromosomal fragments at var- resemble disc florets (type-1); however, the ious stages, and telophase II cells with nuclei corolla tube is distinctly longer than adjacent polymorphisms (e.g., 1 or 2 micronuclei; 5 nu- inner disc florets (Figs. 3B–D, 4). Depth of clei of equal size; 2 large and 2 small nuclei). corolla cleft and androecium length do not In contrast, meiosis in parental exemplars from distinguish type-1 and type-2 florets (Fig. 4). populations distant from the sympatric zone Type-3 outermost florets have a conspicuously appears mostly normal. Although some quad- elongated corolla tube relative to adjacent disc rivalent and anaphase I bridge and fragment florets (Fig. 4). However, the distal end of the configurations are encountered in parental corolla is not radially symmetric but varies populations disjunct from the sympatric zone, from bilabiate (Figs. 2B–D, 3C–G) to cleft more they do not appear systemic (Brown and deeply on the adaxial (posterior) side, resulting Evans in preparation). in an incipient blade (Figs. 2D, 3H). The androe- Mean percent pollen stainability (Fig. 7) of cium of type-3 florets tends to be shorter than disjunct populations of var. foliosa (84.6 ± 7.9) that in type-1 and type-2 florets (Fig. 4). Type- is higher than that of var. monocephala (75.7 ± 4 outermost florets have deeper clefts on the 14.6). Mean percent pollen stainability of pop- adaxial (posterior) side (Fig. 4), a condition ulations of intermediates from the sympatric resulting in a reduced raylike blade (Figs. 2E, zone (84.1 ± 5.4) is more similar to var. foliosa. 114 WESTERN NORTH AMERICAN NATURALIST [Volume 64

A B

C D

E F

Fig. 2. Representative floral head morphology for Oönopsis foliosa var. foliosa, var. monocephala, and putative hybrid intermediates: A, discoid head diagnostic for var. monocephala; all outermost florets are disc florets (type 1); B and C, intermediate heads; outermost florets with elongated corolla tube with bilabiate distal end (type 3); D, intermediate head with both type-3 and type-4 outermost florets (polymorphic); E, intermediate head with stunted ray florets (type 4); F, radiate head characteristic of var. foliosa; all outermost florets are ray florets (type 5). Head width excluding the ring of outermost florets is ~13 mm. 2004] HYBRID SWARM WITHIN OÖNOPSIS FOLIOSA 115

Corolla length Type 1

Type 2

Type 3

Type 4

Type 5

6 10 14 18 22 26 mm

A B C D E Corolla adaxial cleft depth Type 1

Type 2

Type 3

Type 4

Type 5

0 4 8 12 16 20 mm

Androecium length Type 1 F G H I J Type 2 Type 3

Fig. 3. Silhouette drawings of outermost florets from Type 4 Oönopsis foliosa var. foliosa, var. monocephala, and putative hybrid intermediates. For clarity, only the ovary (~4 Type 5 mm) and corolla are shown. The horizontal line to the left of each floret indicates the distal extent of the pappus for 0 2 4 6 8 10 that floret. A, Type 1, normal disc floret characteristic of mm var. monocephala; B, type 2, corolla tube conspicuously elongated relative to adjacent, normal disc florets; C–E, type 3, bilabiate corollas; F–G, type 3, cleft on adaxial Fig. 4. Comparison of summary statistics for corolla (posterior) side of corolla resulting in an incipient blade; length, depth of adaxial (posterior) corolla cleft, and H–I, type 4, distal half of corolla becomes more bladelike; androecium length. Type-1 and type-5 florets correspond note irregular corolla lobe in H; J, type 5, ray floret char- to var. monocephala (disc) and var. foliosa (ray) florets, acteristic for var. foliosa. respectively. For each floret type the minimum and maximum (horizontal line), the mean (vertical line), and 1 standard deviation from the mean (rectangle) are indicated. Sample sizes are as follows: type 1, 25; type 2, 20; type 3, 19; type 4, 16; type 5, 32, except for depth of corolla adaxial cleft, which is 29. Micropollen frequency is highest in var. monocephala (2.4 ± 2.6), lowest in var. foliosa (0.6 ± 0.7), and intermediate in putative hybrids (1.8 zone and distant outliers suggest that the ± 1.5). hybrid zone may be expanding (Fig. 1; Schulz and Shaw 1992). Populations of O. foliosa with- DISCUSSION in the sympatric zone frequently contain individuals with head morphology intermediate to Sympatry Defined with radiate var. foliosa and discoid var. monoce- Floral Morphology phala, suggesting intervarietal hybridization. The area of geographic sympatry for the 2 Away from the sympatric zone, populations of varieties of O. foliosa (Schulz and Shaw 1992) var. foliosa and var. monocephala are mono- is confirmed. Enlarged limits of this sympatric morphic for their respective head morphology, 116 WESTERN NORTH AMERICAN NATURALIST [Volume 64

1.61

A. SUBJECTIVE

0.95 Corolla Andro

0.29

Cleft

-0.37

-1.03 -3.75 -2.33 -0.91 0.51 1.93

Axis 2 1.61

B. WARD'S

0.95 Corolla Andro

0.29

Cleft

-0.37

-1.03 -3.75 -2.33 -0.91 0.51 1.93 Axis 1

Head type Type 1 - Discoid heads - Oönopsis foliosa var. monocephala Type 2 intermediate - Putative hybrid Type 3 intermediate - Putative hybrid Type 4 intermediate - Putative hybrid Type 5 - Radiate heads - Oönopsis foliosa var. foliosa

Fig. 5. Principal components analysis of floral traits including corolla and androecium lengths and corolla cleft depth. Data are standardized. Corolla = corolla length; Cleft = corolla cleft depth; Andro = androecium length. A, Subjective classification of head types; B, classification using similarity measures of raw data and WARD’S cluster analysis (see text). Ellipses include 90% of samples in each group.

which remains the only known discrete morphological feature separating these varieties. Fig. 6 (opposite page). Microsporocyte meiosis representative of individuals collected from the sympatric zone: The morphological range of intermediate A, diakinesis with a chain quadrivalent (IV); other IVs outermost florets (Figs. 3, 4), and thus head may be present; n = 10; B, diplotene/diakinesis with a morphology (Fig. 2), suggests a cline. The 5 complex multivalent that appears to involve 6 to 8 biva- floret categories show morphological continu- lents; arrow indicates a probable quadrivalent; C, diakinesis, with large circle multivalent (arrows) involving at least ity between radiate heads characteristic of var. 6 bivalents; D–G, late metaphase I to early anaphase I: D, foliosa and discoid heads characteristic of var. arrows indicate 3 circle IVs and a possible supernumerary monocephala (Figs. 3, 4). However, the 3 inter- bivalent; E, arrow indicates a circle IV; F, large arrow indicates circle multivalent, a possible hexivalent; small arrows mediate floret categories (types 2, 3, 4) are indicate fragments; G, large arrow indicates circle multi- artificial partitions of the variation found, and valent; small arrows indicate 2 fragments; H, telophase assigning some florets to a particular “type” with bridge between 2 nuclei. 2004] HYBRID SWARM WITHIN OÖNOPSIS FOLIOSA 117

A B

C D

E F

G H 118 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Stained pollen

monocephala

intermediate

foliosa

0 20 40 60 80 100 Percent

Micropollen

monocephala

intermediate

foliosa

0 4 8 12 16 Percent

Fig. 7. Summary statistics for pollen stainability. Categories foliosa and monocephala refer to populations of these varieties disjunct from the sympatric zone. Intermediate category includes sympatric zone individuals with intermediate head morphology. Sample sizes: var. monocephala (25), var. foliosa (14), intermediates (14).

becomes arbitrary. Individuals in the sympatric lopappus, H. aureus was transferred to the zone with intermediate outermost florets are genus Rayjacksonia (R. aurea; Lane and Hart- predominantly monomorphic concerning type man 1996), and H. venetus ssp. venetus was (e.g., Figs. 2C, 2E). The occasional case of mixed placed in Isocoma (I. veneta; Nesom 1991). polymorphism (e.g., co-occurrence of type-3 Thus, in the revised nomenclature (Lane and and type-4 florets in same head; Fig. 2D) sug- Hartman 1996), the hybrids made by Jackson gests developmental plasticity within an inflo- and Dimas (1981) are intergeneric. Based on rescence. the range of outer floret variability found in PCA and WARD’S cluster analysis distin- the sympatric zone, a single-locus, 2-allele guish the corolla morphology of the parental model for ray floret expression within O. foliosa taxa and show that outer-floret morphology of is unlikely. The continuous array of outermost putative hybrids is intermediate to the parental floret morphology in plants from the sympatric exemplars (Fig. 5). Segregation of parental types zone suggests a polygenic system of inheri- and putative hybrids by cluster analysis and tance. ordination supports our subjective determina- A Chromosome Cytology–Pollen tion of morphology types. Stainability Paradox While nothing is known about the genetics of ray floret expression in Oönopsis, informa- Observed deviations from normal meiosis tion is published on related taxa. Jackson and are common in individuals from the sympatric Dimas (1981) report a 3:1 ratio of rayed to ray- zone. In particular, large and complex multiva- less F2 hybrids in a cross between Haplopap- lent formation (Fig. 6) seems to be a general pus aureus (radiate) and H. venetus ssp. vene- feature of sympatric zone individuals, includ- tus (discoid). This ratio indicates that a single ing radiate-headed, discoid-headed, and inter- locus with complete dominance controls ray mediates. Multivalent formation is a manifes- floret expression (Jackson and Dimas 1981). tation of translocation heterozygosity and a With the dismantling of North American Hap- characteristic feature of hybrids involving par- 2004] HYBRID SWARM WITHIN OÖNOPSIS FOLIOSA 119 ental genomes that differ by 1 or more translo- diploid hybrids of Helianthus. These studies cations (Stebbins 1971). These meiotic abnor- show that directional and nonrandom genetic malities suggest divergence between the par- mechanisms facilitate stabilization of hybrid ental varieties and intermediate individuals of heterozygotes (Rieseberg et al. 1995, Ozkan et hybrid origin. However, a more thorough al. 2001). Similar mechanisms may explain the exploration of chromosome pairing is required discordant chromosome cytology and pollen to quantify chromosome differences between stainability of Oönopsis hybrids. parents and putative hybrids and to assess the Hybrid Zone Evolution degree of backcrossing. Radiate and discoid individuals (i.e., apparent ‘parental’ types) from Pollen stainability, morphology of disk flo- the sympatric zone displaying more meiotic rets, and the high frequency of hybrids in the abnormalities than do disjunct parental indi- sympatric zone suggest that hybrid fitness is viduals will suggest that backcrossing occurs equal to or greater than the fitness of the par- and that the sympatric zone may constitute a ents (Fig. 7), which supports bounded hybrid hybrid swarm (Brochmann 1987). superiority (Arnold 1997, Wang et al. 1999). Although meiotic abnormalities are a char- With bounded hybrid superiority, hybrid fit- acteristic of sympatric zone individuals, pollen ness coincides with the sympatric-zone envi- stainability does not reveal reduced hybrid fit- ronment (Freeman et al. 1995, Arnold 1997), a ness, a finding that is contrary to the tension relationship that has failed to be disproved zone model. Mean percent stainability in here. However, the patchy distribution of putative hybrids (84%) is nearly identical to Oönopsis hybrids, which may correlate with var. foliosa (85%) and greater than var. mono- patches of selenium-bearing soils, suggests that cephala (76%), although overlapping standard the sympatric zone may be a hybrid mosaic. deviations show that differences in pollen stain- With this interpretation, patchy environments ability are not significant (Fig. 7). Possible ex- support 1 or the other parental taxon, and dis- planations are that (1) fertility is reduced in turbance creates empty patches that allow 1 putative hybrids, but callose production is not taxon to invade the range of the other without affected, and (2) tetraploidy buffers the puta- the cost of hybridization (Harrison 1986, Arnold tive hybrids against lethal duplications and de- 1997). However, the mosaic hybrid-zone model letions that affect callose production or fertility. assumes that hybrids are unfit compared with It is possible for hybrids to express genetic dis- the parents, a conclusion that available data do equilibrium and equivalent or higher fitness not support. To determine the relative fitness than the parents (Arnold 1997). The level of of hybrids in the sympatric zone, manipulative nucleotide-sequence divergence within Oönop- experiments that include reciprocal transplants sis is low, suggesting a relatively recent origin in combination with soil analyses are required for the genus (Boerema et al. 2001). A recent (Wang et al. 1997, 1999). Results of these ex- origin coupled with limited divergence (i.e., periments would clarify the importance of segmental allopolyploidy; Stebbins 1971) may endogenous and exogenous selection and there- explain the unexpectedly high levels of stain- by distinguish between mosaic hybrid-zone ing pollen in the putative hybrids. and bounded hybrid-superiority models. Using Aegilops and Triticum hybrids, Ozkan Taxonomic Implications et al. (2001) show that genome loss and chromosomal modification in F1 and later genera- High degree of morphological similarity, tion allopolyploids are nonrandom, direction- distinct but regionally close distributions, and al, and reproducible. Ozkan et al. (2001) de- a common ploidy level support the treatment scribe allopolyploidy as 2 genomic shocks: (1) of discoid O. foliosa as a variety. Furthermore, the combination of two divergent genomes and the presence or absence of ray florets within (2) polyploidy. Newly formed allopolyploid ge- Astereae is not a species-defining character nomes respond to these shocks with a “burst (Jackson and Dimas 1981). This restriction of irreversible genomic reorganizations and leads to the implicit assumption that var. modifications” (Ozkan et al. 2001:1739). These foliosa and var. monocephala are sister taxa. changes include modifications at the chromo- Molecular data, however, provide a different some and sequence level. Similarly, Rieseberg perspective for the relationship between these et al. (1995) report directional changes in 2 varieties. Phylogenetic analysis of DNA- 120 WESTERN NORTH AMERICAN NATURALIST [Volume 64 sequence data from 2 chloroplast intergenic AND W.A. TURNER. 1995. Narrow hybrid zone be- spacers (trnL-trnF and psbA-trnH) and the tween two subspecies of big sagebrush, Artemisia tridentata (Asteraceae). III. Developmental instabil- nuclear-ribosomal internal-transcribed spacers ity. American Journal of Botany 82:1144–1152. (Boerema et al. 2001) places var. foliosa and FREEMAN, D.C., W.A. TURNER, E.D. MCARTHUR, AND J.H. var. monocephala in different clades involving GRAHAM. 1991. Characterization of a narrow hybrid different polyploid events. If confirmed, their zone between two subspecies of big sagebrush hybrids would be interspecific. (Artemisia tridentata: Asteraceae). American Journal of Botany 78:805–815. GRAHAM, J.H., D.C. FREEMAN, AND E.D. MCARTHUR. 1995. ACKNOWLEDGMENTS Narrow hybrid zone between two subspecies of big sagebrush (Artemisia tridentata: Asteraceae). II. Selec- Keith Schulz and Randall Terry provided tion gradients and hybrid fitness. American Journal field assistance and discussion. We extend of Botany 82:709–716. HALL, H.M. 1928. The genus Haplopappus: a phylogene- thanks to the United States Army, Fort Carson, tic study in the Compositae. Carnegie Institution of for permission to use the Piñon Canyon Washington 389. Maneuver Site. Reviews by Elizabeth Elle HARRISON, R.G. 1986. Pattern and process in a narrow and Carl Freeman greatly improved the manu- hybrid zone. Heredity 56:337–349. script. The Department of Botany, University ______. 1993. Hybrid zones and the evolutionary process. Oxford University Press, New York. 364 pp. of Wyoming, and a National Science Founda- HAUSER, E.J.P., AND J.H. MORRISON. 1964. The cytochem- tion grant to GKB provided funding. ical reduction of nitro blue tetrazolium as an index of pollen viability. American Journal of Botany 51: LITERATURE CITED 748–752. HEWITT, G.M. 1988. Hybrid zones—natural laboratories ABBOTT, R.J. 1992. Plant invasions, interspecific hybridiza- for evolutionary studies. Trends in Ecology and Evo- tion and evolution of new plant taxa. Trends in Ecol- lution 3:158–167. ogy and Evolution 7:401–405. HUGHES, J.F. 1995. Putative hybridization between two ARNOLD, M.L. 1997. Natural hybridization and evolution. varieties of goldenweed (Oönopsis foliosa: Aster- Oxford University Press, Oxford. 215 pp. aceae). Master’s thesis, University of Wyoming, ARNOLD, M.L., B.D. BENNETT, AND E.A. ZIMMER. 1990a. Laramie. 108 pp. Natural hybridization between Iris fulva and I. hex- JACKSON, R.C., AND C.T. DIMAS. 1981. Experimental evi- agona: pattern of ribosomal DNA variation. Evolu- dence for systematic placement of the Haplopappus tion 44:1512–1521. phyllocephalus complex (Compositae). Systematic ARNOLD, M.L., J.L. HAMRICK, AND B.D. BENNETT. 1990b. Botany 6:8–14. Allozyme variation in Louisiana irises: a test for LANE, M.A., D.R. MORGAN, Y. SUH, B.B. SIMPSON, AND introgression and hybrid speciation. Heredity 65: R.K. JANSEN. 1996. Relationships of North American 297–306. genera of Astereae, based on chloroplast DNA re- BARTON, N.H., AND G.M. HEWITT. 1985. Analysis of hybrid striction site data. Pages 49–77 in D.J.N. Hind and zones. Annual Review of Ecology and Systematics H.J. Beentje, editors, Compositae: systematics. Pro- 16:113–148. ceedings of the International Compositae Confer- BOEREMA, D., T.M. EVANS, AND G.K. BROWN. 2001. Phylo- ence, Kew, volume I. Royal Botanic Gardens, Kew, genetic relationships in Oönopsis (Asteraceae) based United Kingdom. on DNA sequences from three spacer regions. Botany LANE, M.A., AND R.L. HARTMAN. 1996. Reclassification of 2001 Abstracts, Botanical Society of America. North American Haplopappus completed: Rayjack- BROCHMANN, C. 1987. Evaluation of some methods for sonia, gen. nov. American Journal of Botany 83: hybrid analysis, exemplified by hybridization in Argyr- 356–370. anthemum (Asteraceae). Nordic Journal of Botany LEGENDRE, P., AND L. LEGENDRE. 1998. Numerical ecol- 7:609–630. ogy. 2nd edition. Elsevier, Amsterdam. 853 pp. CASGRAIN, P., AND P. L EGENDRE. 2001. The R package for MOORE, W.S. 1977. An evaluation of narrow hybrid zones multivariate and spatial analysis, version 4.0 d5, user’s in vertebrates. Quarterly Review of Biology 52: manual. Département des sciences biologiques, Uni- 263–277. versité de Montréal. http://www.fas.umontreal.ca/ MORGAN, D.R., AND B.B. SIMPSON. 1992. A systematic BIOL/legendre/. study of Machaeranthera (Asteraceae) and related COHN, N.S. 1964. Elements of cytology. Harcourt, Brace, groups using restriction site analysis of chloroplast and World, Inc., New York. 368 pp. DNA. Systematic Botany 17:511–531. COLORADO NATURAL HERITAGE PROGRAM. 1997. Colorado’s NASON, J.D., N.C. ELLSTRAND, AND M.L. ARNOLD. 1992. natural heritage: rare and imperiled animals, plants Patterns of hybridization in populations of oaks, and natural communities. Colorado Natural Heri- manzanitas, and irises. American Journal of Botany tage Program 3(1). 79:101–111. ELLSTRAND, N.C., R. WHITKUS, L.H. RIESEBERG. 1996. NESOM, G.L. 1991. Taxonomy of Isocoma (Compositae: Distribution of spontaneous plant hybrids. Proceed- Astereae). Phytologia 70:69–114. ings of the National Academy of Sciences in the ______. 1994. Hybridization in the tribe Astereae (Aster- United States 93:5090–5093. aceae). Phytologia 77:298–307. FREEMAN, D.C., J.H. GRAHAM, D.W. BYRD, E.D. MCARTHUR, NESOM, G.L., AND D.R. MORGAN. 1990. Reinstatement of 2004] HYBRID SWARM WITHIN OÖNOPSIS FOLIOSA 121

Tonestus (Asteraceae: Astereae). Phytologia 68:174– Annals of the Missouri Botanical Garden 72:824–832. 180. TRELEASE, S.F., AND H.M. TRELEASE. 1938. Selenium as a OZKAN, H., A.A. LEVY, AND M. FELDMAN. 2001. Allopoly- stimulating and possibly essential element for indica- ploidy-induced rapid genome evolution in the wheat tor plants. American Journal of Botany 25:372–380. (Aegiolops-Triticum) group. Plant Cell 13:1735–1747. UNITED STATES DEPARTMENT OF THE INTERIOR. 1993. RIESEBERG, L.H. 1991. Homoploid reticulate evolution in Endangered and threatened wildlife and plants. Helianthus (Asteraceae): evidence from ribosomal Review of plant taxa for listing as endangered or genes. American Journal of Botany 78:1218–1237. threatened species. Federal Register 58:5144–5190. RIESEBERG, L.H., AND N.C. ELLSTRAND. 1993. What can WANG, H., E.D. MCARTHUR, AND D.C. FREEMAN. 1999. molecular and morphological markers tell us about Narrow hybrid zone between two subspecies of big plant hybridization? Critical Reviews in Plant Sci- sagebrush (Artemisia tridentata: Asteraceae). IX. Ele- ences 12:213–241. mental uptake and niche separation. American Jour- RIESEBERG, L.H., C. VAN FOSSEN, AND A.M. DESROCHERS. nal of Botany 86:1099–1107. 1995. Hybrid speciation accompanied by genomic WANG, H., E.D. MCARTHUR, S.C. SANDERSON, J.H. GRAHAM, reorganization in wild sunflowers. Nature 375:313– AND D.C. FREEMAN. 1997. Narrow hybrid zone be- 316. tween two subspecies of big sagebrush (Artemisia ROSENFELD, I., AND O.A. BEATH. 1964. Selenium: geo- tridentata: Asteraceae). IV. Reciprocal transplant ex- botany, biochemistry, toxicity, and nutrition. Acade- periments. Evolution 51:95–102. mic Press, New York. 411 pp. WHITHAM, T.G., G.D. MARTINSEN, K.D. FLOATE, H.S. SCHULZ, K.A., AND R.B. SHAW. 1992. Status of Haplopap- DUNGEY, B.M. POTTS, AND P. K EIM. 1999. Plant hybrid pus fremontii A. Gray ssp. monocephalus (A. Nelson) zones affect biodiversity: tools for a genetic-based Hall [Asteraceae] in Colorado. Prairie Naturalist 24: understanding of community structure. Ecology 80: 143–148. 416–428. STEBBINS, G.L. 1969. The significance of hybridization for plant taxonomy and evolution. Taxon 18:26–35. Received 3 August 2001 ______. 1971. Chromosomal evolution in higher plants. Accepted 13 January 2003 Addison-Wesley Publishing Company, Reading, MA. 216 pp. ______. 1985. Polyploidy and the invasion of new habitats.

Appendixes start on the following page. 122 WESTERN NORTH AMERICAN NATURALIST [Volume 64

APPENDIX A. Voucher specimens of Oönopsis foliosa var. foliosa and var. monocephala from locations that are disjunct from the sympatric zone. All localities are in the state of Colorado and are listed from north to south and east to west. Study-use coding: M = morphology; C = chromosomes; P = pollen. Population location Collection Study use Oönopsis foliosa var. foliosa (radiate heads) Huerfano Co., 10 mi NE Walsenberg Brown 2794 C Huerfano Co., 7.8 mi W Gardner Brown 2710 M,P Kiowa Co., 6.0 mi W Haswell Brown 2777 M,P Kiowa Co., 1.0 mi S Haswell Evans 18 C Otero Co., 17.4 mi N La Junta Brown 2778 P Otero Co., 13.7 mi S La Junta Evans 15 C Otero Co., 16.9 mi W La Junta Brown 2779 P Pueblo Co., 4.0 mi W Pueblo-Otero Co. lines on CO Hwy 10 Brown 2781 P Pueblo Co., 9.6 mi W Pueblo Brown 2795 M,P,C Pueblo Co., 1.1 mi S CO Hwy 10 Brown 2782 P Pueblo Co., 5.0 mi S CO Hwy 10 Brown 2785 P Las Animas Co., 14.1 mi S CO Hwy 10 Brown 2787 P Las Animas Co., 14.5 mi SW La Junta Hughes 154 M Las Animas Co., 0.8 mi N Delhi Hughes 153 M Las Animas Co., 6.4 mi N Thatcher Brown 2788 P Las Animas Co., 11.8 mi NE Thatcher Hughes 124 M,P,C Las Animas Co., 26.8 mi NE Thatcher Hughes 125 M,P Las Animas Co., 1.2 mi S Thatcher Hughes 151 M Las Animas Co., 12 mi NW Tyrone Brown 2789 P Las Animas Co., jct US Hwy 10 & Co. Rd 5 Hughes 126 M,P Oönopsis foliosa var. monocephala (discoid heads) Las Animas Co., 0.9 mi W Model Brown 2791 P Las Animas Co., 1.9 mi W Model Brown 2792 P,C Las Animas Co., 2.0 mi W Model Hughes 139 M,C Las Animas Co., 2.1 mi N Model Brown 2790 P,C Las Animas Co., 2.5 mi W Model Hughes 138 M Las Animas Co., 3.0 mi W Model Hughes 137 M Las Animas Co., 3.1 mi W Model Hughes 75 M,P Las Animas Co., 3.8 mi W Model Evans 10 P Las Animas Co., 4.4 mi W Model Brown 2978 P Las Animas Co., 4.7 mi W Model Evans 9 P Las Animas Co., 9.7 mi W Model Brown 2793 P Las Animas Co., 10.2 mi W Model Evans 8 P Las Animas Co., 13.6 mi W Model Evans 7 P,C Las Animas Co., 15.3 mi W Model Brown 2977 P Las Animas Co., 6.1 mi S US Hwy 160 Brown 3106 M,P Las Animas Co., 1.5 mi W Trinchera Evans 80 P Las Animas Co., 2.0 mi W Trinchera Brown 3107 M,P Las Animas Co., 2.0 mi W Trinchera Evans 81 P Las Animas Co., 6.0 mi W Trinchera Evans 79 P Las Animas Co., 14.0 mi W Trinchera Brown 3108 M,P Las Animas Co., 11.8 mi E US Hwy 350 Brown 3109 P Las Animas Co., 5.1 mi E US Hwy 350 Hughes 77 M,P Las Animas Co., 4.1 mi E Brown Sheep Camp Brown 3110 P Las Animas Co., 1.4 mi S Brown Sheep Camp Hughes 134 M Las Animas Co., 5.6 mi E US Hwy 350 Evans 11 P Las Animas Co., 1.7 mi E US Hwy 350 Evans 74 P Las Animas Co., 0.7 mi W Frijole Creek Evans 76 P Las Animas Co., 22.5 mi ESE Trinidad Evans 82 P Las Animas Co., 0.3 mi N Model Brown 2708 P Las Animas Co., 2.3 mi S Tyrone Hughes 74 M,C 2004] HYBRID SWARM WITHIN OÖNOPSIS FOLIOSA 123

APPENDIX B. Voucher specimens of putative hybrids between Oönopsis foliosa var. foliosa and var. monocephala. All localities are in, or just outside, the U.S. Army Piñon Canyon Maneuver Site (PCMS), Las Animas Co., Colorado. Radi- ate head type corresponds to var. foliosa, discoid to var. monocephala, and intermediate to putative hybrids. Study-use coding: M = morphology; C = chromosomes; P = pollen. Population location Head type Collection Study use 0.3 mi NW Burson Camp discoid Hughes 1 M 0.4 mi NW Burson Camp intermediate Hughes 2 M,P intermediate Hughes 3 M,P discoid Hughes 4M,C intermediate Hughes 5 M,C intermediate Hughes 6 M,P,C intermediate Hughes 8 M,C discoid Hughes 9 M intermediate Hughes 10 M,P intermediate Hughes 12 M 0.8 mi NW Burson Camp intermediate Hughes 13 M,C intermediate Hughes 14 M,P radiate Hughes 15 M discoid Hughes 16 M,C radiate Hughes 18 M discoid Hughes 19 M,C radiate Hughes 20 M intermediate Hughes 21 M,P intermediate Hughes 22 M,P,C intermediate Hughes 23 M,P intermediate Hughes 24 M,C 2.3 mi NW Burson Camp intermediate Hughes 133 M 2.0 mi W Burson Camp discoid Hughes 25 M,C discoid Hughes 26 M discoid Hughes 27 M discoid Hughes 28 M discoid Hughes 29 M 2.3 mi W Burson Camp intermediate Hughes 30 M,P,C intermediate Hughes 31 P,C intermediate Hughes 32 M intermediate Hughes 33 M discoid Hughes 34 M,C intermediate Hughes 35 M intermediate Hughes 36 M,P intermediate Hughes 37 M discoid Hughes 39 M intermediate Hughes 40 M,C intermediate Hughes 42 M,P intermediate Hughes 43 M,P intermediate Hughes 44 M,P,C 2.4 mi W Burson Camp intermediate Hughes 45 M discoid Hughes 46 M,C discoid Hughes 47 M radiate Hughes 48 M,C intermediate Hughes 49 M intermediate Hughes 50 M intermediate Hughes 51 M radiate Hughes 53 C radiate Hughes 54 M,C 0.2 mi SE Burson Camp intermediate Hughes 80 P intermediate Hughes 81 P 0.7 mi SE Burson Camp intermediate Hughes 56 M,P,C radiate Hughes 57 C intermediate Hughes 58 P,C intermediate Hughes 59 M,P,C radiate Hughes 60 M radiate Hughes 61 M 124 WESTERN NORTH AMERICAN NATURALIST [Volume 64

APPENDIX B. Continued. Population location Head type Collection Study use intermediate Hughes 62 M,P discoid Hughes 63 M,C intermediate Hughes 64 M,C radiate Hughes 65 M,C 1.2 mi SE Burson Camp intermediate Hughes 67 M 1.3 mi SE Burson Camp intermediate Hughes 68 M,P 1.4 mi SE Burson Camp radiate Hughes 69 M 1.7 mi SE Burson Camp intermediate Hughes 70 P,C discoid Hughes 71 C radiate Hughes 72 M intermediate Hughes 73 M,P 0.3 mi E Burson Camp intermediate Hughes 145 M,C 1.6 mi E Burson Camp radiate Hughes 89 M,C discoid Hughes 90 C intermediate Hughes 91 M,P radiate Hughes 92 M,C discoid Hughes 93 M radiate Hughes 94 M intermediate Hughes 95 C radiate Hughes 96 M discoid Hughes 97 M,C radiate Hughes 98 M intermediate Hughes 99 M,P 3.7 mi NW Biernacki Ranch intermediate Hughes 100 C radiate Hughes 101 M,C intermediate Hughes 102 M,C intermediate Hughes 103 M,P,C intermediate Hughes 104 P,C radiate Hughes 105 M discoid Hughes 106 C discoid Hughes 107 M 2.0 mi NW Dillingham Point intermediate Hughes 109 M intermediate Hughes 110 M,P intermediate Hughes 111 M 4.5 mi NE Burson Camp intermediate Hughes 113 M intermediate Hughes 115 M,P intermediate Hughes 116 M,P radiate Hughes 117 M radiate Hughes 118 M radiate Hughes 119 M discoid Hughes 120 M radiate Hughes 121 M intermediate Hughes 122 C radiate Hughes 123 M,C 5.3 mi NE Burson Camp intermediate Hughes 130 P 3.0 mi N Tyrone intermediate Hughes 146 M 3.3 mi N Tyrone intermediate Hughes 140 P intermediate Hughes 148 M,P intermediate Hughes 149 M intermediate Hughes 150 M 3.3 mi NE Tyrone radiate Hughes 108 M,C 1.2 mi S Tyrone intermediate Hughes 147 M 1.8 mi N Brown Sheep Camp intermediate Hughes 141 M,P,C 3.0 mi N Brown Sheep Camp intermediate Hughes 142 M,C 0.9 mi S Dillingham Point intermediate Hughes 143 M 0.4 mi NE Dillingham Point intermediate Hughes 144 M Western North American Naturalist 64(1), ©2004, pp. 125–130

SOIL COMPACTION FROM HUMAN TRAMPLING, BIKING, AND OFF-ROAD MOTOR VEHICLE ACTIVITY IN A BLACKBRUSH (COLEOGYNE RAMOSISSIMA) SHRUBLAND

Simon A. Lei1

ABSTRACT.—Soil compaction from human trampling, biking, and off-road motor vehicle traffic was quantitatively investigated in a blackbrush (Coleogyne ramosissima) shrubland in Kyle Canyon of the Spring Mountains in southern Nevada. A significant difference was detected in soil compaction, bulk density, and percent pore space at a particular frequency of visits in each of 4 disturbance types. On average a single vehicle pass was equivalent to 10 human footprints. Ten and 100 footprints were equivalent to 1 motorcycle pass and 10 vehicle passes, respectively. Soil compaction is a product of increased bulk density and decreased pore space. The degree of soil compaction is a function of disturbance type and visit frequency when examining these 2 factors independently. However, interactive effects of disturbance type and visit frequency on soil bulk density, compaction, and percent pore space were not significantly different. The greatest effects occurred during the first few passes, with changes per pass decreasing as the number of passes increased in all 4 trails. Results of this study suggest that the effects of hiking and biking slowly increase over time relative to the effects of motor vehicle traffic in the Coleogyne shrubland of Kyle Canyon in southern Nevada.

Key words: trampling, hiking, biking, motorcycle, vehicle, soil compaction, disturbance, blackbrush, Coleogyne ramosissima, Kyle Canyon, Spring Mountains, southern Nevada.

Kyle Canyon, a part of the Toiyabe National Webb 1982, 1983, 2002, Webb et al. 1986). Forest and located in the Spring Mountains, is Weaver and Dale (1978) determined that soil a popular recreational area for nature-loving bulk density increases with increasing use by tourists and local residents in southern Nevada. horses, hikers, and motorcycles in Montana, Due to rapid population growth and use of off- but they did not relate the changes other than road motor vehicles in recent years, consider- qualitatively to type of impact or number of able increase has occurred in anthropogenic passes. (recreational) activities, and landscape distur- Human trampling, biking, and off-road motor bance has become increasingly noticeable. A vehicle traffic adversely affect soil properties complex network of roads and trails, including by compacting the soil (Wilshire and Nakata those created by human foot, bicycle, and 1976). Severely compacted soils reduce porosity motor vehicle traffic, is evident to casual ob- in southern Nevada (Marble 1985). Compaction servers and appears to have long-term, adverse through animal, human, or vehicle traffic can effects on soil properties in a blackbrush increase soil bulk density due to applied pres- (Coleogyne ramosissima) shrubland. sure or loading (Gill and Vandenberg 1967). Coarse, sandy desert soils on bajadas and Bulk density, which increases with compaction, alluvial fans are significantly compacted by is mainly a function of amount of void and motorcycle traffic in creosote bush (Larrea tri- density of soil minerals. Webb and Wilshire dentata) shrublands of southern California (1980) suggest that severely compacted soils (Webb 1982, 1983, Webb et al. 1986). Motor- may require at least a century for natural re- cycle-induced compaction was studied using covery in southern Nevada. various numbers of passes (Davidson and Fox Soil compaction studies have been con- 1974, Wilshire and Nakata 1976, Webb 1982, ducted in Larrea tridentata–Atriplex spp. (creo- 1983, Webb et al. 1986). Off-road motor vehi- sote bush–salt bush), Larrea tridentata–Ambro- cles can cause significant compaction with as sia dumosa (creosote bush–white bursage), and few as 1 to 10 passes (Davidson and Fox 1974, Coleogyne shrublands in the Mojave Desert. Vollmer et al. 1976, Wilshire and Nakata 1976, Most studies have focused on the influence

1Department of Biology, WDB, Community College of Southern Nevada, 6375 West Charleston Boulevard, Las Vegas, NV 89146-1139.

125 126 WESTERN NORTH AMERICAN NATURALIST [Volume 64 that severe compaction can have on soil physi- tered distribution of other shrub species. Com- cal attributes but have ignored slight soil common associated woody species that occur in paction, such as that occurring from frequency this vegetation zone include Yucca brevifolia of human visits. No comparative data are avail- (Joshua tree), Ephedra nevadensis (Nevada able to quantify the disturbance effect of an ephedra), Menodora spinescens (spiny meno- average human footprint relative to a rolling dora), Thamnosma montana (turpentine bush), bike tire, a motorcycle tire, or the large wheel and Lycium andersonii (Anderson lycium). of a 4-wheeled vehicle. The objective of this The herbaceous species present are primarily study is to examine any significant differences members of the Asteraceae, Brassicaceae, that might exist in soil physical attributes at a Fabaceae, and Poaceae families. particular frequency of visits in each of the 4 Field Design and Sampling disturbance types (trampling, biking, motorcycle, and vehicle). In other words, do the To experimentally investigate recreation- effects of hiking and biking slowly increase induced compaction, my field assistants and I over time relative to the effects of motor vehi- created 4 trails—hiking, biking, motorcycle, cle traffic? and vehicle—with 1, 10, 100, and 200 passes in Kyle Canyon (Table 1). Before this study METHODS these 4 linear trails, each 100 m in length and separated from the others by 50 m, showed no Study Site clear evidence of compaction or other types of Southern Nevada, within the Basin and soil disturbance. The hiking trail was trampled Range geological province, is a region charac- by a 78-kg person in hiking boots, with a pass terized by annual weather extremes and sparse being 1 walk down the lane at a normal gait. vegetation cover (Brittingham and Walker 2000). The mountain bike, motorcycle, and 4-wheeled Most annual precipitation occurs between vehicle were operated by a 78-kg rider who October and April as frontal systems, with the maintained a constant speed (32 kph) to avoid remainder occurring in the summer as con- acceleration, braking, and turning effects. Soil vectional thunderstorms (Smith et al. 1995). measurements were taken at the pre-distur- Summer rains occur as brief, intense, and local- bance level in all 4 trails to ensure comparabil- ized events that are highly variable in both ity. Soil measurements were also made after 1, time and space. Prolonged mountain thunder- 10, 100, and 200 passes (post-disturbance level). storms in the summer can cause flash flooding Within each type of disturbance, I collected in the canyons and nearby washes. Winter rain- 160 soil samples, with 40 from each trail. For falls, on the contrary, are mild and widespread, each sample, I excavated soil in an area 10 cm in and can last up to several days. diameter and 10 cm deep. Intervals within each I conducted field studies in a Coleogyne trail were randomly selected to avoid biased shrubland during fall 2002 in Kyle Canyon sampling. Despite carefully controlling the tire (roughly 36°01′N, 115°09′W; elevation 1475 impact, some bicycle and motor vehicle passes m) on the eastern slope of the Spring Moun- did not overlap completely, leaving a rut approx- tains. Kyle Canyon was selected because soils imately 2 to 3 times the tire width. Hence, soil on its alluvial fans and bajadas are representa- measurements at any point on the trails were tive of a large area in southern Nevada. Soils likely to be underestimates. are calcareous, poorly developed (without dis- Soil samples were sieved through a 2-mm tinct soil horizons), and composed primarily of mesh to remove plant roots and rocks >2 mm weathered granite and limestone bedrock. Mul- in diameter. I performed all tests on <2 mm tiple caliche layers in the subsoil make this soils dried at 65°C for 72 hours, measuring area edaphically arid, which in turn contributes bulk density, percent pore space, moisture, to slow organic decomposition and soil forma- organic matter, and pH. tion. Soil textures near the surface are coarse Fresh soil cores of known volume were and sandy. Numerous rocks litter the terrain, carefully removed from the field and oven- and dry wash channels dissect the rocky slopes dried at 65°C until they reached a constant and alluvial fans. mass. Soil bulk density was estimated by di- The study area is dominated by a closely viding mass by volume. I measured soil com- spaced matrix of Coleogyne shrubs with a scat- paction in the field immediately after the impact 2004] HUMAN RECREATIONAL DISTURBANCE OF SOILS 127

TABLE 1. Description of shoes, bicycle, and motor vehicle used under controlled traffic study in the Coleogyne shrubland of Kyle Canyon. Disturbance type Brand Shoe/Tire width (cm) Gross weight (kg) Foot Outback Footwear 9 0.9 Bicycle Roadmaster Mountain Sport 5 17.3 Motorcycle Yamaha YZ 250 8 88.2 Vehicle Chevy Tahoe 22 3111.4

using a penetrometer, which was inserted into RESULTS the soil after removing stony surface pave- ments. The penetrometer readings were taken Differences in soil pH, texture, moisture, at the point where the cone base reached the temperature, and organic matter were not sig- soil surface (point depth = 3.8 cm). I calcu- nificant at the pre-disturbance level in all 4 lated average pore space using the equation: trails (P > 0.05; Table 2). Nevertheless, soil compaction, bulk density, and percent pore space were significantly different between pre- pore space (%) = 100 – (Db/Dp * 100), disturbance and post-disturbance levels in all 4 trails. Soil compaction and bulk density in- where Db is bulk density of the soil and Dp is average particle density, usually about 2.65 g ⋅ creased, while percent pore space decreased cc–1 (Hausenbuiller 1972, Davidson and Fox significantly, with type of disturbance and 1974). Soil particle size distribution was deter- with increasing frequency of visits when I ex- ≤ mined by the hydrometer method as de- amined these 2 factors alone (P 0.05; Table 3). scribed by Bouycoucos (1951). Soil moisture Soil compaction and bulk density were great- content was measured gravimetrically by com- est in the vehicle trail and least in the hiking puting differences between fresh and oven-dried trail (Table 3). mass. Using a soil thermometer, I recorded Mountain bikes and motor vehicles created soil temperature readings in the field at 15 cm trails that were visible immediately after initial impact. The soil surface, disrupted by the below the soil surface. Soil organic matter was impact, no longer had uniform gravel cover. A computed by mass loss on ignition at 550°C single mountain bike or motor vehicle pass left for 4 hours. Soil pH was determined by definite tire imprints with a slight indentation preparing a 1:1 paste of dry soil:distilled water on the soil surface. Tire imprints became more and by measuring the mixture with an elec- evident after 10 passes and were up to 3 cm trode pH meter. below the level of adjacent undisturbed soil Statistical Analyses after 200 passes. I observed significant differences in soil One-way analysis of variance (ANOVA; compaction, bulk density, and percent pore Analytical Software 1994), followed by Tukey’s space between 1 pass and 100 passes (P ≤ multiple comparison test, was performed to 0.05). Yet, differences between 100 and 200 compare (1) differences in soil attributes (pH, passes in all 4 disturbance types were not sig- moisture, temperature, texture, and organic nificant (P > 0.05; Table 3). Soil bulk density matter) and (2) mean values among the 4 trails increased significantly from 1.30 to 1.51 g ⋅ at the pre-disturbance level. Two-way ANOVA cm–3 after 100 passes, but increased only slight- (Analytical Software 1994) was computed on ly from 1.51 to 1.55 g ⋅ cm–3 after 200 passes soil compaction, bulk density, and percent (Table 2). pore space, with type of disturbance (hiking, In each of the 4 disturbance types, I detected biking, motorcycle, and vehicle) and frequency a significant difference in soil physical charac- of visits (1, 10, 100, and 200 passes) as main teristics at a particular frequency of visits. On effects. Mean values were presented with average a single vehicle pass was equivalent to standard errors, and statistical significance 10 human footprints, and the ratio remained was tested at P ≤ 0.05. constant as the frequency increased to 10 128 WESTERN NORTH AMERICAN NATURALIST [Volume 64

± TABLE 2. Soil properties (mean sx–; n = 40 per treatment per characteristic) of the 4 trails at the pre-disturbance level in the Coleogyne shrubland of Kyle Canyon. Soil temperatures were measured at 10-cm depth. No significant differences were detected in all soil properties among the 4 trails (P > 0.05). Soil property Hiking Biking Motorcycle riding Vehicle driving Soil texture Sand 70.7 ± 3.1 70.9 ± 3.0 69.3 ± 2.8 69.5 ± 2.7 Silt 19.7 ± 1.9 18.3 ± 1.6 19.7 ± 2.0 19.9 ± 2.1 Clay 9.6 ± 0.7 11.8 ± 0.6 11.0 ± 0.5 10.6 ± 0.6 Moisture (%) 2.9 ± 0.4 3.0 ± 0.5 2.8 ± 0.5 3.0 ± 0.5 Temperature (°C) 23.8 ± 0.4 23.1 ± 0.5 23.2 ± 0.3 23.9 ± 0.3 Organic matter (%) 3.4 ± 0.3 3.5 ± 0.3 3.5 ± 0.4 3.4 ± 0.5 pH 7.5 ± 0.1 7.6 ± 0.1 7.6 ± 0.1 7.5 ± 0.5

± TABLE 3. Changes in bulk density, compaction, and percent pore space (mean sx–; n = 40 per treatment per characteristic) of the upper 10 cm of soil in response to various levels of trampling, mountain bike and motorcycle riding, and 4-wheeled vehicle driving in the Coleogyne shrubland of Kyle Canyon. Disturbance Number of Bulk density Compaction Pore space type passes (g ⋅ cm–3) (kg ⋅ cm–2) (%) Human foot 0 1.30 ± 0.07 6.0 ± 0.3 50.9 ± 2.9 11.33 ± 0.06 6.1 ± 0.2 49.8 ± 2.9 10 1.39 ± 0.06 6.4 ± 0.2 47.5 ± 2.8 100 1.51 ± 0.05 7.0 ± 0.2 43.0 ± 2.7 200 1.55 ± 0.05 7.3 ± 0.1 41.5 ± 1.9 Mountain bike 0 1.31 ± 0.07 6.0 ± 0.3 50.6 ± 3.2 11.35 ± 0.08 6.2 ± 0.2 49.9 ± 2.9 10 1.43 ± 0.06 6.6 ± 0.2 46.0 ± 2.7 100 1.55 ± 0.06 7.2 ± 0.1 41.5 ± 2.2 200 1.58 ± 0.06 7.3 ± 0.1 40.4 ± 2.3 Motorcycle 0 1.33 ± 0.07 6.1 ± 0.2 50.2 ± 3.3 11.38 ± 0.07 6.4 ± 0.3 47.9 ± 2.9 10 1.46 ± 0.06 6.8 ± 0.2 44.9 ± 2.7 100 1.57 ± 0.05 7.4 ± 0.1 40.8 ± 2.7 200 1.59 ± 0.06 7.4 ± 0.1 40.4 ± 2.7 Vehicle 0 1.30 ± 0.09 6.0 ± 0.3 50.9 ± 3.3 11.39 ± 0.07 6.4 ± 0.2 47.5 ± 2.7 10 1.51 ± 0.06 7.0 ± 0.2 43.0 ± 2.8 100 1.62 ± 0.06 7.5 ± 0.1 38.9 ± 2.5 200 1.63 ± 0.06 7.5 ± 0.1 38.5 ± 2.4

vehicle passes and 100 footprints. The same tween disturbance type and visit frequency for number of passes made by a mountain bike or soil bulk density, compaction, and percent motorcycle resulted in levels of bulk density pore space (P > 0.05; Table 4). and soil compaction intermediate to the other 2 (Table 3). Soils became significantly com- DISCUSSION pacted after a single pass in the vehicle trail and after 10 passes in the other 3 trails. I studied the interactive effects of distur- The proportional extent of impact and the bance type and visit frequency on soil physical statistical variability of soil compaction, bulk attributes in a Coleogyne shrubland of south- density, and percent pore space decreased ern Nevada to quantify soil compaction under with increasing number of passes. Differences controlled traffic conditions. Human recre- between biking and motorcycle trails, and ational activities significantly increased soil between 100 and 200 passes in all 4 trails, compaction and soil bulk density, but decreased were not significant (P > 0.05; Table 3). More- percent pore space. The greatest effects oc- over, significant interaction was not found be- curred during the first few passes, especially 2004] HUMAN RECREATIONAL DISTURBANCE OF SOILS 129

TABLE 4. Results of 2-way ANOVA showing effects of disturbance type, visit frequency, and their interactions on soil bulk denstiy, compaction, and percent pore space. For disturbance type df = 3, for visit frequency df = 4, and for disturbance type * visit frequency combination df = 12.

Source of variation ______Bulk density ______Compaction ______Pore space FP FP FP Disturbance type 14.41 0.0000 3.93 0.0244 3.45 0.0373 Visit frequency 163.95 0.0000 117.30 0.0000 38.21 0.0000 Disturbance * Frequency 1.51 0.2036 1.61 0.1696 0.27 0.9885

in vehicle trails, with changes per pass decreas- Results of this study illustrate the damage ing as the number of passes increased in all 4 that recreational activities can do to dry desert trails. soils in Coleogyne shrubland. Such activities Soil texture is a major factor in determining cause soil disturbance through disruption and the magnitude of bulk density increases under compaction of surface soil. Soil compaction, applied pressure or loading. Unlike playas and which adversely impacts various soil attributes, sand dunes with mixtures of equal-sized parti- is one aspect of land degradation. Given the cles, many soils of the Mojave Desert are highly fragility of desert soils, random human tram- susceptible to soil compaction. Poorly sorted pling and off-road motor vehicle traffic can soils, particularly those with a loamy sand or severely compact soils across large areas over sandy loam texture with abundant gravel in time. Despite growing evidence that hiking, the profile, are most vulnerable to soil com- biking, and relentless motor vehicle use are paction, and these soils are most common in very damaging to fragile desert soils, the pub- bajadas and alluvial fans of the Mojave Desert lic continues to use Kyle Canyon and other (Webb 1982). areas of the Mojave Desert for recreational Both the proportional extent of impact and activities. It is imperative to educate visitors the statistical variability of soil compaction, about the ecological consequences of soil com- bulk density, and percent pore space decreased paction. With this knowledge, they may more with increasing numbers of passes. Significant likely voluntarily minimize soil compaction by differences were detected between 1 pass and staying within established trails. 100 passes, but differences between 100 passes and 200 passes were not significant in any of ACKNOWLEDGMENTS the 4 disturbance types. This study agrees with Webb’s (1982) study, indicating that soil The author gratefully acknowledges Steven bulk density increases in the upper 6 cm and Lei, David Valenzuela, and Shevaun Valenzuela decreases in proportion to the inverse of num- for valuable field and laboratory assistance. ber of passes. Thus, greatest bulk density in- Steven Lei assisted with statistical analyses. creases and related property changes per pass Critical review provided by David Charlet will occur during the initial few passes (Webb greatly improved the manuscript. The Depart- 1982). Soil compaction is a product of in- ment of Biology at the Community College of creased bulk density and decreased pore space. Southern Nevada (CCSN) provided logistical The degree of soil compaction is a function of support. disturbance type and visit frequency when examining these 2 factors independently. How- LITERATURE CITED ever, interactive effects of disturbance type and visit frequency on soil bulk density, com- ANALYTICAL SOFTWARE 1994. Statistix 4.1, an interactive statistical program for microcomputers. Analytical paction, and percent pore space were not sta- Software, St. Paul, MN. tistically significant. In this study additional BOUYCOUCOS, G.J. 1951. A recalibration of the hydrometer passes caused no significant further compres- method for making mechanical analysis of soils. sion because soils were already compacted. Journal of Agronomy 43:434–438. BRITTINGHAM, S.B., AND L.R. WALKER. 2000. Facilitation Effects of hiking and biking slowly increased of Yucca brevifolia recruitment by Mojave Desert over time relative to effects of motor vehicle shrubs. Western North American Naturalist 60: traffic. 374–383. 130 WESTERN NORTH AMERICAN NATURALIST [Volume 64

DAVIDSON, E., AND M. FOX. 1974. Effects of off-road motor- WEBB, R.H. 1982. Off-road motorcycle effects on a desert cycle activity on Mojave Desert vegetation and soil. soil. Environmental Conservation 9:197–208. Madrono 22:381–390. ______. 1983. Compaction of desert soils by off-road vehi- GILL, W.R., AND G.E. VANDENBERG. 1967. Soil dynamics cles. Pages 51–79 in R.H. Webb and H.G. Wilshire, in tillage and traction. USDA Agricultural Research editors, Environmental effects of off-road vehicles. Service, Agricultural Handbook 316. Springer-Verlag, New York. HAUSENBUILLER, R.L. 1972. Soil science. William C. Brown ______. 2002. Recovery of severely compacted soils in the Company Publishers, Dubuque, IA. Mojave Desert, California, USA. Arid Land Research MARBLE, J.R. 1985. Techniques of revegetation and recla- and Management 16:291–305. mation of land damaged by off-road vehicles in the WEBB, R.H., AND H.G. WILSHIRE. 1980. Recovery of soils Lake Mead Recreational Area. Cooperative National and vegetation in a Mojave Desert ghost town, Ne- Park Resources Studies Unit, University of Nevada, vada, USA. Journal of Arid Environments 3:291–303. Las Vegas. WEBB, R.H., J.W. STEIGER, AND H.G. WILSHIRE. 1986. SMITH, S.D., C.A. HERR, K.L. LEARY, AND J.M. PIORKOWSKI. Recovery of compacted soils in Mojave Desert ghost 1995. Soil-plant water relations in a Mojave Desert towns. Soil Science Society of America Journal 50: mixed shrub community: a comparison of three geo- 1341–1344. morphic surfaces. Journal of Arid Environments 29: WILSHIRE, H.G., AND J.K. NAKATA. 1976. Off-road vehicle 339–351. effects on California’s Mojave Desert. California VOLLMER, A.T., B.G. MAZA, P.A. MEDICA, F.B. TURNER, AND Geology 29:123–132. S.A. BAMBERG. 1976. The impact of off-road vehicles on a desert ecosystem. Environmental Management Received 8 May 2002 1:1–13. Accepted 7 March 2003 WEAVER, T., AND D. DALE. (1978). Trampling effects of hikers, motorcycles, and horses in meadows and forests. Journal of Applied Ecology 15:451–457. Western North American Naturalist 64(1), ©2004, pp. 131–134

GROUNDWATER INVERTEBRATE FAUNA OF THE BEAR RIVER RANGE, UTAH

Candace Brindza Huebner1 and Mark R. Vinson1

Key words: groundwater, aquatic invertebrates, Stygobromus, Utah.

The chance discovery of an undescribed Sampling locations included a stream flow- species of blind subterranean amphipod within ing within Logan Cave and 27 spring outflows the Stygobromus hubbsii group in a cave in (Table 1). These springs generally flow directly northern Utah prompted us to question the or indirectly from fractures, faults, joints, and possible distribution and occurrence of addi- solution channels in carbonate rocks (Bjork- tional groundwater fauna in this area. Ground- lund and McGreevy 1971). Sixty spring sites water biota in western North America have were initially identified for sampling from not been particularly well studied (cf. Stanford topographic maps, but after visiting each spring and Gaufin 1974, Ward 1977, Pennak and Ward we excluded many from the study because of 1986, Stanford and Ward 1988, Stanford et al. lack of water or extensive human alteration, 1994, Ward and Voelz 1994, Ward et al. 1994, e.g., dewatering caused by construction of live- Drost et al. 1997, Canton and Chadwick 2000), stock watering ponds. Most studies of ground- especially when compared with studies of west- water fauna have been based on collections ern surface water systems or groundwater sys- from either wells or caves, but spring outflows tems of eastern North America, where karst top- also can be used as collection sites of ground- ography is more prevalent overall and less dis- water organisms (Nielsen 1950, Erman et al. junct (Ritter 1986, also see Culver 1997 [http:// 1995). www.utexas.edu/depts/tnhc/.www/biospeleol- We sampled aquatic invertebrates at 5 loca- ogy/uskarst.jpg]). The objective of our study tions in fall 1998, and all 28 sites were sampled was to follow up on this chance find by more at various times between May and November intensively sampling the aquatic macroinver- 1999. The stream within Logan Cave was sam- tebrates present in this area of northern Utah. pled at several locations approximately 400– This study was conducted in Cache Valley 1000 m from the cave entrance. Springs were and the Bear River Mountain Range in Cache sampled at their point of discharge. Within County, Utah. Cache Valley is an elongated Logan Cave we set 3 drift nets (10-cm diame- graben formed by normal high-angle faults ter, 250-micron mesh) in the stream at approx- typical of basins in the Basin and Range Pro- imately mid-depth. At each spring location, a vince (Fenneman 1931). The Bear River Moun- short length (0.5 m) of 10- or 16-cm-diameter tains, which are immediately east of Cache polyvinyl chloride (PVC) pipe, depending on Valley, are composed of sedimentary and meta- spring outflow volume, was driven about 0.3 morphic rock of Permian to Precambrian age, m into the hillside where the spring exited the including limestone, dolomite, quartzite, sand- ground. These pipes limited collection of down- stone, mudstone, siltstone, and shale (Kariya stream surface water organisms by capturing et al. 1994). Groundwater in Cache Valley and 100% of the flow at each spring. A 250-micron the Bear River Mountain Range, originating nylon mesh bag installed over the pipe outlet from infiltration of precipitation and seepage filtered all water exiting each spring. Nets were from streams, is considered a shallow and un- left in place at all sites for 1 week, after which confined, unglaciated karst system (Kariya et all material collected in the nets was preserved al. 1994). in 70% ethanol and returned to the laboratory.

1Department of Aquatic, Watershed, and Earth Resources, Utah State University, Logan, UT 84322-5210.

131 132 WESTERN NORTH AMERICAN NATURALIST [Volume 64

TABLE 1. Location of sampling sites within the Bear River Range and number of Stygobromus collected at each site. Latitude Longitude Stygobromus Sampling location (N) (W) County, State collected Logan Cave, 3 sampling locations 41.8000 111.6250 Cache County, Utah 28 Beaver Mountain Spring 41.9611 111.5342 Cache County, Utah 5 Beaver Spring 41.9545 111.5857 Cache County, Utah 1 Box Spring 41.7265 111.6240 Cache County, Utah Card Canyon Spring 41.7530 111.6520 Cache County, Utah 1 Coldwater Spring 41.7178 111.5998 Cache County, Utah Coldwater Spring by Tony Grove 41.8822 111.6462 Cache County, Utah Franklin Basin, unnamed spring # 2 41.9615 111.5952 Cache County, Utah Franklin Basin, unnamed spring # 3 41.9745 111.5987 Cache County, Utah 3 Grey Cliff Spring 41.6731 111.6173 Cache County, Utah Hodges Spring 41.9806 111.4851 Cache County, Utah Millville Srping 41.6587 111.7698 Cache County, Utah Mosslander Spring 41.6578 111.6199 Cache County, Utah 7 North Cheney Spring 41.8930 111.4683 Cache County, Utah Pig Hole Spring 41.6543 111.6355 Cache County, Utah 4 Pine Spring 41.7288 111.6172 Cache County, Utah Pleasant Val Spring 41.6477 111.5787 Cache County, Utah Pole Hole Spring 41.6590 111.5713 Cache County, Utah Porcupine Reservoir Spring 41.5190 111.7545 Cache County, Utah 2 Rock Spring 41.7863 111.5395 Cache County, Utah Sadduccee Spring 41.0513 111.4600 Rich County, Utah Sow Hole Spring 41.6561 111.6068 Cache County, Utah Spring at Spring Campground 41.6333 111.6500 Cache County, Utah Spring Hollow 41.7572 111.7188 Cache County, Utah St. Charles Spring 42.1120 111.4473 Bear Lake County, Idaho 4 Upper Sow Hole Spring 41.6560 111.6047 Cache County, Utah 1 Wind Cave 41.7625 111.7035 Cache County, Utah Worm Fence Spring 41.8597 111.5500 Cache County, Utah 1

All organisms from each sample were removed From the 28 sites, we collected and identi- and identified to the lowest possible taxo- fied 6443 individuals and 44 genera of aquatic nomic level. Dr. John Holsinger (Old Domin- macroinvertebrates (Table 2). Most individuals ion University, VA) verified amphipod identifi- were small and immature, which made identi- cations. fications beyond family and genera difficult if Habitat conditions in Logan Cave and at not impossible. Invertebrate genera richness the point where samples were collected at the varied from 2 to 17 at individual sites. No aqua- spring sites were similar. Substrate consisted tic vertebrates were collected. We did, how- of gravel with some silt deposits. Organic mat- ever, collect an undescribed species within the ter was present at all sites and consisted pri- Stygobromus hubbsii group (Dr. John Holsinger marily of roots, small branches, and fine debris. personal communication) from 12 spring sites and Logan Cave. The most frequently col- Organic matter within Logan Cave was visibly lected organisms were Chloroperlidae, Cap- less than that found downstream of the collec- niidae, Collembola, Dytiscidae, Ameletus, and tion points at the spring sites, as there were no Stygobromus hubbsii group. All taxa collected larger pieces of wood or moss growing within in Logan Cave also were collected at spring the cave. Water temperatures were highly outflows. similar among all sites and ranged from 5°C to The invertebrate fauna we collected is sim- 7°C. Ranging from 7.1 to 8.4, pH was also ilar to that collected in Roaring Springs Cave, fairly similar among sites. Specific conduc- Grand Canyon, Arizona, by Drost and Blinn tance within Logan Cave was 315 µS cm–1 (1997). Stygobromus is the only obligate ground- near the mouth of the cave and 359 µS cm–1 water fauna we collected; the other taxa are deeper into the cave. Specific conductance at considered occasional hyporheic dwellers (Boul- spring sites was 160–525 µS cm–1 and aver- ton 2000) and are commonly found in the aged 305 µS cm–1. Logan River (Vinson unpublished data). Since 2004] NOTES 133

TABLE 2. List of taxa and number of individuals collected in Logan Cave (indicated by an *) and 28 spring outflows in the Bear River Mountain Range, northern Utah. Individuals Individuals Taxon collected Taxon collected Annelida Ephemeroptera Oligochaeta 95 Ameletidae Arthropoda Ameletus*20 Arachnida Baetidae 36 Trombidiformes* 673 Baetis 901 Entognatha Ephemerellidae 4 Collembola* 37 Heptageniidae, unidentified 9 Diplura* 2 Cinygmula 5 Insecta Epeorus 3 Coleoptera 3 Hemiptera Curculionidae 2 Gerridae Dytiscidae, unidentified 9 Aquarius 1 Agabus 75 Lepidoptera, unidentified 1 Deronectes 4 Plecoptera Hydroporus 311 Capniidae, unidentified* 35 Hydrovatus 29 Chloroperlidae Oreodytes 1 Sweltsa*32 Elmidae Nemouridae, unidentified 54 Heterlimnius 22 Malenka 134 Narpus 3 Zapada 260 Ordobrevia 1 Perlidae Stenelmis 2 Hesperoperla pacifica 34 Hydraenidae Perlodidae 215 Hydraena 1 Isoperla 761 Hydrophilidae Trichoptera 192 Ametor 10 Hydroptilidae, unidentified 13 Paracymus 7 Limnephilidae, unidentified 12 Diptera 2 Dicosmoecus 1 Ceratopogonidae, unidentified 5 Hesperophylax 480 Probezzia 3 Psychoglypha 48 Chironomidae, unidentified 1006 Rhyacophilidae Dixidae Rhyacophila 5 Dixa 19 Uenoidae Empididae Neophylax 8 Clinocera 4 Neothremma 296 Ephydridae 1 Oligophlebodes 1 Muscidae 19 Malacostraca Psychodidae Amphipoda Pericoma 2 Crangonyctidae Simuliidae Stygobromus hubbsi group* 52 Simulium 58 Mollusca Stratiomyidae Bivalvia Caloparyphus 10 Veneroida Euparyphus 9 Pisidiidae Tabanidae Pisidium 3 Tabanus 2 Gastropoda Tipulidae, unidentified 1 Basommatophora Dicranota 17 Lymnaeidae, unidentified 9 Limonia 3 Planorbidae, unidentified 8 Ormosia 10 Neotaenioglossa Pedicia 1 Hydrobiidae, unidentified 18 Tipula 2 Platyhelminthes Turbellaria 331 134 WESTERN NORTH AMERICAN NATURALIST [Volume 64 this study was performed, we also have col- Journal of the Kansas Entomological Society 68: lected Stygobromus in southern Utah (Alvey 50–64. FENNEMAN, N.M. 1931. Physiography of the western United Wash, Garfield County). These discoveries were States. McGraw-Hill, New York. 534 pp. predicted by Holsinger (1974) and echoed by HOLSINGER, J.R. 1974. Systematics of the subterranean Ward (1977) and Canton and Chadwick (2000). amphipod genus Stygobromus (Gammaridae), part I: The occurrence and discovery of additional species of the western United States. Smithsonian species of Stygobromus in other areas of the Contributions to Zoology 160:1–63. KARIYA, K.A., D.M. ROARK, AND K.M. HANSON. 1994. western United States are likely as these areas Hydrology of Cache Valley, Cache County, Utah, and are sampled in the future. adjacent part of Idaho, with emphasis on simulation of ground-water flow. Utah Department of Natural We thank Dr. John Holsinger for examina- Resources Technical Publication 108. 120 pp. NIELSEN, A. 1950. On the zoogeography of springs. Hydro- tion of the Stygobromus specimens. Charles biologia 2:313–320. Hawkins offered valuable suggestions regard- PENNAK, R.W., AND J.V. WARD. 1986. Interstitial faunal ing the study and the manuscript. We also communities of the hyporheic and adjacent ground- thank B.C. Kondratieff for reviewing the man- water biotopes of a Colorado mountain stream. uscript. Financial support was provided by Utah Archiv für Hydrobiologie Supplement 74:356–396. RITTER, D.F. 1986. Process geomorphology. 2nd edition. State University. Wm. C. Brown Publishers, Dubuque, IA. 579 pp. STANFORD, J.A., AND R. GAUFIN. 1974. Hyporheic commu- LITERATURE CITED nities of two Montana rivers. Science 18:700–702. STANFORD, J.A., AND J.V. WARD. 1988. The hyporheic habi- BJORKLUND, L.J., AND L.J. MCGREEVY. 1971. Ground-water tat of river ecosystems. Nature 335:64–66. resources of Cache Valley, Utah and Idaho. Utah STANFORD, J.A., ET AL. 1994. Alluvial aquifers of the Flat- Department of Natural Resources Technical Publi- head River, Montana. Pages 367–390 in J. Gilbert, cation 36. 72 pp. D.L. Danielopol, and J.A. Stanford, editors, Ground- BOULTON, A. 2000. The subsurface macrofauna. Pages 337– water ecology. Academic Press, San Diego, CA. 361 in J. Jones and P. Mulholland, editors, Streams WARD, J.V. 1977. First records of subterranean amphipods and groundwaters. Academic Press, San Diego, CA. from Colorado with descriptions of three new species CANTON, S.P., AND J.W. CHADWICK. 2000. Distribution of of Stygobromus (Crangonyctidae). Transactions of the the subterranean amphipod Stygobromus in central American Microscopical Society 96:452–466. Colorado streams, with notes on the interstitial com- WARD, J.V., AND N.J. VOELZ. 1994. Groundwater fauna of munity. Western North American Naturalist 60:130– the South Platte River system, Colorado. Pages 391– 138. 423 in J. Gilbert, D.L. Danielopol, and J.A. Stanford, CULVER, D. 1997. Color map of the karst regions of the editors, Groundwater ecology. Academic Press, San USA. Presented at the Conference on the Conserva- Diego, CA. tion and Protection of the Biota of Karst, Nashville, WARD, J.V., J.A. STANDFORD, AND N.J. VOELZ. 1994. Spatial TN. patterns of Crustacea in the floodplain aquifer of an DROST, C.A., AND D.W. BLINN. 1997. Invertebrate com- alluvial river. Hydrobiologia 287:11–17. munity of Roaring Springs Cave, Grand Canyon National Park, Arizona. Southwestern Naturalist 42: Received 22 January 2002 497–500. Accepted 28 March 2003 ERMAN, N.A., AND D.C. ERMAN. 1995. Spring permanence: Trichoptera species richness, and the role of drought. Western North American Naturalist 64(1), ©2004, pp. 135–136

COLLECTION OF AN ADULT GIZZARD SHAD (DOROSOMA CEPEDIANUM) FROM THE SAN JUAN RIVER, UTAH

Gordon A. Mueller1 and Jim L. Brooks2

Key words: gizzard shad, Dorosoma cepedianum, stocking contamination, range expansion.

We collected an adult gizzard shad (Doro- cooling water for the Four Corners Power Plant soma cepedianum) from the San Juan River located near Farmington, New Mexico. While just upstream of Lake Powell, Utah, on 6 June the reservoir is physically separated from the 2000. This represents the first documented San Juan River, the power plant periodically occurrence of the species in the Colorado flushes the cooling system into a wash that River or its tributaries. The adult male (35 cm empties into the river (Howard Bradley, Ari- TL, 470 g) was taken by trammel net from a zona Public Service, personal communication). small (0.5 ha), shallow (<2 m) backwater along The spread of gizzard shad poses a compet- with several other fish that included 3 endan- itive and predatory threat to native and recre- gered razorback sucker (Xyrauchen texanus). ational fish communities throughout the Colo- The specimen is stored at the Museum of rado River basin. The San Juan River Recovery Southwestern Biology, University of New Mex- Program has stocked large numbers of razor- ico, Albuquerque (curation number 49122). back suckers in an attempt to reestablish river Utah Division of Wildlife (UDOW) intensi- populations (Ryden 1997). Razorback sucker fied their shad surveys in the San Juan arm of and gizzard shad are planktivores (Marsh 1987, Lake Powell as a result of this initial collec- Pflieger 1997) that prefer highly productive tion. No additional gizzard shad were found habitats (Robinson and Buchanan 1988, Mueller during 2001, but 8 young-of-year gizzard shad et al. 2000), and capture of both species at the (x– = 103 mm TL) were taken collectively dur- same site suggests they may compete for simi- ing August and September 2002 (Blommer lar physical and biological resources. If giz- and Gustaveson 2002). The fish were taken 12 zard shad expand further upstream, they could km downstream from the original collection invade floodplain nurseries that are believed site, indicating that gizzard shad are not only critical to the recovery of the razorback sucker present but are successfully producing young. (Wydoski and Wick 1998). While gizzard shad Source of the gizzard shad is unknown. are seldom considered active predators, adults Angler (live bait) introduction is possible but feed on benthic insects (Robinson and Buchanan unlikely because of the fragile nature of the 1988) which may contain larval-native fish. species. More likely, the gizzard shad escaped The spread of nonindigenous fish in the Morgan Lake where it was probably intro- desert Southwest has devastated and, in many duced when largemouth bass were stocked cases, totally eliminated native fish communi- (Brooks et al. 2000). Stocked fish were sup- ties (Minckley and Deacon 1991, Fuller et al. plied by Inks Dam National Fish Hatchery 1999, Tyus and Saunders 2000). The collection (IDNFH) where subsequent shipments were of yet another nonnative reemphasizes the found (by coauthor) to contain gizzard shad in need for diligent inspections of all fish ship- addition to 9 other nontargeted species. Large ments to minimize the risk of further jeopar- numbers of gizzard shad were first observed dizing remaining communities and recovery in Morgan Lake in 1996. The reservoir provides efforts.

1U.S. Geological Survey, Box 25007, Denver, CO 80225. 2U.S. Fish and Wildlife Service, 2105 Osuna Road SE, Albuquerque, NM 87113.

135 136 WESTERN NORTH AMERICAN NATURALIST [Volume 64

LITERATURE CITED razorback sucker (Xyrauchen texanus) in a lower Col- orado River reservoir, Arizona-Nevada. Western North BLOMMER, G.L., AND A.W. GUSTAVESON. 2002. Life history American Naturalist 60:180–187. and population dynamics of threadfin shad in Lake PFLIEGER, W.L. 1997. The fishes of Missouri. Revised edi- Powell 1968–2001. Utah Department of Natural Re- tion. Missouri Department of Conservation, Jeffer- sources, Division of Wildlife Resources Publication son City. 02-23, Salt Lake City. 78 pp. ROBINSON, H.W., AND T.M. B UCHANAN. 1988. Fishes of BROOKS, J.E., M.J. BUNTJER, AND J.R. SMITH. 2000. Non- Arkansas. University of Arkansas Press, Fayetteville. native species interactions: management implication 536 pp. to aid in the recovery of the Colorado pikeminnow RYDEN, D.W. 1997. Five-year augmentation plan for razor- (Ptychocheilus lucius) and razorback sucker (Xyrau- back sucker in the San Juan River. U.S. Fish and chen texanus) in the San Juan River, CO-NM-UT. Wildlife Service, Colorado River Fishery Project, San Juan River Basin Recovery Implementation Grand Junction, CO. Program, U.S. Fish and Wildlife Service, Albuquer- TYUS, H.M., AND J.F. SAUNDERS, III. 2000. Nonnative fish que, NM. control and endangered fish recovery: lessons from FULLER, P.L., L.G. NICO, AND J.D. WILLIAMS. 1999. Non- the Colorado River. Fisheries, American Fisheries indigenous fishes introduced into inland waters of Society 25(9):17–24. the United States. American Fisheries Society Spe- WYDOSKI, R.S., AND E.J. WICK. 1998. Ecological value of cial Publication 27, Bethesda, MD. floodplain habitats to razorback suckers in the upper MARSH, P.C. 1987. Food of adult razorback sucker in Lake Colorado River basin. Upper Colorado River Basin Mohave, Arizona-Nevada. Transactions of the Amer- Recovery Program, U.S. Fish and Wildlife Service, ican Fisheries Society 116:117–119. Denver, CO. 55 pp. MINCKLEY, W.L., AND J.E. DEACON, EDITORS. 1991. Battle against extinction: native fish management in the Received 17 July 2002 American west. University of Arizona Press, Tucson. Accepted 14 December 2002 MUELLER, G., P.C. MARSH, G. KNOWLES, AND T. W OLTERS. 2000. Distribution, movements, and habitat use of Western North American Naturalist 64(1), ©2004, pp. 137–140

GROWTH IN SONORAN DESERT POPULATIONS OF THE COMMON CHUCKWALLA (SAUROMALUS OBESUS)

Brian K. Sullivan1, Matthew A. Kwiatkowski1, and Paul S. Hamilton2

Key words: chuckwalla, growth, Sauromalus, Sonoran Desert.

The common chuckwalla, Sauromalus obe- archies; rather, males were territorial regard- sus (= ater; Hollingsworth 1998), is a large, less of density. However, it is unclear whether herbivorous lizard found throughout the Sono- these density differences are associated with ran and Mojave Deserts of the southwestern variation in other life history characters. We United States and northwestern Mexico (Hol- report on variation in growth rates of males lingsworth 1998). Throughout their range and females of 2 populations in the Phoenix chuckwallas are limited to rocky outcrops typ- area that vary dramatically in density. ically associated with isolated mountain ranges, We established study sites for 2 populations probably because of their unique antipredator of Sauromalus near Phoenix, Arizona, in the behavior whereby they retreat into rock crevices north central Sonoran Desert. The South Moun- and inflate their bodies to prevent removal by tains (“South”) study site (2.0 ha) was immedi- predators (Berry 1974, Hollingsworth 1998). ately south of Phoenix in the west central por- Population subdivision resulting from their tion of the city park surrounding the South patchy distribution might be expected to lead Mountains (33°20′26″N, 112°02′48″W), and to genetic (Lamb et al. 1992) or morphological the Lookout Mountain (“Lookout”) study site (Hamilton 1995) variation. Although dietary (10.4 ha) was in the southwestern section of (Johnson 1965, Nagy 1973) and life history the Lookout Mountain Preserve in the Phoenix (Abts 1987) investigations have been conducted Mountains of northern Phoenix (33°37′22″N, for chuckwalla populations in California, Ari- 112°03′12″W), Maricopa County, Arizona. zona populations remain little known (but see Because plant productivity is thought to Prieto and Sorenson 1977, Prieto and Ryan influence growth in chuckwallas (Berry 1974, 1978). Chuckwalla populations vary considerably Case 1976), rainfall data for the 2 sites were in body size (Case 1976, Tracy 1999) and growth obtained from weather recording stations (Ari- rates (Johnson 1965, Berry 1974, Abts 1987). zona Meteorological Network) at Laveen (South) Additionally, a recent study documented sub- and north Phoenix (Lookout) for 1994–1999. stantial variation in density among populations Each weather station is within 10 km of the of chuckwallas in central Arizona (Kwiatkow- respective study site. ski and Sullivan 2002). Populations in the All chuckwallas encountered on study sites Sonoran Desert near Phoenix, Arizona, range were uniquely and permanently marked by from densities typically found in the Mojave toe-clips. If more than 1 toe had to be removed, Desert of California (i.e., 7–23 individuals ⋅ no more than 1 toe per limb was clipped. No ha–1) to the lowest (3 individuals ⋅ ha–1) and chuckwalla had more than 3 toes clipped and highest (65 individuals ⋅ ha–1) densities yet re- only 2 had more than 2 toes clipped. Individu- corded (Sullivan and Flowers 1998). Contrary als were also marked with paint on the tip of to a hypothesis of Berry (1974), changes in the tail to facilitate identification without cap- population density were not associated with ture. Most animals were initially marked in changes in male behavior; males in high-den- 1995 and 1996 and were recaptured to deter- sity populations did not form dominance hier- mine growth in 1997–1999.

1Department of Life Sciences, Arizona State University West, Box 37100, Phoenix, AZ 85069. 2Department of Biology, Arizona State University–Tempe, Tempe, AZ 85287.

137 138 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Snout-vent length (SVL) and tail length TABLE 1. Rainfall (mm) by year and season (W = Octo- (mm) were measured by pressing the lizard ber–April, S = May–September) for the 2 study sites. Lookout = 19th Ave. and Greenway Road; South = flat against a metal or plastic rule. Growth was Laveen. calculated as follows: recapture SVL—initial SVL / years between captures. Because some Year Season Lookout South processing required off-site activity (e.g., 1995 W 167 135 attachment of radio tags; Kwiatkowski and S3426 1996 W 80 69 Sullivan 2002), some lizards were removed S 105 49 from the site on the day of capture. All such 1997 W 33 33 individuals were released at capture site S3364 within 24 hours. 1998 W 190 152 Nineteen individuals were recaptured dur- S7649 1999 W 85 52 ing a total of 38 occasions at the Lookout site, S84123 while 15 individuals were recaptured on 20 occasions at the South site; on average, 2.5 years elapsed between captures (Fig. 1). Because juveniles exhibit high growth rates (Johnson 1965, Berry 1974), and because some samples regions received somewhat unequal rainfall, had juveniles while others did not, only sexu- plant density was higher in the South site rela- ally mature individuals (initial SVL >135 mm; tive to the Lookout site over the course of this Johnson 1965, Hamilton 1995) were used for study (Kwiatkowski and Sullivan 2002). between-population comparisons (e.g., males Our results indicate that growth of male of South versus males of Lookout). All individ- chuckwallas of the South Mountains exceeded uals (i.e., juveniles and adults) were used for that of males from the Lookout site in the analysis of within-population sample parame- nearby Phoenix Mountains. The higher growth ters (e.g., asymptotic growth with South males), of males at the South site may be the result of and nonparametric tests were used, given the an interaction between resource availability small sample sizes and non-normality of growth and energetic costs of defending territories. data. Despite higher rainfall at Lookout, food re- Sample sizes were small, but growth of sources were considerably higher at the South females from South (n = 4; 1.7 mm ⋅ yr–1; range site over the course of this study, and males = 0–8 mm ⋅ yr–1) and Lookout (n = 10; 1.4 mm had home ranges 6 times smaller than home ⋅ yr–1; range = 0–9.7 mm ⋅ yr–1) was similar, as ranges of males at the Lookout site (Kwiat- was that for males of Lookout (n = 8; 1.4 mm ⋅ kowski and Sullivan 2002). Because males at yr–1; range = 0–3.5 mm ⋅ yr–1). However, both sites were territorial (Kwiatkowski and males of South had significantly (Mann-Whit- Sullivan 2002) and patrolled their territories to ney U = 12.5, P = 0.04, n = 16) higher growth defend against intruders, males at Lookout (n = 8; 5.4 mm ⋅ yr–1; range = 0–25 mm ⋅ may have expended considerably more energy yr–1) than Lookout males. For the individuals in territory defense. However, if growth was evaluated for growth, maximum SVL for Look- dependent on plant resources and energy ex- out males was 194 mm and for females was penditure in territory defense alone, then 173 mm; maximum SVL for South males was females from the South site, which also have 209 mm and for females was 154 mm. small home ranges relative to females at the Interestingly, in spite of the higher growth Lookout site, should have exhibited higher rates of South males over the course of the growth rates. Females from South did exhibit surveys (1995–1999), rainfall was generally a higher growth rate (Fig. 1), but small sample higher at the Lookout site compared with the sizes precluded statistical analysis. South site (Table 1). This pattern is not unex- Berry (1974) found that male chuckwallas pected given the more northern location of grow faster than females; for size classes of Lookout, but it is counter to the higher growth 170–220 mm SVL, females averaged 0.21 mm exhibited by South males. Unfortunately, small ⋅ yr–1 while males averaged 4.41 mm ⋅ yr–1. sample sizes precluded an analysis of individ- For the Phoenix area populations that we stud- ual growth in relation to rainfall. Although the ied, males exceeded females in growth, but 2004] NOTES 139

gesting that the size of the largest individual may serve as a proxy for asymptotic size in populations in which no longitudinal growth data are available.

We thank R. Bowker, S. Heald, E. Stitt, D. Sullivan, J. Sullivan, K. Sullivan, and T. Tuchak for help in the field. Research was supported in part by Heritage Fund grants from the Ari- zona Game and Fish Department, a research grant from Sigma Xi, and by the Department of Biology at Arizona State University.

LITERATURE CITED

ABTS, M.L. 1987. Environment and variation in life history traits of the chuckwalla, Sauromalus obesus. Ecologi- Fig. 1. Amount of growth by initial snout-vent length in cal Monographs 57:215–232. chuckwallas for the Lookout (LO) and South Mountains BERRY, K.H. 1974. The ecology and social behavior of the (SM) populations. Lookout males = open circles; Lookout chuckwalla, Sauromalus obesus obesus Baird. Univer- females = filled circles; South males = open squares; sity of California Publications in Zoology 101:1–60. South females = filled squares. CASE, T.J. 1976. Body size differences between populations of the chuckwalla, Sauromalus obesus. Ecology 57:313–323. HAMILTON, P.S. 1995. Environmental and geographic vari- this was primarily the result of dramatically ation on the expression of sexual dimorphism in the higher growth of South males. chuckwalla, Sauromalus obesus. Master’s thesis, Uni- versity of California, Riverside. Much attention has been paid to the devel- HOLLINGSWORTH, B.D. 1998. The systematics of chuck- opment of growth models to explain lizard wallas (Sauromalus) with a phylogenetic analysis of growth trajectories (Sugg et al. 1995). These other iguanid lizards. Herpetological Monographs models can be used to address hypotheses 12:38–191. concerning proximate and ultimate factors in- JOHNSON, S.R. 1965. An ecological study of the chuckwalla, Sauromalus obesus Baird, in the western fluencing body size, size at maturity, and the Mojave Desert. American Midland Naturalist 73:1–29. asymptotic size (at which point further growth KWIATKOWSKI, M.A., AND B.K. SULLIVAN. 2002. Mating is negligible) reached by each sex (Stamps and system structure and population density in a polygy- Krishnan 1997). While the data presented nous lizard, Sauromalus obesus (= ater). Behavioral here do not lend themselves to such a longitu- Ecology 13:201–208. LAMB, T., T.R. JONES, AND J.C. AVISE. 1992. Phylogenetic dinal analysis of growth, some inferences can histories of representative herpetofauna of the south- be drawn about 1 growth-model parameter: western U.S.: mitochondrial DNA variation in the asymptotic growth. Stamps and Andrews (1992) desert iguana (Dipsosaurus dorsalis) and the chuck- found that in anoles, size of the largest-bodied walla (Sauromalus obesus). Journal of Evolutionary individual in a given population is an accurate Biology 5:465–480. NAGY, K.A. 1973. Behavior, diet and reproduction in a desert estimate of the asymptotic size reached in that lizard, Sauromalus obesus. Copeia 1977:93–102. population. By regressing the average growth PRIETO, A.A., AND M.J. RYAN. 1978. Some observations of of animals on size class (Fig. 1), one can estimate the social behavior of the Arizona chuckwalla, Saur- asymptotic size as the size at which growth omalus obesus tumidus (Reptilia, Lacertilia, Iguan- reaches zero (i.e., where the regression line idae). Journal of Herpetology 12:327–336. PRIETO, A.A., AND M.W. SORENSON. 1977. Reproduction intersects with the x-axis). Our growth data in the Arizona chuckwalla, Sauromalus obesus tumidus show that asymptotic size of males is greater (Shaw). American Midland Naturalist 98:463–469. than females in each population, and there is STAMPS, J.A., AND R.M. ANDREWS. 1992. Estimating asymp- less sexual dimorphism in asymptotic size in totic size using the largest individuals per sample. Lookout than in South lizards. Estimated in Oecologica 92:503–512. STAMPS J., AND V. V. K RISHNAN. 1997. Sexual bimaturation this fashion, asymptotic sizes of each sex and and sexual size dimorphism in animals with asymp- population are concordant with maximum totic growth after maturity. Evolutionary Ecology 11: sizes reached in each population and sex, sug- 21–39. 140 WESTERN NORTH AMERICAN NATURALIST [Volume 64

SUGG, D.W., L.A. FITZGERALD, AND H.L SNELL. 1995. TRACY, C.R. 1999. Differences in body size among chuck- Growth-rate, timing of reproduction, and size dimor- walla (Sauromalus obesus) populations. Ecology 80: phism in the southwestern earless lizard, Copho- 259–271. saurus texanus scitulus. Southwestern Naturalist 40: 193–202. Received 22 July 2002 SULLIVAN, B.K., AND M.A. FLOWERS. 1998. Large iguanid Accepted 16 January 2003 lizards of urban mountain preserves in northern Phoe- nix, Arizona. Herpetological Natural History 6:13–22. Western North American Naturalist 64(1), ©2004, pp. 141–143

REPRODUCTIVE CYCLE OF SMITH’S BLACK-HEADED SNAKE, TANTILLA HOBARTSMITHI (SERPENTES: COLUBRIDAE), IN ARIZONA

Stephen R. Goldberg1

Key words: reproduction, Tantilla hobartsmithi, Smith’s black-headed snake.

Smith’s black-headed snake, Tantilla hobart- ian follicles. I examined tissues from 37 ovaries, smithi, occurs in Arizona, southern California, 34 testes, and 20 vasa deferentia from speci- western Colorado, southern Nevada, southern mens collected between 1948 and 2001. The New Mexico, southwestern Texas, southern left ovary was removed from females and the Utah, and in the Mexican states of Chihuahua, left testis and vas deferens were removed from Coahuila, and Sonora (Cole and Hardy 1981). males for histological examination. Tissues Most individuals are found beneath rocks, were embedded in paraffin, cut into 5-µm sec- mainly in riparian, grassland, chaparral, and tions, mounted on glass slides, and stained woodland communities (Cole and Hardy with Harris’ hematoxylin followed by eosin 1983). Information on clutch sizes and time of counterstain. Slides were examined to deter- oviposition for T. hobartsmithi was given by mine the stage of the testicular cycle and the Stebbins (1985) and Degenhardt et al. (1996). presence of yolk deposition (secondary vitello- Force (1935) provided detailed information on genesis sensu Aldridge 1979). Vasa deferentia reproduction in the congener Tantilla gracilis were examined for sperm. from northeastern Oklahoma as did Aldridge Testicular histology was similar to that re- and Semlitsch (1992a, 1992b) for Tantilla coro- ported by Goldberg and Parker (1975) for the nata from South Carolina. This note’s purpose colubrid snakes Masticophis taeniatus and is to provide information on the reproductive Pituophis catenifer. In regressed testes semi- cycle of T. hobartsmithi in Arizona based on a niferous tubules contained spermatogonia and histological examination of gonadal tissue from Sertoli cells embedded in a Sertoli syncytium. museum specimens. In recrudescent testes I noted a renewal of Data are presented from 74 sexually mature spermatogenic cells characterized by sper- T. hobartsmithi (40 females, mean snout-vent matogonial divisions. Primary and secondary length [SVL] = 175 mm ± 18 [s], range = spermatocytes and occasional spermatids were 141–222 mm; 34 males, mean SVL = 161 mm present. In testes undergoing spermiogenesis, ± 12 [s], range = 138–185 mm) examined from metamorphosing spermatids and mature sperm the herpetology collections of Arizona State were present. University (ASU), Tempe; Carnegie Museum Because males from June were not exam- (CM), Pittsburgh; Museum of Northern Ari- ined, the testicular cycle cannot be completely zona (MNA), Flagstaff; Natural History described. Nevertheless, the presence of males Museum of Los Angeles County (LACM), Los with testes in regression or recrudescence in Angeles; Museum of Southwestern Biology spring (Table 1) and all males undergoing (MSB), at the University of New Mexico, Albu- spermiogenesis in August and September sug- querque; and University of Arizona (UAZ), gests that T. hobartsmithi has an aestival sper- Tucson (Appendix). Data from 2 gravid females matogenesis (sensu Saint Girons 1982) with provided by P. Rosen (personal communica- multiplication of spermatogonia in spring and tion) were also included, but ovaries from spermiogenesis ending in October. This pat- these specimens were not sectioned. Counts tern also occurs in the congener T. coronata in were made of oviductal eggs and enlarged ovar- South Carolina (Aldridge and Semlitsch 1992b).

1Department of Biology, Whittier College, Whittier, CA 90608.

141 142 WESTERN NORTH AMERICAN NATURALIST [Volume 64

TABLE 1. Monthly distribution of reproductive stages in the seasonal testicular cycle of 34 Tantilla hobartsmithi from Arizona. Values shown are the number of males exhibiting each of the 3 conditions. Month n Regression Recrudescence Spermiogenesis March 8 7 1 0 April 14 8 6 0 May 4 2 2 0 July 1 0 1 0 August 5 0 0 5 September 2 0 0 2

TABLE 2. Monthly distribution of reproductive stages in the seasonal ovarian cycle of 40 Tantilla hobartsmithi from Arizona including 2 gravid females* from P. Rosen (personal communication). Values shown are the number of females exhibiting each of the 4 conditions. Early yolk Enlarged follicles Month n Inactive deposition (18 mm length) Oviductal eggs January 1 1 0 0 0 March 6 6 0 0 0 April 16 15 1 0 0 May 4 3 0 0 1* July 5 4 0 0 1 August 6 5 0 1* 0 October 2 2 0 0 0

All vasa deferentia of T. hobartsmithi from the Thus, for T. hobartsmithi in Arizona, clutch sizes following dates contained sperm: 4 March, 10 (n = 3) range between 1 and 2. The remainder April, 2 May, 2 August, 2 September. Force of the T. hobartsmithi females examined were (1935) suggested that Tantilla gracilis mated not reproductively active. The presence of during May in Oklahoma. Aldridge and Sem- 27/31 (87%) reproductively inactive females litsch (1992b) reported that T. coronata mates during April, May, July, and August may sug- during spring and summer in South Carolina. gest that only a portion of the female popula- Most species of Tantilla are thought to mate in tion produces eggs annually. This is typical for the spring (Rossi and Rossi 1995). Tantilla other colubrid snakes from the southwestern hobartsmithi may similarly mate in the spring United States (see, for example, Goldberg 2000, 2001). However, examination of a large sample utilizing sperm stored in the vasa deferentia of T. hobartsmithi females during the months of from autumn spermiogenesis; however, the reproduction from the same population and possibility that fall mating may occur cannot year are needed before this can be known. be dismissed. Aldridge and Semlitsch (1992a) reported an As was reported for other species of Tantilla ovarian cycle for T. coronata similar to that of by Clark (1970), female T. hobartsmithi lack a T. hobartsmithi in which vitellogenesis occurs functional left oviduct. A female T. hobart- in the spring, with ovulation occurring in June smithi (Table 2) from April was undergoing and egg deposition in June and early July. In early yolk deposition with basophilic yolk gran- Oklahoma, T. gracilis egg deposition (2–3 eggs ules in the ovarian follicles (SVL 190 mm, UAZ most commonly) occurred from the middle of 32948), while 1 from May (SVL 208, LACM June to the middle of July (Force 1935). The 20472) contained 2 oviductal eggs. P. Rosen period of egg deposition may be later for T. (personal communication) supplied information hobartsmithi in Arizona and could be timed to on the following 2 gravid T. hobartsmithi. A coincide with the summer monsoon period female from 25 July (SVL 190 mm, Cochise and resultant moisture. However, examination County) contained 1 oviductal egg, and 1 female of additional gravid females will be needed to from 2 August (SVL 222 mm, Pima County) ascertain the time of oviposition for T. hobart- contained 2 enlarged eggs (18 mm length). smithi. 2004] NOTES 143

Previous reports on reproduction in T. FORCE, E.R. 1935. A local study of the opisthoglyph snake hobartsmithi include clutches of 1–3 eggs laid Tantilla gracilis Baird and Girard. Papers of the Michigan Academy of Science, Arts and Letters in June, July, and perhaps August (Stebbins 20:645–659. 1954, 1985). Additional data on clutch size in GOLDBERG, S.R. 2000. Reproduction in the longnose snake, T. hobartsmithi are from females collected in Rhinocheilus lecontei (Serpentes: Colubridae). Texas Big Bend National Park, Brewster County, Journal of Science 52:319–326. Texas. These include 3 reports of clutches of ______. 2001. Reproduction in the night snake, Hypsi- glena torquata (Serpentes: Colubridae), from Ari- 1 egg each from T. hobartsmithi (= T. atriceps) zona. Texas Journal of Science 53:107–114. deposited 23 June, 28 July, and 4 August GOLDBERG, S.R., AND W. S. P ARKER. 1975. Seasonal testic- (Easterla 1975); 1 female with a single egg ular histology of the colubrid snakes, Masticophis ready to be deposited 1 June (Minton 1958 taeniatus and Pituophis melanoleucus. Herpetologica 31:317–322. [1959]); and a clutch of 3 eggs deposited 19 MINTON, S.A., JR. 1958 [1959]. Observations on amphib- June (Degenhardt et al. 1996). Field observa- ians and reptiles of the Big Bend region of Texas. tion and subsequent collection of gravid Southwestern Naturalist 3:28–54. females will improve our understanding of the ROSSI, J.V., AND R. ROSSI. 1995. Snakes of the United States reproductive biology of this species. and Canada: keeping them healthy in captivity. Vol- ume 2, Western area. Krieger Publishing Company, Malabar, FL. I thank G. Bradley (UAZ), J. Gillette (MNA), SAINT GIRONS, H. 1982. Reproductive cycles of male snakes A. Holycross (ASU), D. Kizirian (LACM), and their relationships with climate and female reproductive cycles. Herpetologica 38:5–16. C. Painter (MSB), and J. Wiens (CM) for per- STEBBINS, R.C. 1954. Amphibians and reptiles of western mission to examine specimens and K. Beaman North America. McGraw-Hill, New York. (LACM) for helpful comments on the manu- ______. 1985. A field guide to western reptiles and script. P. Rosen (UAZ) provided information amphibians. Houghton-Mifflin, Boston. on clutch sizes. Received 19 August 2002 Accepted 21 January 2003 LITERATURE CITED

ALDRIDGE, R.D. 1979. Female reproductive cycles of the snakes Arizona elegans and Crotalus viridis. Herpe- tologica 35:256–261. APPENDIX. Specimens examined from the herpetology ALDRIDGE, R.D., AND R.D. SEMLITSCH. 1992a. Female re- collections of Arizona State University (ASU), Carnegie productive biology of the southeastern crowned snake Museum (CM), Museum of Northern Arizona (MNA), (Tantilla coronata). Amphibia-Reptilia 13:209–218. Museum of Southwestern Biology (MSB), Natural History ______. 1992b. Male reproductive biology of the south- Museum of Los Angeles County (LACM) and University eastern crowned snake (Tantilla coronata). Amphibia- of Arizona (UAZ). Reptilia 13:219–225. CLARK, D.R., JR. 1970. Loss of the left oviduct in the colu- LOCALITY.—Cochise Co.: CM 40413–41415; MSB 57234; brid snake genus Tantilla. Herpetologica 26:130–133. UAZ 40061, 43941, 47192, 50656, 50743. Coconino Co.: COLE, C.J., AND L.M. HARDY. 1981. Systematics of North UAZ 28042. Gila Co.: UAZ 42685. Graham Co.: UAZ American colubrid snakes related to Tantilla plani- 43884, 43887. Greenlee Co.: ASU 30456, 30458; UAZ ceps (Blainville). Bulletin of the American Museum 42784. Maricopa Co.: LACM 20472, 103725, 103726, of Natural History 171:199–284. 125277, 125449, 125450; UAZ 26392, 26417, 37454, 37457, ______. 1983. Tantilla hobartsmithi Taylor. Smith’s black- Pima Co headed snake. Catalogue of American Amphibians 43779. .: MNA 27.777, 27.779, 27.790, 27.791, and Reptiles 318:1–2. 27.793; UAZ 26403, 26414, 26416, 26419, 26422, 26428, DEGENHARDT, W.G., C.W. PAINTER, AND A.H. PRICE. 1996. 26433, 26434, 28552, 30474, 30757, 31985, 32948, 32949, Amphibians and reptiles of New Mexico. University 35630, 36426, 39862, 40406, 40458, 42018, 42360, 42496, of New Mexico Press, Albuquerque. 44350, 47180, 50373, 50602, 53402. Pinal Co.: UAZ 26399– EASTERLA, D.A. 1975. Reproductive and ecological obser- 26401, 26405, 26420, 26423, 26432, 30335, 43882, 43883, vations on Tantilla rubra cucullata from Big Bend 44022, 47486. Yavapai Co.: UAZ 49929. National Park, Texas (Serpentes: Colubridae). Her- DISTRIBUTION.—Arizona. petologica 31:234–236. Western North American Naturalist 64(1), © 2004, p. 144

BOOK REVIEW

Weird Nature. John Downer. Firefly Books a velvet worm (phylum Onychophora) squirt- (U.S.) Inc., Buffalo, NY. 2002. 156 pp. ing sticky “glue” to ensnare prey! The discus- ISBN 1-55297-586-X. sion and photo of the parasitic candiru catfish approaching the midsection of a male in the This is a beautifully illustrated and composed Amazon will assuredly attract the attention of book telling short stories of some of the weird at least half of the readers. ways animals make their livings. Written as a On the whole, the individual written treat- companion to a Discovery Channel series from ments are too short for my tastes. But, of cable television, this book is heavy on pictures course, one must remember the attention span but light on text and explanations. of the particular television audience and the I was thrilled to read of strange things ani- goals of the author. Despite considering this, I mals do in such chapters as Marvelous Motion, was disappointed that the weird and bizarre Bizarre Breeding, Fantastic Feeding, Devious stories and amazing photographs were not fol- Defenses, Puzzling Partners, and Peculiar lowed by some kind of bibliography or note on Potions. Many of the examples, which are how to read more about the facets covered. indeed weird by human standards, are familiar This, for me, was frustrating, and for this rea- to the professional natural historian. To illus- son I would suggest finding the book at a pub- trate, nice photos of flying lizards, hovering lic library and quickly reading it in a quiet nautilus, and head-to-head spiny lobsters are corner. The content is wonderful but too brief given. The explanation of jumping beans being for use by biologists in the museum, classroom, caterpillars in desert seeds and the artistry of or on a field course. bower birds follow. The piece on foam-nesting frogs is certain to grab the reader’s attention C. Riley Nelson and must have been even more striking in Department of Integrative Biology video. But enough of the segments are new Brigham Young University and fresh that it was exciting to turn each Provo, Utah 84602 page. I found particularly amazing the shot of [email protected]

CONTENTS (Continued from back cover) Notes Groundwater invertebrate fauna of the Bear River Range, Utah ...... Candace Brindza Huebner and Mark R. Vinson 131 Collection of an adult gizzard shad (Dorosoma cepedianum) from the San Juan River, Utah ...... Gordon A. Mueller and Jim L. Brooks 135 Growth in Sonoran Desert populations of the common chuckwalla (Sauromalus obesus) ...... Brian K. Sullivan, Matthew A. Kwiatkowski, and Paul S. Hamilton 137 Reproductive cycle of Smith’s black-headed snake, Tantilla hobartsmithi (Serpentes: Colubri- dae), in Arizona...... Stephen R. Goldberg 141 Book Review Weird Nature by John Downer ...... C. Riley Nelson 144

144