Full Issue, Vol. 61 No. 4

Western North American Naturalist

Volume 61 Number 4 Article 18

11-15-2001

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

Recommended Citation (2001) "Full Issue, Vol. 61 No. 4," Western North American Naturalist: Vol. 61 : No. 4 , Article 18. Available at: https://scholarsarchive.byu.edu/wnan/vol61/iss4/18

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 61(4), © 2001, pp. 381–394

BIOLOGY AND CONSERVATION OF THE CORAL PINK SAND DUNES TIGER BEETLE, CICINDELA LIMBATA ALBISSIMA RUMPP

C. Barry Knisley1 and James M. Hill2

ABSTRACT.—This study investigated the distribution, abundance, and biology of Cicindela limbata albissima Rumpp, an endemic tiger beetle known only from the Coral Pink Sand Dunes (CPSD) in southwestern Utah. A recently imple- mented conservation agreement between BLM, USFWS, Utah State Parks, and Kane County protects most of the known habitat of this beetle from off-highway vehicle (OHV) use. A search of collection records and field surveys of 19 Great Basin sand dune sites indicated that this species occurs only at CPSD. Yearly index counts of adults (1992–1998) during peak season in May ranged from 331 in 1997 to 895 in 1993, but the actual population size is probably 2–3 times higher than the index counts. Nearly all of the population is found in the primary habitat, a 300-m-wide × 2.7-km-long area in the southern part of the dune field. Small numbers of adults and larvae have been found at the far north end of the dune field. Mark-recapture studies indicated that most adults moved only short distances (<300 m), but a few moved 1000 m. This beetle has a 2-year, modified spring–fall life cycle. Adults are most abundant from April through early June, but some adults from the following year’s adult cohort emerge and can be found from late August to early October. Adults are active on warm or sunny days, but they dig burrows which they use at night or during unfavorable weather. We observed little evidence of parasitism or predation of larvae or adults, but these limiting factors were not fully studied. Surveys of dominant plant species, arthropod (potential tiger beetle prey) abundance, and OHV activity indicated that these vary throughout the dune field and may explain, in part, the distribution of C. l. albissima. The primary habitat is a transition area between the highly dynamic south end of the dune field and the more stabilized north end. Psoralidium lanceolatum Rybd., Sophora stenophylla Gray, and Stipa hymenoides R.&S. are the dominant plants in the interdunal swales of the primary habitat, but other species are dominant in other parts of the dune field. Numbers of arthropod individuals and taxa are greatest in the primary habitat. Off-highway vehicle activity was greatest at the south end of the dune field and lowest at the far north end. Run-over trials and observational data revealed that adult beetles are killed by OHVs, but more important impacts may be damage to vegetation, reduction of arthropod prey of C. l. albissima, and disturbance and increased desiccation of the larval microhabitat. We anticipate that the conservation agreement will provide long-term protection for this species at CPSD.

Key words: Cicindela limbata albissima, Coral Pink Sand Dunes, tiger beetle, conservation, off-highway vehicle impacts, population dynamics.

Tiger beetles of the genus Cicindela are subspecies, including Cicindela arenicola predatory insects that prefer open, sparsely Rumpp from the Bruneau and St. Anthony vegetated habitats. Adults are active, visual dunes of Idaho, C. theatina Rotger from the hunters that use their large mandibles to cap- Great Sand Dunes of Colorado, C. scutellaris ture and eat small arthropods. Larvae are yampae Rumpp from northwestern Colorado, sedentary predators that live in permanent and C. formosa rutilovirescens Rumpp from burrows in the ground. They also use large the Mescelaro dunes of southeastern New mandibles to capture small arthropods that Mexico. The only published study of these pass near the mouth of their burrows. There is species is that of Bauer (1991) on C. arenicola. considerable interest in tiger beetle conserva- Cicindela limbata albissima was described tion. Two species (Cicindela d. dorsalis Say, C. by Rumpp (1961) from specimens collected at puritana G. Horn) are listed as threatened by the Coral Pink Sand Dunes in southwestern the U.S. Fish and Wildlife Service (USFWS). Utah. He distinguished it from other subspecies Several others are under consideration for list- of C. limbata by the greatly reduced elytral ing. Among the 10+ Cicindela species occur- pigmentation (the elytra are white except for a ring in sand dune habitats in the western narrow medial line) and its disjunct geographic United States are several endemic species or range in southern Utah, a distance of over 600

1Department of Biology, Randolph-Macon College, Ashland, VA 23005. 2RR 1, Box 2746, Reedville, VA 22539.

381 382 WESTERN NORTH AMERICAN NATURALIST [Volume 61 km from other populations of the species. The of the dunes. An additional 370-acre conserva- other subspecies of C. limbata have separate, tion area at the northern part of the sand more northern geographic ranges, are east of dunes has also been proposed. In this paper the Continental Divide, and have more pigmen- we present the results of our 1992–1999 stud- tation on the lateral portions of their elytra. ies of this rare insect, including information on Cicindela limbata limbata occurs in the western distribution and abundance, seasonality, habi- Great Plains, C. limbata nympha occurs from tat, and possible OHV impacts. North Dakota northwest into southern Canada, and C. l. hyperborea is found in northern METHODS Canada. Johnson (1989) recently described C. l. Geographic Distribution labradorensis from Goose Bay, Labrador, although Larson (1986) suggested this form was Because tiger beetles are a popular, well- C. l. hyperborea. Rumpp (1961) believed that collected group of insects, our initial work in C. l. albissima was more closely related to C. l. determining the distribution of C. l. albissima nympha that, he said, followed a southward involved a compilation of collection records migration route during the Pleistocene when from various sources. We examined tiger bee- climatic conditions were more favorable. He tle specimens in 23 museum and university suggested that the lack of elytral pigmentation collections that were rich in tiger beetle hold- ings or were likely to have many Utah records. may be a thermoregulatory adaptation to the Among the most important collections were warmer temperatures where C. l. albissima those at Brigham Young University, California occurs. Acorn (1992) provided support of the Academy of Sciences, Utah State University, relationship between coloration and thermo- U.S. National Museum, American Museum of regulation in C. limbata and other dune Cicin- Natural History, and Yale Peabody Museum. dela. He also reported on the ecology of C. We also obtained label information from sev- limbata nympha and other sand dune Cicin- eral individuals who had large cicindelid col- dela in the southern Canadian plains (Acorn lections. We conducted field surveys of all 1991). A recently completed analysis of the major sand dunes and many other sandy areas mtDNA in C. limbata and its relatives found in Utah and adjacent states where potential that C. l. albissima is genetically distinct from habitat for this species might occur (Fig. 2). All the other subspecies and that it should be of these sites were visited 1 or 2 times in May given full species status (Morgan et al. 2000). 1993, 1994, and 1995. At each site we spent Cicindela l. albissima was first listed as a several hours to several days searching for category II species in 1984 (Federal Register tiger beetles in areas of suitable habitat. The 49:21664). In 1994 the USFWS was petitioned searching method involved walking through by the Southern Utah Wilderness Alliance to the open areas of potential habitat and looking list C. l. albissima as endangered and to desig- on the ground 5–10 m ahead for the adults to nate critical habitat. The service’s response run or fly up as they were approached. This (Federal Register 59[178]:47293–47294) indi- visual search method is commonly used to sur- cated that the petition presented substantial vey for tiger beetles (Knisley and Schultz 1997) information in support of listing, but progress and is effective when done during a species’ toward its listing was interrupted by the list- peak activity period (warm, sunny days from ing moratorium in 1996. Currently, C. l. albis- mid-April to early June for C. l. albissima). At sima is a candidate species. In 1997 a conserva- most of these sand dune sites we also sur- tion agreement was signed by the USFWS, veyed dune arthropods using pitfall traps. We Bureau of Land Management (BLM), Utah set out 20–32 sixteen-oz cup traps (Carolina State Parks and Recreation, and Kane County. Biological Supply Co., Burlington, NC) with The provisions of this agreement include the approximately 100 mL of ethylene glycol (no establishment of a 350-acre conservation area longer recommended for this use) for 5- to 7- in the southern part of the dune field that is day periods in May 1993, 1994, and 1995. the primary beetle habitat (Fig. 1B). The western portion of this area is signed off to prevent Studies on Distribution, Abundance, vehicle access, while the eastern portion and Biology serves as a travel corridor for vehicles to move Much of our initial work at Coral Pink Sand between the northern and southern portions Dunes (hereafter CPSD) in May 1992 involved 2001] BIOLOGY AND CONSERVATION OF C. L. ALBISSIMA 383

Fig. 1. A, Map of Coral Pink Sand Dunes showing different study areas; B, map of the primary habitat, area C, showing 6-year mean numbers of adults in different swales. The conservation area is divided into a non-motorized area and a travel corridor as indicated. determining the distribution of adults and lar- Relative population size of adult C. l. albis- vae within the dune field. We spent 4–6 hours sima at the Coral Pink Sand Dunes was deter- per day for several days walking over all parts mined each year in May from 1992 to 1998 of the dune field and making preliminary counts using index counts and mark-recapture. In of adults and larvae and recording their distri- this paper we include only the results of the bution relative to physical landmarks, Global index counts, which were conducted through- Positioning System (GPS) readings, and wooden out the entire primary habitat during 2- to 3- stakes which we placed in interdunal swales day periods each year in mid-May from 0930 (see below). We separated the CPSD dune field to 1230 hours on sunny, mild (>20°C) days. into 7 areas, AAA at the north end to E at the During this time of day, most beetles were south end (Fig. 1A). Our preliminary surveys active and primarily concentrated along the indicated that adults and larvae were largely edges of the interdunal swales and on adjacent concentrated in an area ~300 m wide × 2.7 lower slopes. In making the counts, 2 individu- km long in the southern portion of the dune als positioned themselves 20–25 m apart and field (area C). Most of our studies were within walked back and forth across the width of the this area, which we call the primary habitat dune field, progressing from north to south (Fig. 1B). and counting all observed beetles. We kept 384 WESTERN NORTH AMERICAN NATURALIST [Volume 61

To establish fixed survey points, we placed wooden stakes within these swales of the primary habitat near the centers of patches where larvae were aggregated and counted all larval burrows within a 10-m-diameter circle drawn in the sand around each stake. GPS readings were taken at each stake location so lost or vandalized stakes could be replaced. New stakes were added as we discovered new concentrations of larvae. Additionally, in 1996, 1997, and 1998 we also estimated total numbers of larval burrows in each swale (including those outside the fixed plots) by walking back and forth through the swale areas and counting all observed burrows. This number was added to plot counts to obtain a total count of larval burrows in each swale. We also compiled April–October rainfall records for Kanab, the nearest weather station to CPSD, because we hypothesized that rainfall might be an important factor affecting the population dynamics of C. l. albissima. Movement of adults was determined by marking and releasing 200–400 adults each year during a 2–3 day period in mid-May. Fig. 2. Map of Great Basin sites surveyed for Cicindela Beetles were marked by removing a 1-mm l. albissima and other tiger beetles (see Table 1 for site section from the tip of one of their elytra. By information). cutting at different angles, we had unique marks for many of the swale areas. Beetles were recaptured 3–10 days after marking, and separate counts for each swale-slope area. We distances between mark and recapture loca- also counted adults of another tiger beetle, tions were measured. Seasonal activity, life Cicindela tranquebarica, which co-occurred cycle, and other natural history information of with C. l. albissima in some swales. C. l. albissima were determined from our The distribution and abundance of tiger adult and larval surveys and field observations beetle larvae can be determined by searching during our May and September visits to the the ground and counting numbers of burrows Coral Pink Sand Dunes between 1992 and during times when larvae are active (Knisley 1998. Feeding habits of adult C. l. albissima 1987, Knisley and Schultz 1997). The 3 larval were determined by capturing individuals that instars can be distinguished by differences in were feeding and then examining their prey. burrow diameter, which is fixed within an instar. Accurate estimates of larval numbers are diffi- Habitat Characteristics and cult to obtain for sand dune species like C. l. OHV Impact Studies albissima because their larval burrows may be We conducted surveys of dominant plant plugged and thus unrecognizable during the species and arthropod abundance in different day when the sand surface becomes warm and areas of CPSD to determine if these might aid dry, and at other times when conditions are in explaining the distribution of C. l. albissima unfavorable (Knisley 1987). We surveyed lar- within the dune field and to characterize its val burrows from 0700 to 1100 hours when a habitat. Within each of the 7 areas of the dune high percentage of larvae are active. Prelimi- field (Fig. 1A), we estimated the percentage nary surveys in 1992 indicated that larvae cover of each plant species within 10-m-diam- were most abundant in open or sparsely vege- eter circular plots located at 25-m intervals tated edges of interdunal swales and low slopes. along a transect line running across the length 2001] BIOLOGY AND CONSERVATION OF C. L. ALBISSIMA 385 of each interdunal swale. A wire hoop subdi- Direct effects of OHVs on tiger beetles vided into 4 quarters was placed on the ground were determined from experimental run-over at each sample point and the percentage cover trials and by examination of individual beetles of each species within each quarter estimated. collected during mark-recapture studies. Run- The sum of the 4 quarters gave the total per- over trials were performed at CPSD in May cent cover for a species. In the primary habitat 1994 on 3 different substrates (dry sand, wet we estimated the percent cover of plant species sand, mixed sand and stones). In preparation in most interdunal swales. for the trials, we tied 1 end of a 50-cm length Surveys for swale arthropods (the prey re- of thread around the thorax of a beetle and source base of adults and especially larvae of tied a piece of plastic surveyor flagging to the C. l. albissima) were conducted by placing 6 other end. The beetles were held in place on pitfall traps (see above) in 2 different swales in the sand surface by covering the middle por- each of the 7 areas of the dune field for 6-day tion of the string with a handful of sand. The periods (9–15 July 1993, 5–11 September 1993, groups of beetles were run over by 10 passes 21–27 May 1994, and 18–24 May 1996). Three of a 1994 Honda 4-wheeled vehicle. They traps were placed around the perimeter of the were examined after 1, 5, and 10 passes of the swale and 3 around the interior near the swale vehicle. Another indication of probable direct bottom. We transferred arthropods from each effects of OHVs was noted during our 1994 trap to separate plastic bags and later identi- mark-recapture studies when we found adult fied them to species (where possible). Mean beetles with injuries similar to those in the numbers of individuals and taxa were deter- run-over trials. In 1994 and subsequent years, mined from 12 traps (6 in each swale). Another we examined all beetles captured during pitfall trap survey from 16–22 May 1998 was mark-recapture studies and recorded the designed to determine if there was a relation- numbers that were injured. ship between OHV activity and arthropod abundance. In this experiment we set 4 sets of RESULTS 6 traps in each of 4 different areas of the dune Distribution and Abundance field (AA, B, C, E). Two sets in each area were placed in 2 swales that were in areas of higher Only 1 of 65 locality records researched for OHV activity (determined by higher numbers the distribution of C. l. albissima was from a of OHVs and OHV roads), and 2 sets were site other than CPSD. That record was of a placed nearby in 2 swales with lower OHV specimen collected by H.P. Boyd (personal activity. A paired t test was used to compare communication 1994) from a sandy floodplain means for low- versus high-use swales in each of the Virgin River south of Mt. Carmel Junc- area. tion (~15 km north of CPSD). We did not find Off-highway vehicle activity at CPSD was C. l. albissima at any of the 19 sites that we assessed on Memorial Day weekends in 1997 surveyed (Fig. 2), although other Cicindela sp. and 1998 and on Labor Day weekend in 1998. were found at some sites (Table 1). Five sites These holiday weekends are among the high- (2, 4A, 4B, 4C, 5) seemed likely prospects for est use times for OHV activity at CPSD. We C. l. albissima because of their close proximity counted the number of OHVs that crossed 8 to CPSD and the presence of Asclepias welchii, transect lines (2 in area C and 1 in each of the a rare milkweed that may have similar habitat other dune areas; Fig. 1A) across the width of requirements because it occurs only at these the dune field during 4 one-hour periods, 2 on sites and at CPSD. Saturday and 2 on Sunday, of each weekend. The index counts in the primary habitat, We also measured the width of all OHV paths where nearly all adults were found, ranged from or roads that crossed each of these 8 transect 331 in 1997 to 895 in 1993 (Table 2). Most of lines and summed them to obtain a total road these adults (60–79%) were in swale rows and width for each area. We determined the amount low slopes in the northern part of the primary of motorized play, primarily the riding back habitat, swales 4 through GH. Highest mean and forth over the dunes, for the 8 areas of the adult numbers (based on annual index counts dune field by obtaining the mean minutes per from 1993 to 1999) were in swales KJ (97), 2–3 hour that all OHVs (vehicle minutes) were en- (85), 1 (73), and H (60; Fig. 1B). Numbers of gaged in this activity. adults in adjacent areas to the north and south 386 WESTERN NORTH AMERICAN NATURALIST [Volume 61

TABLE 1. Great Basin sand dunes sites surveyed for Cicindela limbata albissima. Elevation Site Location (m) Species found 1–UT, Coral Pink 15 km W Kanab 1830 C. l. albissima C. tranquebarica 2–UT, Sand Hills 14 km NNE Kanab 1830 none 3–UT, Sand Mountain 10 km SE Washington 1250 none 4A–AZ, Sand Cove 45 km E Kanab 1615 none 4B–UT, Coyote Buttes 45 km E Kanab 1740 none 4C–UT, AZ, state line 45 km E Kanab 1585 none 5–AZ, Thousand Pockets 12 km W Page 1540 none 6–UT, Holden Dunes 10 km WNW Holden 1430 C. tranquebarica 7–UT, Oak City 5 km W Oak City 1435 C. lepida 8–UT, Little Sahara 12 km W Jerico Junction 1640 C. lepida 9–UT, San Raphael Desert Rt. 24, 12 km S I-70 1341 none 10–UT, Hanksville 22 km N Hanksville 1585 none 11–UT, Snow Canyon SP 12 km N Santa Clara 1310 none 12–UT, Green River Green River at Utah Co. 1700 C. formosa formosa state line, Brown Park C. tranquebarica 13–CO, Maybell Dunes 2 km E Maybell 1835 C. formosa formosa C. scutellaris yampae 14–CO, Great Sand Dunes Great Sand Dunes National 2130 C. theatina Monument 15–WY, Boars Tusk Dunes 43 km N Rock Springs 1890 C. limbata limbata? C. tranquebarica 16–ID, St. Anthony Dunes 11 km N St. Anthony 1524 C. arenicola 17–ID, Bruneau Dunes Bruneau Dunes State Park 980 C. arenicola

TABLE 2. Index counts of adult Cicindela l. albissima in the primary habitat at Coral Pink Sand Dunes, 1992–1998. 1992 1993 1994 1995 1996 1997 1998 Total for primary habitat 651 895 511 513 843 331 758 Swales north of swale 4 49 46 25 38 107 23 50 Swales south of GH 107 139 91 116 234 48 220 % of total north of 4 and south of GH 24 21 23 30 40 20 36 of the high-density area varied among years, other years. We found a significant positive but highest total percentages were in 1998 correlation (Spearman rank correlation, r = (36%) and 1996 (40%; Table 2). Adults were 0.036, P = 0.035) between the total April– always more abundant in the adjacent area to October rainfall in one year and adult num- the south, which extended for about 1300 m bers the following year (Fig. 3). For example, south of GH and included many swales having very low rainfall totals in 1993, 1994, and 1996 at least some adults (Fig. 1B). The adjacent were associated with very low adult numbers area of primary habitat to the north extended the following years. only about 500 m beyond swale 4 and Checks of recaptured beetles during 6 years included only a few swales with adults. The (1993–1998) of mark-recapture studies indi- only adults we found beyond the primary cated that C. l. albissima adults moved very habitat were small numbers (<20 per year) of little throughout the dune field at CPSD. Of scattered individuals in 1994, 1996, and 1998 275 adults recaptured, 205 did not move be- south of the primary habitat (areas D and E) yond the swale row (<200 m) in which they or in the far northern end of the dune field were marked, 34 moved 200–300 m, 16 moved (areas AAA and AA). 300–500 m, 14 moved 500–700 m, and 6 moved Total rainfall from April through October 700–1000 m. (the activity period for adults and larvae of C. Our initial observations and subsequent l. albissima) varied greatly from 1991 to 1998. surveys indicated that larvae were most com- Totals for the last 3 years were higher than all mon in sparsely vegetated outer edges (away 2001] BIOLOGY AND CONSERVATION OF C. L. ALBISSIMA 387

TABLE 3. Mean numbers of larvae in specific swale groups in the primary habitat at CPSD. Means are based on single counts in May and September 1993–1998 for individual swales listed in each row.

______Mean number of larvae Swales May September 5–9 76.8 97.9 4 193.4 189.8 1 247.8 227 PB,2,3 518.8 294.1 IJK 295.2 377.4 GH 244.4 129.1 LMN 78.8 69.6 OP 190.8 111.7 QR 414.4 281.2 ST 38.2 22.6 UVW 201 262.2 Fig. 3. Rainfall amounts and adult and larval numbers. XYwY 46.3 29.4 Rainfall amounts are total mm from April through October each year at Kanab. Adult numbers are index counts in mid-May. Larval numbers are totals of all larval burrows counted in all swales of the primary habitat. scattered burrows in 1 or 2 swales in this area, but in 1996 we counted a total of 74 larvae in 5 swales, in 1997 we found 67 in 8 swales, and from the slipface) of the interdunal swales and in 1998 we found 352 in 8 swales. in nonvegetated adjacent low dune slopes. Some larvae were also found in the more Seasonal Activity, Life Cycle densely vegetated interior of some swales, and and Behavior small numbers were occasionally seen on mid- We first saw adults of C. l. albissima on the and upper slopes, especially after rains. We dunes in late March or early April after they also noticed that on days after a substantial emerged from overwintering (Fig. 4). Emer- rainfall many larvae appeared at the base of gence continues through May, with peak num- some slipfaces. (Tiger beetle larvae frequently bers occurring from early to mid-May. Num- clear out their burrows after rains.) Total numbers begin to decline in late May, and by late bers of larval burrows counted (permanent June most adults have died off. Few adults are plots and other areas within the swales) in the found from July to mid-August. A small pro- primary habitat during May and September portion of the previous year’s larvae completes varied more than threefold over the years. its development, pupates, and emerges in late May counts ranged from 3567 in 1996 to 908 August. These adults are active until late Sep- in 1998, while September counts ranged from tember or early October, then dig overwinter- 2830 in 1996 to 1880 in 1993 (Fig. 3). As with ing burrows. Our estimates of adult numbers adults, highest numbers of larvae were in the during this period were ~300 in 1991, ~100 in area from swale 4 south to swale row GH (Table 1993 and 1994, and 160 in 1998. These fall 3). Most swale rows in this area had means adults will reemerge in spring along with the near or >250 larvae in both May and Septem- larger proportion of the cohort, which develops ber survey dates. Other swales with similar more slowly and does not emerge in the late high mean larval numbers were QR, UVW, summer. and OP. However, larval numbers in QR and Mating and ovipositing were observed soon UVW declined to <100 in 1996, 1997, and after spring adults emerged and continued 1998, while numbers in OP increased about throughout the adult activity period. First- twofold during this same period. Mean number instar larvae were first seen in late May, and of burrows for the 5 swales north of swale 4 (5 by July most of the new larval cohort had pro- to 9) was <100 in both May and September gressed to the second instar (Fig. 4). Many counts. Small numbers of larvae were also found second instars completed their development ~4 km from the primary habitat in 12 swales by October before overwintering as third in- in the far northern end of the dune field (areas stars, but slower developing individuals over- AAA, AA). Prior to 1996 we found only a few wintered as second instars. Third instars had 388 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Fig. 4. Seasonal abundance of the life history stages of Cicindela l. albissima. Fig. 5. Daily activity pattern for adults and larvae in May at CPSD. the longest developmental period, and most of the larval cohort continued in this stage through most of the 2nd year before pupating exhibit contact guarding (Knisley and Schultz and emerging as adults during the 2nd fall or 1997) and use the mandibles to grasp the female the following spring. The rate of larval devel- as she digs the burrow. Sixty-six percent of opment is greatly influenced by climate, the these burrows were within interdunal swales amount of food obtained, and perhaps other and low slopes, 24% in the mid-slope area, factors that may cause the larvae in a cohort to and 10% at or near the dune crest. The 85 prey develop asynchronously. In Manitoba the life items observed being eaten by adults included cycle of C. l. nympha was 3 years (Criddle 27 ants, 21 already dead insects (mostly orthop- 1910), probably because lower temperatures terans and large ants), 11 flies, 9 homopterans, in the spring and fall reduce seasonal activity 7 beetles, and 10 unidentified insects. of larvae. Another tiger beetle, Cicindela tranquebar- Observations on warm, sunny days in May ica, also common at CPSD, co-occurred (both indicated that adults begin to emerge from adults and larvae) with C. l. albissima in some their overnight burrows and appear in swales interdunal swales of the primary habitat where and low slopes at about 0830–0930 hours (Fig. there was a layer of clay in the sand and/or 5). Numbers increase rapidly, with peak abun- presence of very moist soil (Romey and Knis- dance from about 1000 to 1230 hours, and ley in preparation). Adults and larvae of C. then decline from about 1300 to 1600 hours as tranquebarica were also common in similar many adults dig burrows in the dune slopes to types of swales outside the primary habitat avoid high surface temperatures. Some of these (especially E, AA, A) where C. l. albissima was adults reemerge from their burrows about absent. 1600 hours and remain active until 1700–1900 We observed little evidence of natural ene- hours, when they again dig burrows on dune mies attacking C. l. albissima. On 2 June 1994 slopes or interdunal swales to spend the night. we found an individual of the small, antlike On cloudy or windy days when surface tem- parasitoid wasp, Methocha sp., entering a sec- peratures remain <35°C, adults may be active ond-instar larval burrow. After the wasp exited most of the day. For several years in May we the burrow, we dug out the larva and found it made observations on adults digging burrows. paralyzed and with an egg on it. In May 1995 Adult beetles used their mandibles and legs to we found 2 third-instar C. l. albissima larvae dig the burrow and push out the sand with an with larvae of the bee fly (possibly Anthrax alternating back-and-forth, sweeping movement sp.). Several days before, we observed an adult of the meso- and metathoracic legs. These fly in the same area. None of the other 600+ burrows were 4–9 cm deep, slightly angled, larvae that we dug out between 1991 and 1998 and 7–16 cm long. Fifty-two of 111 observa- were parasitized. Both parasitoids are specific tions were of females, 37 were of males, and to tiger beetle larvae, but their effects on tiger 22 were of mated pairs. Adult males of C. l. beetle populations are difficult to determine albissima, like those of many other species, unless individual burrows are marked and 2001] BIOLOGY AND CONSERVATION OF C. L. ALBISSIMA 389 monitored. We frequently observed asilid flies absent from swales of the primary habitat. on the dunes but did not see any instances of Dicoria brandegei was observed to be extreme- this known predator of adult tiger beetles ly abundant and widespread in primary habitat attacking C. l. albissima. swales during September, but it was absent or present only as seedlings in May when vegeta- Habitat Characteristics tion surveys were conducted. Total percent veg- Dominant plant species within the swales etation cover was higher in all swales of the pri- of the primary habitat (area C) were different mary habitat (means of 23–57%) than in swales from those in other dune areas (AAA, AA, A, in other areas (means of 12–17%; Table 4). B, D, E). Psoralidium lanceolatum Rydb. had Mean numbers of arthropod taxa and indi- the highest mean coverage percent in all but 1 viduals per area (12 traps in 2 swales) from the of the swale rows of the primary habitat, while 1993–1996 pitfall surveys were significantly Sophora stenophylla Gray, Stipa hymenoides different (Kruskal-Wallis ANOVA, P < 0.05) R. & S., and Reverchonia arenaria Gray were among the 7 areas surveyed. Highest means also abundant and widespread within this area were in areas C and AAA (Table 5). Mean (Table 4). These species were absent or much numbers of both taxa and individuals collected less common within swales of other dune in the 1998 surveys were significantly higher areas. Wyethia scabra Hook. was absent or rare in the low OHV-use swales than in the high- in primary habitat swales, but it was a domi- use swales (paired t test, P < 0.05). Important- nant plant species in areas AAA, AA, A, B, and ly, the highest numbers of taxa and individuals E, along with Chrysothamnus sp., Eriogonum were in the low-use swales of area C, and the spp., and several other species (Gilia congesta greatest difference in numbers of individuals Hook, Redfieldia flexosa) included in the (1027 and 422) and taxa (51 and 39) between “other” category of Table 4. The northern half of low- and high-use swales was also in area C. the dune field (areas AAA to B) had a greater This great difference in area C probably reflects diversity of plant species and was dominated the greater difference in amount of OHV use by Wyethia scabra, which also was the domi- in high- and low-use swales of this area. nant plant species at the south end of the dune field (area E). Juncus arcticus Willd., abundant OHV Activity and Impacts in many of the wetter swales in the northern Mean numbers of OHVs, mean play time, and southern parts of the dune field, was nearly and total width of OHV roads were greatest in

TABLE 4. Mean percent cover of plant species in swales in different areas at CPSD. Plant species are: S.s. = Sophora stenophylla Gray, P.l.= Psoralidium lanceolatum Rybd., S.h.= Stipa hymenoides R.&S., R.s. = Reverchonia arenaria Gray, D.b. = Dicoria brandegei Gray, W.s. = Wyethia scabra Hook., C.s. = Chrysothamnus sp., E.a.= Eriogonum.

______Plant species Other Total % S.s. P.l. S.h. R.s. D.b. W.s. C.s. E.a. sp. cover

C (PRIMARY HABITAT OF C. L. ALBISSIMA) 5–9 10 60 20 16 4 25 29 46 4 13 75 8 15 2 8 55 1 682112194 3 46 2,3 22 71 6 20 2 48 IJK 26 65 12 14 1 5 57 GH 23 64 9 11 1 45 L–P 32 21 9 4 5 32 QR 10 58 27 8 5 5 1 52 ST 18 70 9 34 6 28 U–Y 16 62 8 5 1 6 23

OTHER AREAS AAA 2 6 12 31 10 22 17 16 AA 17 33 3 47 15 A 2 12 12 9 57 12 B 12 4 19 30 9 31 16 E 8 12 40 15 25 17 390 WESTERN NORTH AMERICAN NATURALIST [Volume 61

TABLE 5. Mean numbers of arthropod taxa and individuals collected from pitfall traps in different areas of CPSD. Means for 1993–1996 are based on total numbers in 6 pitfall traps each in 2 different swales in each area averaged for 4 survey periods (see text). Numbers for 1998 are totals from 6 pitfall traps each in 2 low- versus 2 high-use OHV swales in each area.

______1993–1996 ______1998 low OHV ______1998 high OHV No. of No. of No. of No. of No. of No. of Area taxa indiv. taxa indiv. taxa indiv. AAA 43 285 AA 18 124 27 194 22 90 A 24 139 B 23 133 44 260 36 245 C 27 341 51 1027 39 422 D 22 222 E 21 148 24 282 26 184 MEAN TOTALS 36.5 441 30.8a 235a aMeans are significantly different for low- versus high-OHV swales (paired t test, P < 0.05). area E and lowest at the far northern end (area and surrounding areas may be due to the AAA; Table 6). Motorized vehicle play time absence of other high-elevation sand dunes. was over 4 times higher in area E than in any The Great Sand Dunes of Colorado is the only other dune area. OHV activity parameters in similar high-elevation site, and the closely the other dune areas were quite variable but related species C. theatina occurs there. Cicin- indicated intermediate level of OHV activity dela. l. albissima, like its more northern rela- between the 2 ends of the dune field. Areas E tives, is apparently restricted to habitats with and A are major points of vehicle access to the cooler temperatures. Rumpp (1961) suggested dunes. that the distinct lack of elytra pigmentation in Numbers of injured adults of C. l. albissima C. l. albissima may be a thermoregulatory adap- found during our mark-recapture studies var- tation to its presence in a more southern loca- ied greatly among years. Numbers injured and tion than the other subspecies of this species. total number captured and examined during The single individual that was found north of Memorial Day periods were as follows: 1993, CPSD may have been a dispersing individual 14 of 179 injured in 1–2 days after Memorial since only one was encountered. Several Day; 1994, 0 of 363 injured before Memorial searches of this area by us revealed no beetles Day weekend, 6 of 125 after; 1996, 2 of 231 or suitable habitat. before and 41 of 287 after; 1997, 2 of 256 before We believe that the index counts give an and 6 of 64 after; and 1998, 1 of 168 before accurate representation of the year-to-year vari- and 8 of 278 after. Results of run-over trials ation in numbers of adult C. l. albissima, but indicated that the type and/or condition of sub- the counts clearly underestimate the actual strate and number of passes were important population size at CPSD. Lincoln index values in the fate of beetles run over by an OHV. No from mark-recapture studies that we conducted beetles on dry sand were injured or killed coincidentally with the 1993–1998 index counts when run over once, but a majority of those on (Knisley and Hill unpublished studies), other either wet sand or gravel were injured or killed studies we have done with Cicindela (Knisley by a single vehicle pass (Table 7). Most or all and Schultz 1997), and a study comparing dif- beetles were killed or injured by 5 and 10 ferent survey methods in 1999 (Gowan and passes on all substrates. Knisley in preparation) all suggest the actual population size may be 2–3 times higher than DISCUSSION the index counts. The underestimate is probably the result of some beetles flying off before Distribution, Abundance, they are counted or occurring in patches of and Biology overlooked habitat or in burrows. Our search of collection records and our The significant year-to-year variation in C. field surveys indicate that C. l. albissima occurs l. albissima numbers is typical of many desert only at CPSD. Its limited distribution in Utah arthropods that are greatly affected by climatic 2001] BIOLOGY AND CONSERVATION OF C. L. ALBISSIMA 391

TABLE 6. OHV activity throughout different areas of CPSD. OHV numbers are means of two 1-hour counts each on Saturday and Sunday of Memorial Day weekends in 1997 and 1998 and Labor Day 1998. Vehicle play time is the mean number of OHVs multiplied by minutes per hour averaged for the 3 weekends (Cn is the northern part of C, Cs the southern part).

______Area of dune field AAA AA A B Cn Cs D E Mean number of OHVs per hour 3 19 69 60 67 70 51 94 Mean vehicle play (vehicle 22 62 138 161 50 92 138 771 minutes per hour) Total width of OHV roads (m) 75 125 188 181 122 108 226 230

TABLE 7. Effects of OHV run-over trials on adults of Cicindela l. albissima on 31 May 1994.

Substrate No. in No. of ______Effect on beetles type trial passes No effect Injured Killed Dry 5 1 5 0 0 Dry 5 5 2 1 2 Dry 5 10 0 0 5 Wet 15 1 5 7 3 Wet 15 5 2 0 13 Wet 15 10 1 0 14 Gravel 5 1 2 3 0 Gravel 5 5 2 1 2 Gravel 5 10 0 1 4 factors, especially rainfall (Seely 1991). The than those in May. We also found survivorship positive correlation we found between April– was <20% for several patches of larvae that October rainfall amounts one year and adult were marked as first instars in May 1996 and numbers the following year suggests rainfall monitored through part of the third instar may be one of the key factors affecting popula- (Knisley unpublished study). tion dynamics of C. l. albissima. Studies with Much of what we learned about the biology other desert tiger beetle species in Arizona of C. l. albissima was similar to that of other demonstrated that higher rainfall increases tiger beetles that have been studied. Its biol- survivorship of larvae by reducing mortality ogy is especially similar to that of C. l. nympha from desiccation and starvation, especially of in Canada (Acorn 1991) and C. arenicola from first instars (Knisley 1987, Knisley and Juliano the St. Anthony Sand Dunes in Idaho (Ander- 1988). Prey was more abundant during wet son 1988, 1989, Bauer 1991). The amount of years, and this not only reduced the effects of movement of marked individuals of C. l. albis- starvation but also decreased development sima was, like C. arenicola, quite limited com- time and increased fecundity. We expected pared to many tiger beetle species. Anderson that these effects and the importance of soil (1989) found that most marked adults of C. moisture for oviposition and egg hatch should arenicola remained in the same general area similarly result in higher larval numbers in where they were marked and only a few dis- September counts in years with high rainfall, persed (in a “step-stone” manner) >500 m. The but our data do not show this. Obviously, other C. l. albissima found near Mt. Carmel Junction factors are important in population dynamics suggests that, at times, adults may disperse of this species, and clear patterns of cause and much greater distances than what we observed. effect are difficult to determine without more The adults and larvae found at the far north- detailed studies. Our study does provide some ern end of the dune field probably represent a evidence that larval survivorship is very low colonizing event by adults that dispersed from for this insect. Total adult numbers are consis- the primary habitat. Dispersal may be greater tently much lower than larval numbers, and during times of high density, as is common in larval numbers in September counts are lower many other animals. For example, we found that 392 WESTERN NORTH AMERICAN NATURALIST [Volume 61 highest proportions of adults in areas adjacent the geology, vegetation, prey arthropods, and to the swale 4 to GH concentration area were OHV activity throughout the CPSD dune in 1996 and 1998, 2 of the years of greatest field, and all of these could significantly affect adult abundance. the distribution and abundance of C. l. albis- The seasonal pattern we observed for C. l. sima. Importantly, the primary habitat is very albissima represents a variation of the typical closely associated with distribution of the spring–fall life cycle common in many Cicin- transverse type of sand dunes at CPSD, while dela sp. In this pattern most larvae complete other dune types are present north and south their development in the summer, pupate, and of the transverse dunes where C. l. albissima emerge as sexually immature adults in late occurs (Ford and Gillman 2000). There is also August and September. They dig burrows in a distinct transition in the dynamics, physical October to overwinter and then reemerge in characteristics, and elevation of the dune field the spring. The very low counts of adult C. l. progressing from southwest to northeast. The albissima in September suggest that most of south end of the dune field (area E) is lowest in the maturing larval cohort do not emerge in elevation (<1800 m) and nearest the sand the fall, although they may molt into the adult source that feeds the dunes. It is the most stage and remain in their pupal burrows dynamic area with the least vegetated inter- (Schultz 1998). Acorn (1991) found that num- dunal swales (with little P. lanceolatum and S. bers of adults of C. l. nympha in the Canadian stenophylla) and the best developed and most dunes peak in June and September. This is the active dunes. Wyethia scabra is the dominant typical spring–fall pattern and similar to what cover plant in this sparsely vegetated part of Bauer (1991) found for Cicindela arenicola. the dune field. This is also the part of the dune The pattern of adult burrowing during midday field with the highest level (by far) of OHV to escape heat or unfavorable weather condi- activity (a primary vehicle play area) and relations and at night is similar to that of C. areni- tively low numbers of arthropods. The high cola (Anderson 1988) and other Cicindela OHV use probably contributes to the low veg- (Willis 1967, Knisley and Schultz 1997). Our etation cover, an effect that is well documented finding of more females and mated pairs dig- (Vollmer et al. 1976, Hosier and Eaton 1980, ging burrows may indicate that females could Luckenbaugh and Bury 1983). We have ob- be using burrows to oviposit and thus position served that many interdunal swales in this eggs deeper into the soil where more moisture area had OHV tracks indicative of heavy use is present. Such behavior has been observed and were nearly devoid of vegetation. We can- in other species in the laboratory and in the not determine which of these differences in field (Anderson 1989, Knisley and Schultz this part of the dune field explains the absence 1997). Our observations of predation and of C. l. albissima. Adults and larvae of C. tran- potential predators on adults indicate that pre- quebarica were mostly found in swales with dation may not be a major limiting factor for damp soils and lower OHV activity. adults. Interestingly, Acorn (1991) found evi- The northern half of the dune field (areas A dence that adults of Cicindela formosa are an to AAA) is the highest in elevation (>1900 m) important predator of adult C. l. nympha. More and the least dynamic area, with large num- studies are needed to determine the impor- bers of ponderosa pine stabilizing the dunes. tance of parasitism and predation for C. l. Most of the swales, however, are sparsely veg- albissima. etated, similar to that at the south end, and dominated by Wyethia scabra and a greater What Factors Explain the Distribution variety of plant species. Arthropod abundance of C. l. albissima at CPSD? is low and OHV activity moderate to heavy in Results of this study provide some insight areas B and A, but in the far north end (AAA) but do not fully explain the localized distribu- this pattern is reversed. Low OHV activity tion of C. l. albissima within the CPSD dune and high numbers of arthropods in AAA may field. The concentration of adults and larvae partly explain the presence of C. l. albissima within the primary habitat and the presence of there and its absence in areas A and B. The very few adults with moderate numbers of larvae increase in numbers of larvae in the past 3 in swales at the far north end of the dune field years may be due partially to the exceptionally are puzzling. There is significant variation in high rainfall and its effect on larval recruitment 2001] BIOLOGY AND CONSERVATION OF C. L. ALBISSIMA 393 and survival (see above). However, the very habitat (Schultz 1988). Soil moisture measure- low adult numbers here suggest that this north ments taken in 1996 in different portions of end of the dune field may be unfavorable for the CPSD dune field indicated that areas in complete development and successful emer- and near heavy-use OHV roads had signifi- gence of adults. cantly lower soil moisture readings than undis- The primary habitat of C. l. albissima (area turbed areas nearby (Gwilliam, Hill, and Knis- C) has some distinctive features that may con- ley unpublished study). Although it cannot be tribute to the beetle’s predominance there. concluded from our study, it is possible that This area is a transitional zone between the high levels of OHV activity in area E and parts highly dynamic south end and the stabilized of areas B and A, just north of the primary dunes north of area B. Dune slopes and ridges habitat, prevent the successful colonization of are open and very active, but interdunal swales C. l. albissima. have higher percent vegetation cover relative Despite the potential negative impacts from to other areas. The dominance of Psoralidium, OHV activity, there is no evidence of a pro- Sophora, and Stipa in these swales may progressive decline of the population of C. l. vide a different and more favorable habitat for albissima at CPSD. There are no records of larvae by supporting a greater abundance of this species’ abundance prior to the beginning arthropods as a food source for larvae and of OHV activity >20 years ago, and so it is adults. The level of OHV activity in this area unknown whether the population was histori- is moderate, but numbers of OHV roads are cally larger. Year-to-year fluctuations in adult fewer, play time less, and apparent OHV dam- population size have been significant, but age to swales more limited than in adjacent declines have been followed by a rebound in areas to the north and south. We have also abundance. It is believed that implementation observed during the years of our study that of the conservation plan in 1998, which pre- swales in the northwestern part of the primary vents or reduces vehicle activity in most of the habitat supporting most of the adults and lar- primary habitat, will provide long-term pro- vae have received less OHV use than swales tection for this species at CPSD. Continued along the eastern side of the primary habitat. monitoring and study of the C. l. albissima Our studies show that some adults are killed population will help to determine this. each year from run-overs by OHVs, but it is not certain if this significantly impacts the pop- ACKNOWLEDGMENTS ulation. A greater effect may occur when adults are run over and crushed in their shallow adult Logistic support and assistance in these burrows on the dune slopes. We have not ob- studies were provided by Rob Quist, Dan served this effect. Low levels of OHV activity Richards, and the staff at Coral Pink Sand may not impact larvae because their burrows Dunes State Park. Tim Smith was especially are deep enough (>20 cm) for them to avoid helpful in our early work at CPSD. Field assis- being crushed. In experimental run-over tri- tance was provided by Kevin Fielding, Charles als, Anderson (1989) found no effects to larvae Davis, Ryan Knisley, Charles Gowan, Bruce of C. arenicola after 10 OHV passes. Heavy Gwilliam, and Larry England (USFWS). Charles OHV activity, such as seen at some areas of Davis identified the plants. Charles Gowan CPSD, may impact the population through produced several of the figures. We greatly other direct or indirect effects on adults and appreciate the assistance, cooperation, and fi- larvae. Adult feeding, oviposition, and mating nancial support made available by Ron Bolander may be disrupted by OHVs, and this effect of the Utah BLM through a Challenge Cost could reduce recruitment, as was reported for Share Agreement and Larry England of the C. dorsalis on Virginia’s coastal beaches (Knis- U.S. Fish and Wildlife Service through a Coop- ley and Hill 1992). The reduction of swale erative Agreement. vegetation may coincidentally reduce prey arthropod abundance and negatively affect lar- LITERATURE CITED val survival. Off-highway vehicles may cause a mixing of the upper sand layers, which can ACORN, J.H. 1991. Habitat associations, adult life histories, and species interactions among sand dune tiger bee- increase desiccation (Webb et al. 1978) or alter tles in the southern Canadian prairies (Coleoptera: the soil moisture gradient of the larval micro- Cicindelidae). Cicindela 23:17–48. 394 WESTERN NORTH AMERICAN NATURALIST [Volume 61

______. 1992. The historic development of geographic Atlantic states. Virginia Museum of Natural History. color variation among dune Cicindela in western Special Publication 5. 210 pp. Canada. In: G.E. Noonan, G.E. Ball, and N.E. Stork, LARSON, D.J. 1986. The tiger beetle, Cicindela limbata editors, The biogeography of ground beetles in moun- hyperborea LeConte, in Goose Bay, Labrador (Cole- tains and sslands. Intercept Press, Andover, UK. 256 optera: Cicindelidae). Coleopterists Bulletin 40: pp. 249–250. ANDERSON, R.L. 1988. The sand dunes tiger beetle. Final LUCKENBAUGH, R.A., AND R.B. BURY. 1983. Effects of off- report to the Bureau of Land Management, Idaho road vehicles on the biology of the Algodones Dunes, Falls District, ID. Imperial County, California. Journal of Applied ______. 1989. The sand dunes tiger beetle. Final report to Ecology 20:265–286. the Bureau of Land Management, Idaho Falls Dis- MORGAN, M., C.B. KNISLEY, AND A.P. VOGLER. 2000. New trict, ID. taxonomic status of the endangered tiger beetle BAUER, K.L. 1991. Observations on the developmental Cicindela limbata albissima (Coleoptera: Cicindeli- biology of Cicindela arenicola (Coleoptera: Cicindel- dae): evidence from mtDNA. Annals of the Entomo- idae). Great Basin Naturalist 51:226–235. logical Society of America 93:1108–1115. CRIDDLE, N. 1910. Habits of some Manitoba tiger beetles. RUMPP, N.L. 1961. Two new tiger beetles of the genus No. 2 (Cicindelidae). Canadian Entomologist 42:9–15. Cicindela from the southwestern United States. Bul- FORD, R.L., AND S.L. GILLMAN. 2000. Geology of Coral letin of the Southern California Academy of Science Pink Sand Dunes State Park, Kane County, Utah. 60:165–187. Pages 365–389 in D.A. Sprinkel, T.C. Chidsey, Jr., SCHULTZ, T.D. 1988. Destructive effects of off-road vehicles and P.B. Anderson, editors, Geology of Utah’s parks on tiger beetles habitat in central Arizona. Cicindela and monuments. Utah Geological Association Publi- 20:25–29. cation 28. ______. 1998. Verification of an autumnal diapause in HOSIER, P.E., AND T.E. EATON. 1980. The impact of vehi- adults of Cicindela sexguttata. Cicindela 30:1–6. cles on dune and grassland vegetation on a south- SEELY, M.K. 1991. Sand dune communities. Pages 348–382 eastern North Carolina barrier beach. Journal of in G.A. Polis, editor, The ecology of desert communi- Applied Ecology 17:173–182. ties. University of Arizona Press, Tucson. JOHNSON, W.N. 1989. A new subspecies of Cicindela lim- VOLLMER, A.T., B.G. MAZA, P.A. MEDICA, F.B. TURNER, AND bata Say from Labrador (Coleoptera: Cicindelidae). S.A. BAMBERG. 1976. The impact of off-road vehicles Naturaliste Canadien (Revue d’Ecologie Systema- on a desert ecosystem. Environmental Management tique) 116:261–266. 1:115–129. KNISLEY, C.B. 1987. Habitats, food resources, and natural WEBB, R.H., H.C. RAGLAND, W.H. GODWIN, AND D. JEN- enemies of a community of larval Cicindela in south- KINS. 1978. Environmental effects of soil property eastern Arizona (Coleoptera: Cicindelidae). Canadian changes with off-road vehicle use. Environmental Journal of Zoology 65:1191–2000. Management 2:219–233. KNISLEY, C.B., AND J.M. HILL. 1992. Effects of habitat WILLIS, H.B. 1967. Bionomics and zoogeography of the change from ecological succession and human impact tiger beetles of the saline habitats in the central on tiger beetles. Virginia Journal of Science 43: United States (Coleoptera: Cicindelidae). University 335–340. of Kansas Science Bulletin 48:145–313. KNISLEY, C.B., AND S.A. JULIANO. 1988. Survival, development and size of larval tiger beetles: effects of food Received 5 November 1999 and water. Ecology 69:1983–1992. Accepted 30 August 2000 KNISLEY, C.B., AND T.D. S CHULTZ. 1997. The biology of tiger beetles and a guide to the species of the south Western North American Naturalist 61(4), © 2001, pp. 395–402

VARIATION IN THE BARK CALL OF THE RED SQUIRREL (TAMIASCIURUS HUDSONICUS)

Osamu Yamamoto1, Barry Moore1, and Leonard Brand1,2

ABSTRACT.—Calls of the red squirrel (Tamiasciurus hudsonicus; n = 122) were recorded in wild populations from 15 localities in Arizona, New Mexico, Colorado, Utah, Wyoming, Montana, Idaho, and Washington. Computer-generated audiospectrograms of 20- or 30-second samples from a calling bout of each individual were analyzed. Eighteen bark types (distinct forms of the bark call) were identified plus a 19th category that included rarely used, longer bark calls. The frequency of use of each bark type within the sample was recorded for each squirrel. Differences in frequency of use of the various bark types were found among subspecies, within subspecies, and within populations; additionally, the southern subspecies utilized a reduced number of bark types. The large number of different bark types and the variation in bark type usage within populations suggest the potential for communication of such information as individual identification, behavioral states, or gender identification.

Key words: vocalizations, behavior, geographic variation, Sciuridae, Rodentia, Tamiasciurus.

The red squirrel (Tamiasciurus hudsonicus) The bark is one of the most frequently heard is a small, semi-arboreal mammal of the Hud- vocalizations of the red squirrel (Embry 1970). sonian and Canadian life zones of North Amer- Barking bouts can last for just a few seconds ica (Hall 1981). It inhabits coniferous and de- or can continue for nearly an hour (Embry ciduous forests throughout the Rocky Moun- 1970). The bark is also reported by both Smith tains, most of Canada, the Great Lakes states, (1978) and Embry (1970) as being the most and New England (Hall 1981). The abundant variable of the 5 calls, and its function is some- literature on the ecology, behavior, and taxon- what contested. Smith (1978) interpreted the omy of this species has been summarized by call strictly as an alarm call and invoked kin Steele (1998). Tamiasciurus hudsonicus has been selection to explain its origin. Greene and divided into 25 subspecies, with the greatest Meagher (1998) also consider at least some diversity occurring in the Rocky Mountains bark calls to be alarm calls, which differ accord- south of Canada (Hall 1981, Lindsay 1987). ing to the type of predator. Searing (1977) Western populations of pine squirrels of interpreted the bark call as a low-intensity, both T. hudsonicus and the closely related T. aggressive call. Other authors (Embry 1970, douglasii exhibit exclusive territoriality (Smith Nodler 1973, Lair 1990) questioned its func- 1968). According to Smith (1968), the basis of tion as an alarm, preferring the broader inter- territorial behavior is the need for individual pretation mentioned above, and Lair’s (1990) squirrels to harvest, store, and defend a sea- work seems to be consistent with this concept. sonal supply of food so that it will be available In his analysis of the behavioral context of red all year long. Vocal display is an important part squirrel calls, Lair (1990) concluded that the of this territory defense behavior. Four of 5 bark was a poor indicator of the caller’s behav- different calls used by T. hudsonicus (growl, ior. Embry (1970) quantified the variability in buzz, rattle, and screech) are related to territo- this call and identified at least 7 different types rial behavior (Embry 1970, Lair 1990, Price et of bark calls. al. 1990). The 5th call (bark) has been inter- There is a high degree of variability in the preted as expressing fear, anger, frustration, or bark call of Tamiasciurus, which may indicate a conflict of motivation (Klugh 1927, Embry that the call conveys different meanings in dif- 1970, Nodler 1973, Lair 1990), or as an alarm ferent contexts (Lair 1990). Embry (1970) call (Smith 1978, Price et al. 1990, Greene and found this variation to exist within individuals, Meagher 1998). among individuals, between sexes, among age

1Department of Natural Sciences, Loma Linda University, Loma Linda, CA 92350. 2Corresponding author.

395 396 WESTERN NORTH AMERICAN NATURALIST [Volume 61 classes, and among subspecies in western Mon- tana and an adjacent locality in northwestern Wyoming. Our study extended the analysis of geographic variation begun by Embry. The purposes of this study were to examine variation of the bark call over a broader geographic range, to better understand the geographic variability of this call, and to test the hypothesis that significant variation exists between the calls of different subspecies. This study aims to provide a basis for further research on the function of the call.

MATERIALS AND METHODS Recording Method While walking along a road or trail, we made tape recordings of Tamiasciurus bark calls at 15 localities from the western United States (Fig. 1). One call was recorded from each of 122 individuals representing 7 subspecies. As Fig. 1. Range map of the western subspecies of Tamias- ciurus hudsonicus showing study sites. T. h streatori: (1) much of each call bout was recorded as possi- Washington, Stevens Co., Orient; (2) Idaho, Kootenai Co., ble (range = 30 seconds to 10 minutes). To Hayden Lake; T. h. richardsoni: (3) Montana, Broadwater prevent recording the same squirrel twice, we Co., 20 mi E Townsend; (4) Idaho, Lemhi Co., Bannack did not make 2 recordings in the same area Pass, 32 mi SE Salmon; (5) Idaho, Idaho Co., 13 mi E Slate Creek; (6) Idaho, Valley Co., McCall; (7) Idaho, unless we could determine that the calls were Boise Co., 12 mi N Boise; T. h. ventorum: (8) Wyoming, produced by different squirrels. Recordings Fremont Co., 6 mi SW Lander; (9) Idaho, Bear Lake Co., were made at any time of day during daylight 14 mi NNE Montpelier; (10) Utah, Cache Co., 15 mi NE hours (not on stormy days) from any squirrel Logan; T. h. fremonti: (11) Utah, Summit Co., 6 mi E that began calling, and all recordings were Kamas; (12) Colorado, Teller Co., 3 mi NW Woodland Park; T. h. mogollonensis: (13) Arizona, Greenlee Co., 32 made during August and September, after mi N Clifton; T. h. grahamensis: (14) Arizona, Graham young of that year reached adult size. After Co., Mt. Graham; T. h. lychnuchus: (15) New Mexico, completing the recordings at a locality, the Otero Co., 5 mi S Cloudcroft. Map after Hall (1981). individual making the field recordings estimated the size of the area that would include the position of all recorded squirrels at that Analysis of Recordings locality: 3, 10, or 20 km2. Recordings from 1992 were analyzed on a Fieldwork was done in 1992 by Barry Moore Gateway 2000 personal computer with Kay (Moore 1993) and in 1995 and 1996 by Osamu Elemetrics Corporation Computerized Speech Yamamoto (Yamamoto 1998). There were dif- Lab (CSL) model 4300 hardware and version ferences in the recording and sound-analysis 4 software, with sound digitized at 40,000 Hz. equipment available to us at these 2 time peri- Audiospectrograms were generated on the CSL ods, but the differences did not affect our abil- system with the following parameters: frame ity to accurately identify peeps and barks. length of 256 points, 0.80 pre-emphasis, Black- Recordings were made with a Uher 4000 Report man window weighting, 18.00–48.00 darkness L tape recorder at a tape speed of 19 cm ⋅ s–1, scale, 0 dB gain, 20 kHz display range, 2 × 2 or with a WM-D6C Sony professional walk- pixels grid size, and linear display (Kay Ele- man cassette-tape recorder. Filters were not metrics Inc. 1991, Moore 1993). used. The microphone was either a Senn- Recordings from 1995 and 1996 were ana- heiser MKH 404 or ME-62, mounted on a 61- lyzed using a Macintosh computer with Canary cm parabolic dish. Recordings were stored on 1.2.1 software. With this system, calls were laboratory standard polyester magnetic tape by digitized at 44,100 Hz and 16-bit sampling size, Realistic, or on TDK MA110 metal bias and audiospectrograms were generated with IECIV/type IV cassette-tapes. the following parameters: analysis resolution 2001] VARIATION IN TAMIASCIURUS BARK CALLS 397 of 69,940 Hz filter bandwidth and frame length of 256, time grid resolution of 2.902 ms with 50% overlap, frequency grid resolution of 86.13 Hz with 512 point FET size, and 20 kHz display range (Bioacoustics Research Program 1995, Yamamoto 1998). Calls were played into the audiospectrogram equipment on the same tape recorder with which they were recorded. We analyzed a 20- or 30-second sample from each recording (localities 2, 5, 6, 7, 10, 11: 20 seconds; localities 1, 3, 4, 8, 9, 12–15: 30 seconds), with a total of 6373 notes (syllables). Fig. 2. An audiospectrogram containing (left to right) a With the system used in 1992, it was not prac- bark (bark type B; 1 syllable), a peep and 2 barks (bark type PBB; 3 syllables), and a peep and bark (bark type PB; tical to analyze samples longer than 20 sec- 2 syllables). onds because of the time required for producing sonograms with this equipment. In the 2nd part of the study, each 30-second sample con- valid discriminant analysis. After the discrimi- sisted of three 10-second segments, one each nant analysis calculated its own canonical vari- from the beginning, middle, and end of the ables from the data, individuals were entered recording. as unknowns, and the test was used to identify Embry (1970) described 2 components that the population to which each belonged, as a form all variations of the bark call. She called test of consistency of the interpopulation dif- these the alpha and the beta components, but ferences in call parameters. we call them, respectively, the peep and the We performed a test to determine how many bark syllables (Fig. 2). The peep has a chevron- 10-second segments were needed to adequately shaped structure, often with 3 harmonics. The represent the number of bark types used in a bark is a column of noisy sound that com- call. For each of 9 locations distributed through- monly reaches 12+ kHz and frequently con- out the research area, the longest recording tains numerous harmonics. (range = 1–10 minutes) was analyzed in 10- In this paper we use the following defini- second segments. The cumulative number of tions of terms: call bout—all of the vocaliza- bark types given by the end of each 10-second tions in one continuous interaction by a single segment of the recording was plotted against squirrel; bark call—one of 5 types of calls number of 10-second segments analyzed to used by red squirrels, composed of bark and that point. The number of 10-second segments peep syllables; syllable—a single sound, either after which number of bark types ceased to a bark note or a peep note (Fig. 2); bark type— increase indicated the total length of sample one of the combinations of peeps and/or barks needed, on average, for a complete count of used by red squirrels (Fig. 2); vocabulary—the bark types. We then tested variation in vocab- number of bark types used by a given individ- ulary size in different segments of a call by ual or population of squirrels. comparing number of bark types in three 10- The variables we analyzed were number of second segments in each of 73 recorded calls. bark types, variation in frequency of use of The hypothesis that vocabulary size was not bark types among populations, and variation significantly different among 10-second seg- in bark rate (syllables per second). The hy- ments of the call was tested with a chi-square pothesis that populations had equal vocabu- test, with the actual number of call types in lary sizes (in number of bark types) was tested each segment of an individual call bout tested with 1-way ANOVA with the Tukey-B post against the expected ratio of 1:1:1. test. Consistency of differences between pop- Variation in bark rate (syllables per second) ulations in number of bark types, usage of bark between populations was tested with 1-way types, and bark rate were evaluated with a dis- ANOVA with the Tukey-B post test. Consis- criminant analysis using Wilk’s routine. This test tency in bark rate between segments of a call analyzed data only for the 11 most frequently was tested by comparing bark rate in three 10- used bark types because the 8 remaining bark second segments in each of 73 recorded calls. types were performed too infrequently for a The hypothesis that bark rate was consistent 398 WESTERN NORTH AMERICAN NATURALIST [Volume 61 throughout each call was tested with a chi- square test, as described above. All tests were considered statistically significant at P < 0.05. Chi-square tests were done with software written by David Cowles. All other statistical computations were carried out on Statistical Package for the Social Sciences 6.0 (SPSS; Norusis 1993).

RESULTS Variation in Bark Types This study describes 19 bark types made up of different combinations of peeps and barks (Figs. 2, 3). The most common bark calls are a single bark (B bark type; 30.0%), a single peep (P bark type; 14.6%), and the PB bark type (a peep followed by a bark; 41.5%). In contrast, a Fig. 3. Frequency of usage of bark types by 7 subspecies number of the longer bark types each consti- of T. h. hudsonicus. In the abbreviations of bark types, a P tuted a fraction of 1% of the sample. Longer indicates a peep and a B indicates a bark. combinations of barks and peeps are most commonly used by the northern subspecies in this study, T. h. streatori and T. h. richardsoni, and much less by the southern subspecies (Fig. 3). Also, in the 2 northern subspecies the single bark is the most common bark type, replaced in most southern subspecies by the PB bark type. In T. h. grahamensis the single peep is the most common type. The column labeled COMB in Figure 3 includes rarely heard combinations of alternating barks and peeps longer than those in the other columns. The longest complex consisted of 8 peeps alternating with 7 barks. This call begins to sound like the territorial trill or rattle call which consists of closely spaced barks, but the rattle call was never found to have associated peeps. These COMB call types occurred 23 times (0.4%) in this study. Some individuals from the T. h. ventorum population ± at locality 10 and a T. h. dixiensis population in Fig. 4. Means 1sx– for vocabulary size in number of central Utah (not included in this paper bark types, and bark rate in syllables per second. N indi- because of small sample size) followed bark cates number of squirrels (1 recording per squirrel). calls with a heavy wheezing sound. Vocabulary size, expressed as number of In most cases there were no significant dif- bark types used, was significantly different ferences in vocabulary size among populations among populations (Fig. 4; F = 4.65; df = 14; within a subspecies. Of 15 pairs of populations P < 0.0001). The southern subspecies used that differed significantly (Tukey-B test; Fig. 5), fewer bark types and had lower variability in only a single pair was within the same sub- number of bark types used. Although these species. This pair consisted of 2 populations (8 subspecies used the 3 most common bark types, and 10) on opposite sides of the species range they used the more complex bark types much of T. h. ventorum, which differed significantly less often, or not at all. in vocabulary size. 2001] VARIATION IN TAMIASCIURUS BARK CALLS 399

Fig. 5. Results of Tukey-B test. Numbers along the axes identify study localities, which are shown on the map. An r (bark rate) or a v (vocabulary size in number of bark types) indicates that these parameters are significantly different for that pair of populations. Lightly stippled areas outlined with a heavy line include the populations within a single subspecies.

The test of the adequacy of our sampling The frequency with which different bark method supported the use of 20- to 30-second types were used varied between individuals in samples from each recorded call. A 20- or 30- a subspecies. Figure 7 shows individual varia- second sample included on average all bark tion within one population each of T. h. rich- types for the 4 southernmost subspecies (local- ardsoni and T. h. ventorum, which represent ities 11–15) but not for most populations in the extremes in individual variation found in the 3 northern subspecies (localities 1–10; Fig. this study. Individuals from the T. h. richard- 6). Consequently, increasing the length of our soni population show much more variation samples would likely have yielded increased than those from the T. h. ventorum population. vocabulary size only for the northern sub- Despite this individual variation, there are species, which in our study already had the highly significant differences in the frequency largest vocabulary sizes. From this we infer distribution of bark type usage between popu- that if we had used longer samples from each lations and subspecies (discriminant analysis; call, our results would likely have increased Table 1). When each individual was treated as the difference between the vocabulary of the an unknown in a discriminant analysis, 95% of northern and southern populations, reinforcing the calls were placed in the correct population rather than reducing our documented difference on the basis of bark type usage. Variability did between northern and southern populations. not seem to correlate with the size of sampling This probable increase in vocabulary size would area from which calls were recorded (and thus have involved the rarely used bark notes and the potential relatedness of the squirrels; Fig. is unlikely to have changed the conclusions 4). Several populations with the smallest range reached in this paper. Also, the vocabulary size of variation represented recordings from the was in most cases not significantly different largest sampling areas. between different 10-second segments of a Variation in Bark Rate recorded call. Of 73 individuals (all with samples of 30 seconds) tested for variation in Bark rate was significantly different among vocabulary size between the three 10-second populations (Figs. 4, 5; F = 5.84; df = 14; P < segments of the sample, 65 (89%) showed no 0.0001), but there was no consistent geographic significant difference between segments (χ2 = trend. Most populations within the same sub- 0.0–2.0 [x– = 0.6]; P ≥ 0.05). The other 8 indi- species were not significantly different from viduals showed significant differences between each other in bark rate. The only 2 exceptions sample segments (χ2 = 2.4–6.3 [x– = 4.2]; P < were 2 populations (6 and 7) on the west side 0.05). of the range of T. h. richardsoni. Each differed 400 WESTERN NORTH AMERICAN NATURALIST [Volume 61

this common call could contribute to reproductive isolation if secondary contact between subspecies occurs. The behavior of these organisms has not been worked out in sufficient detail to test these ideas. Red squirrels distinguish the rattle calls of neighbors and strangers and respond differently to each (Price et al. 1990). The extent of variation that exists within the bark call of T. hudsonicus suggests that bark calls may also contain the potential for individual recognition and perhaps also for information about the sex, age, and behavioral states of the indi- Fig. 6. Cumulative vocabulary size (number of bark vidual. Whether it does convey this informa- types given) in successive 10-second segments of longest tion is not known. Full understanding of these recorded call at each of 9 localities. Labels indicate local- questions will require study of the amount and ity number (see Fig. 1) followed by name of the subspecies. context of intra-individual variability in bark calls (which our study did not include) compared with inter-individual variation. significantly from population 3, on the oppo- Greene and Meagher (1998) found that red site side of the subspecies range. squirrels used different bark types in response Bark rate tended to be consistent through- to aerial and ground predators. The squirrels out the length of a recorded call. Bark rate did used “seet” calls (the same as peeps in this not differ between the three 10-second seg- paper) or seet-barks (PB bark type) in response ments of individual calls in 69 of 73 individu- to raptors, and barks (B bark type) in response als tested (χ2 = 0.0–1.98 [x– = 0.5]; P ≥ 0.05). to humans or dogs. They indicated that only The remaining 4 showed significant differences one seet call was given in response to a bird between segments (χ2 = 2.33–3.64 [x– = 2.8]; but did not say if the seet-bark or bark vocal- P < 0.05). izations were used only as a single syllable in each predator encounter. The peeps and barks DISCUSSION that we studied were not given as single syllables but were all part of extensive calling bouts. Our data indicate considerable geographic The difference between these single-syllable variation in characteristics of the bark call in alarm calls and the longer bark call bouts that red squirrels. Variation exists at several levels: we studied deserves more study. within individuals, among individuals of a Our research does not address whether red population, and among subspecies (this study), squirrel bark calls are alarm calls. When we and between sexes and among age classes recorded calls, the squirrels did not necessar- (Embry 1970). The differences between some ily appear to be calling in response to our populations are consistent enough and large enough that bark-type frequencies could be presence, since the calls often began in the considered in studies on the systematics of distance and it was necessary to quickly move this group. The southern subspecies of red close enough to record the call. We could not squirrel have a conspicuous lack of the more determine if they were responding to other complex forms of the bark call, relative to the sources of alarm. 2 northern subspecies (T. h. streatori and T. h. Greene and Meagher’s (1998) study of barks richardsoni). and peeps as alarm calls occurred in Montana, Perhaps these more southerly populations, where we found peep notes to be rare compo- with their more homogeneous and unique call- nents of bark bouts. How might alarm calls ing patterns and reduced number of bark types, differ in a population like T. h. grahamensis, in are farther from the center of origin of the which the peep is the most common bark type? species, and part of the call variability has Will that subspecies still use a single peep as been lost. If the bark call is involved in species an alarm call? More study of the relationship recognition, the interpopulation differences in between documented alarm calls, as studied 2001] VARIATION IN TAMIASCIURUS BARK CALLS 401

Fig. 7. Percentage distribution of call types used by all recorded individuals in 1 population each of T. h. richardsoni and T. h. ventorum. Each vertical bar gives the data for 1 individual.

TABLE 1. Results of discriminant analysis of differences between populations in bark type usage. Canonical Wilk’s Function Eigenvalue correlation lambda Chi-square df P 1 3.1546 .8714 .0495 321.6 156 0.00001 2 1.4239 .7665 .1200 226.8 132 0.00001 by Greene and Meagher (1998), and geographic Lair (1990), summarizing the varied con- variation in bark calls would be beneficial. texts in which barks are used, reported that It may be that these peep or bark notes are she could distinguish at least 4 distinct vari- used differently in different contexts, with sin- ants of the bark, some of which seemed to be gle syllables given as the animal responds to a given in a restricted set of contexts. Available predator, or the same syllables given in long data are not adequate to determine how those sequences in other contexts. If this is so, it 4 variants relate to the bark types reported in seems to parallel the use of chips by chipmunks. this paper. More comparative study of red Chipmunks give long series of chip calls, last- squirrel calls and associated behavior is needed ing for up to 20 minutes or more, and these before we will understand the contexts and chips are fairly consistent in structure. When functions of the different forms of bark calls. chipmunks are startled by a ground predator, Another variable that has not yet been stud- they often give a brief, rapid chippering call as ied is variation in the acoustic environment of they escape (Brand 1976). Chip syllables are these squirrels and possible environmental used in both, but the context and length of the influences on their calls. It has been shown calling bout and specific parameters of the that physical differences between habitats can chips are consistently very different for the 2 influence the properties of sound that are types of calls. Chippering lasts a few seconds effectively transmitted in those habitats (Blum- or less, with great variation in syllable struc- stein and Daniel 1997). Some features of bird ture, but chipping bouts last many minutes vocalizations, for example, are apparently and have little variation in syllable structure. adaptations to the acoustic structure of their 402 WESTERN NORTH AMERICAN NATURALIST [Volume 61 habitats (Morton 1975, Nottebohm 1975). KAY ELEMETRICS INC. 1991. Operations manual: Comput- Whether similar factors influence red squirrel erized Speech Lab model 4300 software version 3.X. Kay Elemetrics Inc., Pine Brook, NJ. 432 pp. calls is not yet known. KLUGH, A.B. 1927. Ecology of the red squirrel. Journal of Understanding the function of this call will Mammalogy 8:1–32. contribute significantly to understanding the LAIR, H. 1990. The calls of the red squirrel: a contextual behavior of this ubiquitous mammal. Further analysis of function. Behaviour 115:255–282. detailed research on the characteristics and LINDSAY, S.L. 1987. Geographic size variation and non- size variation in Rocky Mountain Tamiasciurus hud- context of bark calls of marked individual sonicus: significance in relation to Allen’s rule and squirrels is needed to increase our under- vicariant biogeography. Journal of Mammalogy 68: standing of bark calls. Our data document sig- 39–48. nificant geographic variation in the bark calls MOORE, M.B. 1993. Variation and pattern in the bark call of the red squirrel (Tamiasciurus hudsonicus). Mas- of red squirrels and suggest profitable lines of ter’s thesis, Loma Linda University, Loma Linda, research regarding the adaptive significance of CA. 80 pp. this variation. MORTON, E.S. 1975. Ecological sources of selection on avian sounds. American Naturalist 109:17–34. NODLER, F.A.1973. Food habits, vocalizations, and territo- ACKNOWLEDGMENTS riality of Alaskan red squirrels (Tamiasciurus). Mas- ter’s thesis, University of Alaska, College. 86 pp. We thank Dr. Jean Maki and the Depart- NORUSIS, M.J. 1993. SPSS® base system user’s guide. ment of Speech and Language Pathology at SPSS Inc. 520 pp. Loma Linda University for unlimited free use NOTTEBOHM, F. 1975. Continental patterns of song vari- of their audio lab, and David Cowles for assis- ability in Zonotrichia capensis: some possible ecological correlates. American Naturalist 109:605–624. tance with statistical analysis. PRICE, K., S. BOUTIN, AND R. YDENBERG. 1990. Intensity of territorial defense in red squirrels: an experimental LITERATURE CITED test of the asymmetric war of attrition. Behavioral Ecology and Sociobiology 27:217–222. BIOACOUSTICS RESEARCH PROGRAM. 1995. Canary version SEARING, G.F. 1977. The function of the bark call of the 1.2 user’s manual. Cornell Laboratory of Ornithol- red squirrel. Canadian Field-Naturalist 91:187–189. ogy, Ithaca, NY. 229 pp. SMITH, C.C. 1968. The adaptive nature of social organiza- BLUMSTEIN, D.T., AND J.C. DANIEL. 1997. Inter- and tion in the genus of three [sic] squirrels Tamiasciurus. intraspecific variation in the acoustic habitats of Ecological Monographs 38:31–63. three marmot species. Ethology 103:325–338. ______. 1978. Structure and function of the vocalizations BRAND, L.R. 1976. The vocal repertoire of chipmunks of tree squirrels (Tamiasciurus). Journal of Mammal- (genus Eutamias) in California. Animal Behaviour ogy 59:793–808. 24:319–335. STEELE, M.A. 1998. Tamiasciurus hudsonicus. Mammalian EMBRY, P.C. 1970. Vocal communication of the red squir- Species 586:1–9. rel, Tamiasciurus hudsonicus. Master’s thesis, Uni- YAMAMOTO, O. 1998. Geographic variation in the bark versity of Montana, Missoula. 51 pp. calls of the red squirrel, Tamiasciurus hudsonicus. GREENE, E., AND T. M EAGHER. 1998. Red squirrels, Tami- Master’s thesis, Loma Linda University, Loma Linda, asciurus hudsonicus, produce predator-class specific CA. 62 pp. alarm calls. Animal Behaviour 55:511–518. HALL, E.R. 1981. The mammals of North America. 2nd edi- Received 8 February 1999 tion. John Wiley & Sons, New York. 1:1–600 +90. Accepted 27 April 2000 Western North American Naturalist 61(4), © 2001, pp. 403–408

NANONEMOURA, A NEW STONEFLY GENUS FROM THE COLUMBIA RIVER GORGE, OREGON (PLECOPTERA: NEMOURIDAE)

R.W. Baumann1 and G.R. Fiala2

ABSTRACT.—Nanonemoura, a new genus of Nearctic Nemouridae, is described to accommodate Nemoura wahkeena Jewett. New descriptions and illustrations are provided for the adult male, adult female, and nymph collected at Wah- keena Spring, Columbia River George, Oregon. A diagnosis is furnished to distinguish the new genus from related genera of the subfamily Nemourinae.

Key words: Plecoptera, Nemouridae, Nanonemoura, Columbia River Gorge, Oregon.

Nearly 50 years ago Jewett (1954) named a nymph of this species. We discovered that the nemourid species from the Columbia River species does not live in the falls or even in the Gorge. He chose the name Nemoura wah- large spring source. Instead, it occupies some keena because he had collected his type series large seepage areas along the trail to the upper at Wahkeena Falls in 1947. Although he stated spring. that it was distinctive, he tentatively assigned Since that day in 1982, we have visited the it to the subgenus Zapada. In his monograph area several times and have always been able of the Pacific Northwest stoneflies (Jewett to find specimens at the seeps during the 1959), he added that it was rare and had been spring months. In addition, repeated attempts recorded only from Wahkeena Falls, but prob- have been made to find this stonefly in other ably had a wider distribution. Illies (1966) in localities in the gorge and especially along the his world catalog listed it in the genus Zapada, trail to, and at, nearby Multnomah Falls. How- which he had elevated along with the other ever, it still has not been found anywhere but subgenera of Nemouridae. Since no additional at the Wahkeena Falls site. specimens were available and the generic placement was questionable, Baumann (1975) Nanonemoura, new genus listed it under species incertae sedis. During the intervening 25-year period, it has been TYPE SPECIES.—Nanonemoura wahkeena, by included under the genus Zapada in the fol- monotypy. lowing publications: Stark et al. (1986), Stew- ADULTS.—General facies like a small grass- art and Stark (1988), and Stark et al. (1998). hopper nymph (Fig. 1). Body dark brown dor- However, because the true generic identity sally, lighter ventrally. Legs yellow, darker at was questionable, new material was obtained joints, proportionally long, especially hind for a detailed study. This remarkable little legs, which are more than twice the length of stonefly seems to be endemic to the Wahkeena abdomen. Wings very small, micropterous, and Falls springs, and it represents an undescribed seldom extending beyond thorax, veination genus in the subfamily Nemourinae. reduced to major veins, with few crossveins. Our search for specimens and additional Head with large, prominent eyes, antennae information on what is often called the Wah- long with ca 35 segments. Thorax stout with keena stonefly (Federal Register 1991) has oversized legs, wings found on dorsolateral been aided by our colleagues. Stan Jewett, margins. Abdomen as long as head and thorax before his untimely death in 1991, led the combined, cerci one segmented. authors and Riley Nelson to his study site at MALE GENITALIA.—Epiproct with large Wahkeena Falls, where we collected a single dorsal sclerite extending over entire dorsal

1Department of Zoology and Monte L. Bean Life Science Museum, Brigham Young University, Provo, UT 84602. 21274 NE 30th Lane, Gresham, OR 97030.

403 404 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Fig. 1. Nanonemoura wahkeena male adult, habitus.

surface (Fig. 7), lightly sclerotized apically and FEMALE GENITALIA.—Seventh sternum en- laterally at median expanded areas, covered larged into a pregenital plate, broadly rounded with numerous rows of small, wartlike processes and extending over most of eighth. Eighth (Fig. 6); ventral sclerite heavily sclerotized, sternum more darkly sclerotized, forming subequal in area to dorsal sclerite, and bearing small subgenital plate, with narrow, dark scle- ca 30 stout, ventrally directed spines that are rotized band over vaginal opening. Cerci randomly spaced over most of ventral surface membranous but large and long, almost like (Fig. 8). Paraprocts with heavily sclerotized tiny ears (Fig. 5). inner lobe that terminates in a sharply bifur- NYMPH.—Color uniformly brown, with cate apex; outer lobe partially membranous, darker pattern on head and thorax; head cov- with sclerotized outer margin, bearing ca 25 ered with numerous tiny hairs, eyes promi- short stout spines (Fig. 4). Hypoproct broadest nent, cerci well developed; thorax bearing at base, anterior 1/3 tapering to truncate apex, many long, darkly sclerotized spines, prono- bearing a thin, sharp, pointed projection tum completely encircled, meso- and meta- medially; vesicle present, thin and elongate notum bearing a pair of long, hairlike spines (Fig. 3). Cerci sclerotized dorsolaterally, drawn lateral to midline; legs short and stout, femur out into a narrow apex that ends in a hooked bearing 8–10 long spines scattered randomly tip (Fig. 2). on lateral margins, tibia with 2 rows of short, 2001] NEW OREGON STONEFLY GENUS 405

5 3

Figs. 2–5. Nanonemoura wahkeena adult terminalia: 2, male, dorsal; 3, male, lateral; 4, male, right paraproct; 5, female, ventral. stout spines, one on each lateral margin, a 3 lobes each, lobes arising from midlength to sparse fringe of long, thin hairs present on apex, not from a common stalk (Fig. 11). ventral margin (Fig. 10); abdomen without DIAGNOSIS.—Adults are easily distinguished setae or spines, except for 2 rows of long, thin from all other Nemourinae by their very long spines, one on each side of midline, running legs, especially those on metathorax, long max- entire length of abdomen; cerci with ca 20 illary palpi, and micropterous wings (Fig. 1). segments, intercalary spines present, anterior The male epiproct consists of dorsal and ven- 2/3 of cercal segments encircled with whorls tral sclerites that are large and flat and approx- of long spines (Fig. 9). Two cervical gills pre- imate each other in size, dorsal sclerite with sent on each side of midline, thin and divided, lightly sclerotized, slightly swollen areas medio- 406 WESTERN NORTH AMERICAN NATURALIST [Volume 61

and terminating in a large bifurcate process; male cerci are sclerotized and modified into long structures with a pointed apex; the female has a small, lightly sclerotized pregenital plate and long, thin cerci; the nymph has 2 pairs of gills with 3 branches each; spines on the 6 femur are sparse and randomly spaced rather than in definite rowlike whorls. Nanonemoura clearly belongs to the subfamily Nemourinae and would fit with the Nearctic genera Lednia and Visoka in the clad- ogram in Baumann (1975). Male terminalia of Nanonemoura are really quite similar to those of Lednia and Visoka. The epiprocts of all 3 genera are composed of large, nearly subequal dorsal and ventral sclerites. Nanonemoura and 7 Visoka both have rows of wartlike structures on their dorsal sclerite, while Lednia is covered by small spines. Their paraprocts consist of large, membranous outer lobes and thin, heavily sclerotized inner lobes. The inner lobes of Lednia and Visoka end in single pointed processes, while those of Nanonemoura have a birfurcate apex. Lednia lacks a vesicle on the hypoproct, but it is present in the other 2 genera. Nanonemoura and Visoka exhibit sclerotized, highly modified cerci, but those of Led- 8 nia are simple and unmodified. DISTRIBUTION.—Known only from tiny spring seeps along Wahkeena Creek in the Figs. 6–8. Nanonemoura wahkeena male, epiproct: 6, Wahkeena Falls area of the Columbia River dorsal; 7, lateral; 8, ventral. Gorge in northwestern Oregon. SPECIMENS EXAMINED.—Oregon, Multno- mah County, Wahkeena Creek near Wahkeena laterally, covered by rows of small, wartlike Falls, Columbia River Gorge: 5 April 1945, processes, ventral sclerite bearing ca 30 stout, S.G. Jewett, Jr., holotype , allotype , and 1 ventrally directed spines scattered over entire , 1 paratype (CAS, BYU); 16 April 1955, ventral surface (Figs. 6–8). Male paraprocts S.G. Jewett, Jr., 2 , 3 (USNM); 26 April with 2 well-developed lobes, inner lobe darkly 1955, Jewett & Wilson, 3 , 1 (ROM); 4 May sclerotized, narrow, and with bifurcate apex 1982, Baumann & Jewett, 9 , 4 (BYU); 29 (Fig. 4). Female with well-developed pregeni- February 1984, Baumann, Jewett, Nelson, & tal plate that covers most of next segment, Fiala, nymph (BYU); 29 March 1984, G.R. Fiala, small sclerotized bar over vaginal opening 1 (BYU); 17 April 1984, G.R. Fiala, 5 , 10 (Fig. 5). Nymph with 2 pairs of cervical gills, (BYU); 6 April 1985, G.R. Fiala, 8 , 5 each with 3 branches arising linearly and not (BYU); 9 April 1988, G.R. Fiala, 5 , 8 (BYU); from a common stalk (Fig. 11), leg setation 8 May 1988, G.R. Fiala, 4 , 4 (BYU); 23 mostly random on femora (Fig. 10). March 1992, G.R. Fiala, 8 , 2 (BYU); 23 Nanonemoura can easily be separated from March 2001, G.R. Fiala, 3 , 7 (BYU); 28 Zapada by using the following characters: the March 2001, G.R. Fiala, 9 , 11 (BYU). Most male epiproct lacks lateral knobs at the basal of the specimens are deposited at Brigham corners and consists of only dorsal and ventral Young University (BYU). sclerites that are distinctly flattened doroven- ETYMOLOGY.—The prefix nano is from the trally; male paraprocts are composed of 2 defi- Greek and is defined as small. Combining nite lobes, the inner being heavily sclerotized nano with nemoura, which signifies sylvan, 2001] NEW OREGON STONEFLY GENUS 407

Fig. 9. Nanonemoura wahkeena nymph, habitus. 408 WESTERN NORTH AMERICAN NATURALIST [Volume 61

National Museum (USNM) for the loan of specimens collected in 1955. Glenn Wiggins made his vial of specimens available when he was at the Royal Ontario Museum (ROM). Riley Nelson accompanied us twice when we collected specimens and was along the day we found the nymph. Ken Stewart and Bill Stark 10 are to be thanked for encouraging us to describe this genus so that it could be included in their revision of the North American stonefly nymph book. Finally, the excellent illustrations were made by Jean Stanger Leavitt.

LITERATURE CITED

BAUMANN, R.W. 1975. Revision of the stonefly family Nemouridae (Plecoptera): a study of the world fauna at the generic level. Smithsonian Contributions to Zoology 211:1–74. FEDERAL REGISTER. 1991. Endangered and threatened wildlife and plants; animal candidate review for listing as endangered or threatened species, proposed rule. Department of the Interior, Fish and Wildlife Service, 50 CFR Part 17, 56(225):58804–58836. 11 ILLIES, J. 1966. Katalog der rezenten Plecoptera. Walter de Gruyter and Company, Das Tierreich 82. 631 pp. JEWETT, S.G., JR. 1954. New stoneflies from California and Figs. 10, 11. Nanonemoura wahkeena nymphal struc- Oregon. Pan-Pacific Entomologist 30:167–179. tures: 10, right foreleg; 11, cervical area showing proster- ______. 1959. The stoneflies (Plecoptera) of the Pacific num and gills. Northwest. Oregon State Monographs, Studies in Entomology 3:1–95. STARK, B.P., S.W. SZCZYTKO, AND R.W. BAUMANN. 1986. results in a feminine name that means small, North American stoneflies (Plecoptera): systematics, distribution, and taxonomic references. Great Basin woodland stonefly. Naturalist 46:383–397. STARK, B.P., K.W. STEWART, S.W. SZCZYTKO, AND R.W. ACKNOWLEDGMENTS BAUMANN. 1998. Common names of stoneflies (Ple- coptera) from the United States and Canada. Ohio We are especially indebted to the late Stan Biological Survey Notes 1:1–18. STEWART, K.W., AND B.P. STARK. 1988. Nymphs of North Jewett, who shared his enthusiasm with us American stonefly genera (Plecoptera). Thomas Say and showed us where to collect his prized lit- Foundation, Entomological Society of America 12. tle stonefly. Thanks to Vincent Lee of the Cali- 460 pp. fornia Academy of Sciences (CAS), who helped Received 18 March 2001 us examine the holotype and allotype. We also Accepted 14 September 2001 appreciate Oliver Flint of the United States Western North American Naturalist 61(4), © 2001, pp. 409–416

SITE AND STAND CHARACTERISTICS RELATED TO WHITE PINE BLISTER RUST IN HIGH-ELEVATION FORESTS OF SOUTHERN IDAHO AND WESTERN WYOMING1

Jonathan P. Smith2 and James T. Hoffman3

ABSTRACT.—Successful infection of white pine species by white pine blister rust (WPBR) is contingent upon environmental conditions that are favorable to the spread and development of Cronartium ribicola. Site and stand factors related to this process have been studied elsewhere within the distribution of the disease, but few studies have concentrated on the high-elevation white pine forests of southern Idaho and western Wyoming. We found that mean summer precipitation, average tree diameter, and elevation were the most important variables in 3 logistic regression models of WPBR presence and intensity. The models were tested on a randomly chosen portion of our data set. The model with 9 variables correctly predicted categories of low-, moderate-, and high-disease incidence in 79% of cases. The 2 models with fewer variables had lower predictive efficiencies but were more parsimonious and generally easy to measure. The ability to use easily measured or remotely sensed site and stand characteristics to predict WPBR spread or intensification could be an important asset to land managers who need to decide where to focus disease mitigation efforts and predict disease effects on water quality, wildlife habitat, recreation potential, and other land-management activities.

Key words: white pine, whitebark pine, limber pine, white pine blister rust, Cronartium ribicola, tree diseases, Rocky Mountain forests, subalpine forests.

White pine blister rust disease (WPBR), spores to infect white pine needles. Fungal caused by the introduced fungus Cronartium hyphae spread into woody tissue causing ribicola, is the most widespread and serious cankers, where the 5th type of spore-bearing disease of Pinus albicaulis (whitebark pine; structure, the pycnium, is produced. Upon Arno and Hoff 1989) and P. flexilis (limber pine) completion of the pycnial stage, which proba- in the Rocky Mountains (Smith and Hoffman bly involves mating, aecia are produced, com- 2000). The disease is also a potential threat to pleting the life cycle. most, if not all, other white pine species (genus Like other pine rusts, transmission of spores Pinus, subgenus Strobus, section Strobus, sub- and host infection depends on a favorable sections Cembrae and Strobi, and section Par- temperature and moisture environment, an rya, subsection Balfourianae; Hoff et al. 1980). abundance of spores (inoculum), and availabil- The rust causes branch and stem cankers that, ity of susceptible hosts (Mielke 1943, Charlton in most cases, girdle and kill the host tree. 1963). These conditions may be affected by Cronartium ribicola has a complex life cycle physical factors such as slope, aspect, eleva- that is characterized by 5 spore-producing tion, and precipitation, as well as biological stages that alternate infection between white factors such as structure of the forest canopy pine species and plants of the genus Ribes and proximity of Ribes spp. (currants and gooseberries). Aeciospores are Site and stand factors associated with rust small, light spores that are produced on pine incidence have been identified by studying cankers and can travel long distances to infect the distribution of WPBR and endemic pine the leaves of Ribes. Urediniospores emerge rusts. Van Arsdel (1972) found that the size of on Ribes leaves and spread to other leaves on forest canopy openings and certain topographic the same plant, or other nearby Ribes plants. features were related to WPBR incidence in Teliospores, produced on Ribes, germinate and Pinus strobus (eastern white pine). In British form the basidium, which releases basidio- Columbia, Hunt (1983) reported more WPBR

1This article was originally presented as part of the 5th Biennial Scientific Conference on the Greater Yellowstone Ecosystem held at Yellowstone National Park in October 1999. The conference was entitled “Exotic Organisms in Greater Yellowstone: Native Biodiversity Under Siege.” The July 2001 issue of the WESTERN NORTH AMERICAN NATURALIST (Volume 61, No. 3) published presentations made at the conference. Three additional articles are included in this issue. 2Northern Arizona University, School of Forestry, PO Box 15018, Flagstaff, AZ 86011. 3USDA Forest Service, Forest Health Protection, 1249 S. Vinnell Way, Boise, ID 83709.

409 410 WESTERN NORTH AMERICAN NATURALIST [Volume 61 cankers in P. monticola (over 2.5 m high in the also several disjunct P. flexilis and P. albicaulis tree) as slope increased. Jacobi et al. (1993) populations in isolated mountain ranges of found Cronartium comandrae (comandra blister eastern Oregon and northern Nevada, and both rust) incidence in Pinus contorta subsp. latifolia species occur in the Sierra Nevada (Critch- (lodgepole pine) positively correlated with tree field and Little 1966). Our study area encom- diameter, and negatively correlated with stand passes those Rocky Mountain white pine pop- density and distance to the rust’s telial host. ulations that lie within southern Idaho and Beard et al. (1983) found a greater incidence western Wyoming (Fig. 1). Within this region of C. coleosporioides (stalactiform blister rust) P. albicaulis and P. flexilis populations extend in central Idaho Pinus contorta forests at mid- upward from the lower subalpine zone to the dle to upper elevations, and in Abies lasiocarpa/ upper (cold) tree line. Pinus flexilis also has Xerophyllum tenax and Abies lasiocarpa/Vac- the unique ability to grow at lower (dry) tree cinium scoparium habitat types. Endocronar- line (Arno and Hammerly 1984). tium (=Peridermium) harknessii (western gall rust) stem infections were negatively corre- FIELD METHODS lated with stand age in British Columbia Pinus contorta forests (van der Kamp 1988). Van Ars- In 1995 we installed 10 rectangular plots del (1965) constructed a formula based on according to the methods specified by the slope and canopy openings and predicted Whitebark Pine Monitoring Network (Kendall WPBR presence in southwestern Wisconsin 1995). In 1996 we used randomly located strip with 89% accuracy. Charlton (1963) used aspect, transects rather than rectangular plots to delin- elevation, slope, topographic position, and vege- eate trees. We switched to transects because tation structure, along with climatic factors, to white pine species in our study area tend to assess WPBR infection hazard in the eastern grow as dispersed woodlands or as infrequent U.S. A comprehensive site-specific WPBR haz- seral components in subalpine forests. Obtain- ard model based on site, stand, alternate host, ing 50 white pines in a rectangular plot of a and physiological factors was developed by reasonable size was often not possible. For the McDonald et al. (1981) for P. monticola in 68 sites sampled during 1996, we established northern Idaho. a 4.6-m (15-ft)-wide strip transect, along the Very little of this type of work has been contour of the slope, from a random starting conducted in the southern portion of C. ribi- point. We traversed this transect until 50 white cola’s range in the Rocky Mountains because, pines had been inspected or until we encoun- historically, disease surveys revealed only trace tered a change in the character of the site or levels of infection (Brown 1967, Brown and stand that did not match our sampling criteria, Graham 1969). However, WPBR has recently such as a different canopy structure, a suffi- intensified and spread to new locations in the ciently different aspect (>10° difference), slope southern portions of the Northern Rocky Moun- (>5% difference), habitat type or phase, or a tain and Middle Rocky Mountain provinces topographic change. Rather than cross this (Kendall et al. 1996, Smith and Hoffman 2000). environmental gradient, we changed the direc- As an initial step in modeling WPBR spread tion of the transect by 180°, displaced it uphill and intensification in this region, we used USDA or downhill 4.6 m (15 ft), and continued to Forest Service disease survey data (Smith and sample until 50 trees had been inspected. Hoffman 1998) to look for relationships between For each tree we recorded the presence of WPBR incidence and several site and stand WPBR cankers and DBH (diameter at breast characteristics. height, 1.37 m above the ground), in 5.1-cm (2-in) size classes. At the midpoint of each tran- STUDY AREA sect, we measured or calculated habitat type (Steele et al. 1981, 1983), presence/absence of Pinus albicaulis and P. flexilis populations in Ribes sp., basal area, trees per hectare, canopy the U.S. extend southward along the Rocky closure, elevation, aspect, slope angle, and Mountains from the Canadian border to south- topographic position (Table 1). eastern Idaho and southwestern Wyoming. An additional variable, estimated mean sum- Pinus flexilis extends even further south, mer precipitation, was generated from climate throughout the mountains of Utah. There are maps (Martner 1986, Molnau and Newton 2001] WHITE PINE BLISTER RUST ECOLOGY 411

Fig. 1. Northern and middle Rocky Mountain ranges, sample locations, and white pine blister rust intensity for 78 sample sites inspected in 1995–1996. Distribution of white pine species (shaded areas) derived from Little (1971).

1994). To estimate mean summer precipita- analysis. We grouped the categorical indepen- tion, we multiplied regionalized estimates of dent variables, habitat type, canopy cover, and the summer (June, July, and August) propor- topography to reduce the number of cate- tion of total precipitation by the mean annual gories for model calculation. For example, we precipitation values taken from these maps. identified 20 habitat type classes in the field We interpolated precipitation values between but combined these into 4 categories based isohyetal contours for each of our sample sites. on a multidimensional scaling procedure that groups habitat series based on moisture re- STATISTICAL METHODS quirements of understory plants (McDonald unpublished data). We used dummy coding for Because sampling location criteria and data the categorical variables. Presence or absence collection procedures were identical for plots of Ribes spp. was entered as a binary variable and transects, the data were combined for our (i.e., a value of 0 for absent, 1 for present). 412 WESTERN NORTH AMERICAN NATURALIST [Volume 61

TABLE 1. Variables used in the stepwise logistic regression procedure and parameter estimates for the 3 models using the recombined (full) data set. χ2 ______Standardized parameter estimate (Wald ) Variable (description) Model 1 Model 2 Model 3 Elevation (meters, from topo map) –0.613 (313) –0.606 (321) –0.546 (308) Average DBH (diameter at breast height in 2-cm classes) 0.458 (212) 0.454 (210) 0.517 (301) Mean summer precipitation (interpolated from maps) 0.278 (109) 0.304 (145) 0.250 (111) Ribes (present/absent in stand) 0.237 (62) 0.239 (69) Stand density (trees ⋅ ha–1, all species) –0.258 (61) –0.222 (55) Topographic position 3 (upper slopes) 0.215 (56) 0.212 (55) Habitat type 1 (wet subalpine fir group) 0.123 (28) Slope (percent) –0.109 (12) –0.118 (15) Habitat type 4 (whitebark pine series) 0.007 (7) Basal area (m2 ⋅ ha–1, all species, white pines only) Aspect (cosine of aspect in degrees) Latitude (UTM–northing) Canopy (open, broken, closed) Topography (valley/lower, mid, upper, ridge) Habitat type (wet subalpine fir, cool/moist subalpine fir, cold/dry subalpine fir, whitebark series, Douglas-fir)

Elevation, slope, latitude (UTM northing), aver- predictor variables contribute no more than age tree size, and mean summer precipitation chance to the explanation of the dependent were entered as continuous variables (i.e., variable with the Gm statistic (the model chi- their actual measured values). The aspect square statistic). The Bayesian information cri- measurement, which is azimuthal (circular terion (BIC) was calculated as a selection de- data), was linearized by taking the cosine of vice because it emphasizes parsimony by penal- the aspect in radians. izing models with a large number of parameters (Ramsey and Schafer 1997). To measure Logistic Regression Analysis the predictive efficiency of each model, we Percent of trees infected in a sample stand arbitrarily assigned broad classes of WPBR was the dependent variable, which was treated incidence, low (<25% incidence), medium statistically as the number of successful events (26–50%), and high (>50%), to the observed (infected trees) per number of trials (trees and predicted values and then calculated how sampled) at each sample site. We performed a frequently each model correctly predicted the stepwise logistic regression procedure with observed category, was 1 category off, or was the model development data set using the off by 2 categories. PROC LOGISTIC STEPWISE option in SAS Model Testing (SAS Institute Inc. 1996). This procedure iden- We used a split-sample validation tech- tifies predictive variables when the number of nique to develop and test the logistic regres- potential explanatory variables is large relative sion models. Each record was assigned a ran- to the number of samples (Hosmer and Leme- dom number, sorted by this number, and then show 1989). We constructed 3 models of WPBR split into a model-development data set (2/3 of incidence with combinations of the variables the data), which was used to develop the mod- selected by the stepwise procedure. els. The remaining 1/3 of the data (n = 23) was To determine if models were statistically treated as an independent data set to test the significant, we compared 4 criteria to assess models’ statistical significance, fit, and predic- how well the models fit and to compare how tive efficiency, and to assess the importance of well each model predicted WPBR incidence. the independent variables. We estimated the 2 First, we calculated r L, which is a measure of predicted proportion of trees infected in each the reduction in the log-likelihood as a result sample with the predicted probability of infec- of including the independent variables (Menard tion (presence or absence of WPBR) for each 1995). We tested the null hypothesis that the tree in that sample. 2001] WHITE PINE BLISTER RUST ECOLOGY 413

TABLE 2. Fit statistics and prediction efficiency for 3 logistic regression models of WPBR incidence using the model- developing data set (A) and the model-validation data set (B).

A ______Prediction of incidence category χ a b 2c 2 d e f Model ’s GM P R R L BIC n Correct Under 1 Over 1 Under 2 Over 2 1 9 502 0.001 0.483 0.179 2326 49 37(76%) 3 5 2 2 2 7 487 0.001 0.473 0.173 2339 49 37(76%) 5 4 2 1 3 3 370 0.001 0.347 0.129 2502 50 33(66%) 8 7 2 0

B ______Prediction of incidence category Model χ’s Pb R2c nf Correct Under 1 Over 1 Under 2 Over 2 1 9 0.001 0.448 24 19(79%) 3110 2 7 0.001 0.384 24 18(75%) 3210 3 3 0.001 0.405 24 15(62%) 6300 aModel chi-square bStatistical significance of model cCoefficient of determination dReduction in log-likelihood due to the model eBayesian information criterion fNumber of observations; differences due to missing values for some variables

The probability of WPBR infection in a Analysis of the Independent tree [P(Y)] was obtained by inserting the test Variables in the Model data independent variables into the equation To assess the importance of the indepen- for each model. The equations calculated logit(Y) dent variables, we evaluated the odds ratio, (the natural logarithm of the odds of WPBR which approximates how much more likely infection) rather than P(Y) directly. The form of the event (WPBR presence in a tree) becomes the equation was with increases or decreases in the value of each independent variable (SAS 1996). We βˆ βˆ × βˆ × βˆ βˆ × logit(Y) = 0 + 1 x1 + 2 x2 +..... + k xk, also used the standardized logistic regression βˆ coefficients to evaluate the strength of the where logit(Y) = ln {P(Y)/[1 – P(Y)]}, 0 is the relationship between each independent vari- Y-intercept, x1 through xk are the independent able and the dependent variable (Menard variables identified by the stepwise procedure 1995). as important predictors of WPBR incidence, and βˆ through βˆ are the coefficients for these 1 k RESULTS independent variables. It was necessary to lin- earize the predicted value to compare it to the Model Development linear observed proportion of trees infected. and Validation To accomplish this, logit(Y) was converted to The stepwise logistic regression identified odds(Y) by exponentiation, and then to P(Y) 13 variables that were potentially related to by the formula P(Y) = odds(Y)/[1 + odds(Y)], WPBR infection. We used these to develop 3 where P(Y) is the predicted probability of infec- candidate models. For the 1st model we re- tion in an individual tree and odds(Y) is the moved 4 variables that were highly correlated ratio of the probability that Y = 1 to the prob- (r > 0.6) or that were not significant (P > ability that Y ≠ 1. 0.05). We created the 2nd model by removing We used least-squares regression to com- the variables with Wald-χ2 values <20. The pare the predicted proportion of trees infected 3rd model contained only the 3 variables that with our observed proportion of infected trees stood apart from the others because of their and to calculate the significance of the regres- very high Wald-χ2 values (>100). Fit statistics, sion and the coefficient of determination. significance, and predictive efficiency for the Finally, we assigned the low, medium, and 3 models are shown in Table 2A. high classes to the predicted and observed When applied to the validation set, all 3 values and performed a simple error assess- regression models were statistically significant ment to see how well the model predicted (P ≤ 0.001). The coefficient of determination incidence. (r2) for the models ranged from 0.38 to 0.45. 414 WESTERN NORTH AMERICAN NATURALIST [Volume 61

The level of classification accuracy was highest regional moisture characteristics. For example, for model 1, which correctly classified 79.2% Van Arsdel et al. (1956) attributed low WPBR of the cases. Model 3 correctly classified 62.5% incidence in southwestern Wisconsin to the of the test cases and had a higher r2 than dry climate of the region. model 2 (Table 2B). Optimal temperature and moisture conditions for survival of Cronartium ribicola have Importance of Independent been well documented (Mielke 1943, Van Ars- Variables in the Model del et al. 1956). Infection of pines requires The most important variables in all 3 models extended periods of time (Charlton 1963) dur- were elevation, mean summer precipitation, ing late summer and early autumn with night- and average DBH. Although other variables time temperatures below 19.4°C (67°F) and free were also statistically significant, when com- moisture on the needle surfaces (Kimmey and bined, these variables accounted for a much Wagener 1961). Van Arsdel et al. (1956) con- smaller proportion of the variation in WPBR cluded that at least 2 consecutive days of these incidence than the first 3 variables. Table 1 favorable conditions are required for infection lists the parameter estimates for the variables of pines. in each model. Extended temperature data from high-elevation weather stations within our study area DISCUSSION were not available, and interpolating temperatures between low-elevation weather stations Interpretation of Independent is inappropriate because of local temperature Variable Selection inversions that are common in mountain envi- ELEVATION.—In Yellowstone National Park, ronments (Baker 1944). Thus, we did not include Berg et al. (1975) reported that WPBR inci- a temperature variable in our analysis. We were dence in Pinus albicaulis and Pinus flexilis also unable to locate climate data for moun- decreased with elevation. These researchers tainous areas that included summer moisture found that 92% of all infections occurred estimates other than mean precipitation below 2591 m (8500 ft) elevation. Our results amounts. It is generally thought that moist suggest a similar negative relationship between summers are conducive to WPBR development elevation and WPBR incidence. We found that and spread; however, mean summer precipita- 97% of the sample stands below 2591 m had tion alone is probably not the best indicator of WPBR, while only 53% of the stands above favorable climate conditions. For example, 2591 m were infected. However, the average Mielke (1943) noted that a heavy “flare up” of proportion of trees infected in these stands WPBR incidence occurred in Idaho during a did not decrease with elevation. In fact, the summer of relatively low mean precipitation proportion of high-infection sites above 2591 in 1937. In fact, dew may be an equally impor- m, 33%, was slightly greater than the propor- tant source of moisture (Mielke 1943). Cloudy tion of high-infection sites below this eleva- summer periods and high relative humidity tion, 31%, suggesting that once WPBR is able periods may be better indicators of WPBR to infect a high-elevation site, it is able to con- incidence than precipitation. tinue to intensify. However, this phenomenon AVERAGE TREE DIAMETER.—The importance was apparent only in the Greater Yellowstone of average tree diameter at breast height (DBH) Ecosystem portion of our study area. Some fac- in the logistic regression model may be due to tors involved in the decrease in WPBR inci- 2 factors. First, smaller-diameter trees tend to dence with increasing elevation may include have less foliage than larger-diameter trees earlier Ribes leaf senescence, cooler tempera- and are therefore smaller targets for spores. tures at key times of development or spore Second, most cankers we inspected were in dispersal, less susceptible Ribes species, or a the upper portion of tree crowns in the inte- less favorable spatial pattern of hosts at higher rior of stands or throughout the crown of trees elevations. on an open edge of the stand. We speculate PRECIPITATION.—Mean summer precipita- that wind patterns during times of basidio- tion was an important predictor variable in spore dispersal from Ribes to pines concen- our model. Other researchers have observed a trate infections along the windward and upper relationship between WPBR incidence and sides of a stand. Wind-dampening effects of 2001] WHITE PINE BLISTER RUST ECOLOGY 415 the forest canopy and screening of spores by Management Implications larger trees may reduce the transfer of spores The ability to identify areas of potential rapid to smaller, more sheltered trees. intensification or areas with a low probability While diameter could reflect the length of of infection or intensification over time would exposure, the length of exposure is probably help land managers direct mitigation efforts. not important because even the smallest trees For example, a spatial model that identifies in our samples likely pre-date WPBR presence these areas of intensification could aid the on- in the region. going search for phenotypically resistant trees, Average DBH appears to be more impor- which are highly visible in severely infected tant to the intensity of infection on sites that stands. In some areas vegetative competition are infected than to WPBR incidence. Of 16 from Abies lasiocarpa (subalpine fir) is as much stands with an average DBH of <10 cm, all of a concern as WPBR (Keane et al. 1994). A but 2 were infected, with an average infection spatio-temporal WPBR spread and intensifica- level of 19.5% (2–85%) for the infected stands. ≥ tion model would help managers decide where Of 14 stands with 20 cm average DBH, 4 were treatments to reduce this competition would uninfected, and the average infection rate for be effective. Where WPBR intensification prob- the infected stands was 46.3% (2–87%). ability is low, silviculture and/or prescribed OTHER VARIABLES.—Other variables were fire could be used to reduce competition and statistically significant in the stepwise logistic provide regeneration opportunities for white regression analysis. However, these variables pines. Conversely, conducting these activities χ2 had much lower Wald- values and con- in areas with a high probability of WPBR tributed proportionally much less to explain- intensification could potentially increase inocu- ing observed variability in incidence than ele- lum levels through the regeneration of suscep- vation, mean summer precipitation, and aver- tible white pines or an inadvertent increase in χ2 age tree diameter. Due to the low Wald- val- Ribes abundance. A predictive model could ues and potential correlations between these also help resource planners assess the future variables, their statistical and biological signif- effects of white pine mortality on wildlife, water icance is suspect. Also, since we did not test quality and quantity, avalanche activity, and each of the independent variables, it is possi- recreation. ble that we included irrelevant variables in the model. ACKNOWLEDGMENTS Implications for Future Research Funding for this project was provided by the USDA Forest Service, the Stillinger Grant The potential relationships between site for Forest Pathology Research, and the Univer- and stand characteristics that we identified in sity of Idaho. We thank the Forest Health Pro- this analysis represent a “snapshot” in time for tection Office staff, especially Russ Hirschler, the current stage of the developing WPBR epi- Bob Albano, Phil Mocettini, Dick Halsey, and demic in our study area. These relationships many other Forest Service, Bureau of Land help identify areas where WPBR will likely Management, National Park Service, and Idaho spread and/or intensify first. Aging cankers Fish and Game personnel for their help with could help researchers (1) determine how data collection. Kirk Steinhorst and Chris WPBR has moved and intensified in the region Williams provided valuable advice on statisti- and (2) differentiate between sites susceptible cal methods. We also thank Kara Leonard, Leon to long-range transmission and those where Neuenschwander, Art Partridge, and Maurey WPBR intensifies quickly. Such a study could Wiese for their support and many helpful dis- also help researchers predict future spread cussions. and intensification of WPBR in the region. However, the characteristics of spread and LITERATURE CITED intensification may change in the future due to genetic adaptations by Cronartium ribicola, ARNO, S.F., AND R.J. HAMMERLY. 1984. Timberline: moun- an exponential increase in inoculum availabil- tain and arctic forest frontiers. The Mountaineers, Seattle, WA. 304 pp. ity, changes in host distributions, or shifts in ARNO, S.F., AND R.J. HOFF. 1989. Silvics of whitebark pine regional climate patterns. (Pinus albicaulis). U.S. Department of Agriculture, 416 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Forest Service, Intermountain Research Station LITTLE, E.L., JR. 1971. Atlas of United States trees. Vol- General Technical Report INT-253. 11 pp. ume 1. Conifers and important hardwoods. Miscella- BAKER, F.S. 1944. Mountain climates of the western United neous Publication 1146, U.S. Department of Agricul- States. Ecological Monographs 14:225–254. ture, Forest Service, Washington, DC. 200 pp. BEARD, T.H., N.E. MARTIN, AND D.L. ADAMS. 1983. Effects MARTNER, B.E. 1986. Wyoming climate atlas. University of habitat type and elevation on occurrence of stalac- of Nebraska Press, Lincoln. Prepared in cooperation tiform blister rust in stands of lodgepole pine. Plant with the Wyoming Water Research Center, Univer- Disease 67:648–651. sity of Wyoming. Map. BERG, M.J., C.E. CARLSON, L.P. LOUNSBURY, AND O.J. DOOL- MCDONALD, G.I., R.J. HOFF, AND W. W YCOFF. 1981. Com- ING. 1975. Effectiveness of Ribes eradication in Yel- puter simulation of white pine blister rust epidemics. lowstone National Park: study establishment and sta- I. Model formulation. U.S. Department of Agriculture, tus report. U.S. Department of Agriculture, Forest Forest Service, Intermountain Research Station Service, Northern Region, State and Private Forestry Research Paper INT-258. 77 pp. Report. 15 pp. MENARD, S. 1995. Applied logistic regression analysis. In: BROWN, D.H. 1967. White pine blister rust survey in M.S. Lewis-Beck, series editor, Sage Publications Montana and Wyoming: 1966. U.S. Department of Series 7, No. 106, Sage University Papers, Thousand Agriculture, Forest Service, Northern Region, State Oaks, CA. 98 pp. and Private Forestry Report. 11 pp. MIELKE, J.L. 1943. White pine blister rust in western BROWN, D.H., AND D.A. GRAHAM. 1969. White pine blis- North America. Yale University School of Forestry ter rust survey in Wyoming, Idaho, and Utah: 1967. Bulletin 52. 155 pp. U.S. Department of Agriculture, Forest Service, MOLNAU, M., AND C. NEWTON. 1994. Mean annual precip- Northern Region, State and Private Forestry Report. itation and variability in Idaho. Unpublished paper 11 pp. presented at the 1994 Western Snow Conference. CHARLTON, J.W. 1963. Relating climate to eastern white RAMSEY, F.L., AND D.W. SCHAFER. 1997. Statistical sleuth. pine blister rust infection hazard. U.S. Department Duxbury Press, Belmont, CA. 742 pp. of Agriculture, Forest Service, Eastern Region Report. SAS INSTITUTE INC. 1996. SAS/STAT software: changes 38 pp. and enhancements through release 6.11. SAS Insti- CRITCHFIELD, W.B., AND E.L. LITTLE, JR. 1966. Geographic tute Inc., Cary, NC. distribution of the pines of the world. U.S. Depart- SMITH, J.P., AND J.T. HOFFMAN. 1998. Status of white pine ment of Agriculture, Forest Service, Miscellaneous blister rust in Intermountain region white pines. Publication 991. 97 pp. U.S. Department of Agriculture, Forest Service, HOFF, R.J., R.T. BINGHAM, AND G.I. MCDONALD. 1980. Intermountain Region, State and Private Forestry Relative blister rust resistance of white pines. Euro- Report R4-98-02. 24 pp. pean Journal of Forest Pathology 10: 307–316. ______. 2000. Status of white pine blister rust in the HOSMER, D.W., AND S. LEMESHOW. 1989. Applied logistic Intermountain West. Western North American Nat- regression. Wiley, New York. 307 pp. uralist 60:165–179. HUNT, R.S. 1983. White pine blister rust in British Colum- STEELE, R., S.V. COOPER, D.M. ONDOVE, D.W. ROBERTS, bia: II. Can stands be hazard rated? The Forestry AND R.D. PFISTER. 1983. Forest habitat types of east- Chronicle, Canadian Forestry Service, Pacific Forest ern Idaho–western Wyoming. U.S. Department of Research Centre. 5 pp. Agriculture, Forest Service, Intermountain Forest JACOBI, W.R., B.W. GEILS, J.E TAYLOR, AND W.R. ZENTZ. and Range Experiment Station General Technical 1993. Predicting the incidence of comandra blister Report INT-144. 122 pp. rust on lodgepole pine: site, stand, and alternate- STEELE, R., D. PFISTER, A.R. RUSSELL, AND J.A. KITTAMS. host influences. Phytopathology 83:630–637. 1981. Forest habitat types of central Idaho. U.S. KEANE, R.E., P. MORGAN, AND J.P. MENAKIS. 1994. Land- Department of Agriculture, Forest Service, Inter- scape assessment of the decline of whitebark pine mountain Forest and Range Experiment Station (Pinus albicaulis) in the Bob Marshall Wilderness General Technical Report INT-114. 138 pp. Complex, Montana, USA. Northwest Science 68: VAN ARSDEL, E.P. 1965. Micrometeorology and plant dis- 213–229. ease epidemiology. Phytopathology 55:945–950. KENDALL, K.C. 1995. June 29, 1995, memorandum on ______. 1972. Environment in relation to white pine blis- monitoring methods, field forms, and instructions for ter rust infection. Pages 479–493 in Proceedings of a the Whitebark Pine Monitoring Network. On file at NATO-IUFRO Advanced Study Institute. Biology of U.S. Department of Agriculture, Forest Service, For- rust resistance in forest trees. U.S. Department of est Health Protection Office, Boise, ID. Agriculture, Forest Service Miscellaneous Publica- KENDALL, K.C., D. SCHIROKAUER, E. SHANAHAN, R. WATT, tion 1221. D. REINHART, R. RENKIN, S. CAIN, AND G. GREEN. VAN ARSDEL, E.P., A.J. RIKER, AND R.F. PATTON. 1956. The 1996. Whitebark pine health in northern Rockies effects of temperature and moisture on the spread of national park ecosystems: a preliminary report. Page white pine blister rust. Phytopathology 46:307–318. 16 in R.E. Keane, editor, U.S. Department of Agri- VAN DER KAMP, B.J. 1988. Temporal and spatial variation in culture, Forest Service, Intermountain Research Sta- infection of lodgepole pine by western gall rust. tion, Nutcracker Notes 7. Plant Disease 72:787–790. KIMMEY, J.W., AND W. W. W AGENER. 1961. Spread of white pine blister rust from Ribes to sugar pine in Califor- Received 10 March 2000 nia and Oregon. U.S. Department of Agriculture, Accepted 17 July 2001 Forest Service, Pacific Southwest Forest and Range Experiment Station Report. 71 pp. Western North American Naturalist 61(4), © 2001, pp. 417–427

EXOTIC PLANTS IN EARLY AND LATE SERAL VEGETATION OF FIFTEEN NORTHERN ROCKY MOUNTAIN ENVIRONMENTS (HTs)1

T. Weaver2,4, D. Gustafson2, and J. Lichthardt3

ABSTRACT.—We determined the capacity of exotic plants to invade major environmental types of the northern Rocky Mountains. We did this by observing their presence on disturbed and undisturbed sites in relatively well inoculated locations—corridors adjacent to highways—on transects across the mountains in Glacier National Park and Grand Teton National Park and on low-altitude sites between them. We draw 3 primary conclusions. First, of 29 exotics commonly found, the most dominant are intentionally introduced grasses (Agrostis, Bromus, Dactylis, and especially Phleum pratense and Poa pratensis) and legumes (Melilotus, Medicago, and Trifolium) rather than the forbs more often listed as noxious. Second, in the environmental types studied, disturbed sites are invasible, except in the alpine. Third, invasion of undisturbed sites declines from grasslands and open forests to alpine to moist forests. This gradient probably represents a decline in resource (light, water, nutrients) availability for herbs, except in the alpine, where a physical limitation is suggested by the poor performance of exotics on noncompetitive disturbed sites.

Key words: exotic plants, aliens, weeds, Phleum pratense, Poa pratensis, Bromus inermis, Bromus tectorum, Tri- folium, Melilotus, Centaurea, Chrysanthemum leucanthemum, habitat types, environmental types, Bouteloua gracilis, Agropyron spicatum, Artemisia tridentata, Pseudotsuga menziesii, Abies lasiocarpa, mountain meadows, alpine, seral stages, disturbance, climax, northern Rocky Mountains, Grand Teton National Park, Glacier National Park, Yellowstone National Park.

National forests and parks have a mandate 1980, Steele et al. 1983). And, within each of to manage against exotic plants both in their these, one compares sites on the competitive charters (U.S. Congress 1872) and recent exec- spectrum from freshly disturbed (noncompeti- utive directives (Clinton 1994, 1999). tive) to late seral (very competitive; Grime Management of exotics requires their iden- 1979, Despain 1990). tification. Plants exotic to specific regions (e.g., The objectives of this paper are thus to list Whitson 1992) and management units (e.g., the common exotics of the northern Rocky Whipple 2001) have often been listed to facili- Mountains, to provide separate lists of the tate recognition and identification. A listing by exotics present in major upland environmental ecological zones within a region would refine types of the region, and to compare exotic this capacity. presence in an early (less competitive) and late In addition, a listing by environmental types (more competitive) seral stage in each envi- within a region would provide a key to environmental type. A companion paper extends ronments (or sites) the plant might invade or our results to separately map the potential dis- might already have invaded. Identification of tribution of major exotics on disturbed and occupiable environmental types will enable undisturbed sites in Yellowstone National Park managers to concentrate control efforts in a (Despain et al. 2001). fraction of the management area. Two environmental qualities are important. First, one con- METHODS siders environmental types (defined in Meth- ods), determined by physical characteristics Our term environmental type is synonymous such as climate and substrate (Holdridge 1947, with Daubenmire’s (1968a, 1968b, 1970) habi- Daubenmire 1968, 1970, Whittaker 1975) and tat type (HT). (1) An environmental type (ET indicated locally by potential natural vegeta- = HT) includes all environments (equivalent, tion (Pfister et al. 1977, Mueggler and Stewart but not identical) capable of supporting a climax

1Presented at the 5th Biennial Scientific Conference on the Greater Yellowstone Ecosystem, October 1999. See footnote 1 on p. 409. 2Ecology Department, Montana State University, Bozeman, MT 59717. 3Conservation Data Center, Idaho Department of Fish and Game, Boise, ID 83707. 4Corresponding author.

417 418 WESTERN NORTH AMERICAN NATURALIST [Volume 61 association (series of plant communities of the To complete the list of exotics likely to invade same kind). (2) The concept is useful because our region, we needed to sample other major it groups discrete sites (or environments as environmental types representative of the opposed to plant communities) likely to respond environmental gradient from steppe upward similarly to similar managment (Daubenmire through forest to the alpine. Thus, we sampled 1968b, 1976, Pfister et al. 1977, Mueggler and 15 environmental types along highways cross- Stewart 1980). (3) The term environmental type ing the mountains in Glacier and Grand Teton is preferred because it clarifies the concept national parks (1984–1985) and in intervening that the types represent physical environment lowland areas (in 1986). We identified major (e.g., moisture and temperature) rather than environmental types by using late seral vege- habitat for a particular organism, since the lat- tation as indicators (Holdridge 1947, Dauben- ter “habitat type” may either extend across mire 1968a, 1968b, 1970, Whittaker 1975). several environmental types (e.g., wide-rang- While our exotic lists for major environmental ing plants in Table 3) or may not exist in the types approach completeness, our regional list organism’s optimal environmental type (e.g., if is incomplete because we omitted less wide- vegetation of the seral stage present provides spread types such as those along streams or on too much competition [Walter 1960] or fails to unusual substrates. The environmental types provide necessary nutrients, cover, or struc- (HTs) sampled are listed, in altitudinal order, ture). Daubenmire (personal communication) in Table l, with abbreviations, general loca- recognized the environmental type/habitat type tions, and sample size (~10). Underlying confusion—especially among zoologists—and changes in climate and soils along the gradient wished he had called his types “environmental are compared in Table 2, as well as by Dauben- types.” mire (1968a, 1970), Pfister et al. (1977), Mueg- To determine which exotics might invade a gler and Stewart (1980), and Steele et al. (1983). specific environmental type (e.g., a montane Thus, our sample design included 15 envi- environment occupied by Pseudotsuga men- ronmental types (HTs), 2 treatments reported ziesii/Symphoricarpos albus at climax), we here (and 3 others [Weaver et al. 1993]), and needed to observe exotic colonization of well- approximately 10 replications (sites). Vegeta- inoculated sites in that type. Thus, we sampled tional characteristics of each of the approxi- several (7–11) sites jointly in that type and mately 800 sites studied were recorded with near a major highway that has long delivered measures of presence, frequency, and cover of seed to it (Table 1). both native and exotic species present. (1) To determine which “seral” vegetation types Presence was recorded by listing all exotic and in that environmental type could be invaded, native plant species present in a 1 × 25-m plot we needed to compare invasion of highly dis- representative of the zone and parallel to the turbed (low competition), less disturbed (early highway traveled. We separately noted any seral), and high competition (late seral) sites other species present in adjacent similar vege- (Grime 1979) occupying that physical environ- tation. Natives in the plots, not discussed here, ment. Thus, we sampled relatively gentle road are listed in Weaver et al. (1993). (2) Cover of cuts, logged right-of-way (not reported here), a species was measured by recording the per- and nearby undisturbed vegetation at each of centage of 75 points covered by that species. the 7–11 sites studied. The fact that our work The 75 points were located by lowering 3 pins was primarily in national parks facilitated loca- into the vegetation in each meter point along tion of undisturbed sites adjacent to highly the plot’s center line. Cover was integrated disturbed sites. The relatively low establish- over a type by averaging cover measurements ment of a species on a late seral site, perhaps across sites, but only at sites where the species 20–30 m away, is attributed to competition but occurred. We omitted unoccupied plots in could also be due to failure to disperse. We these calculations to measure the success of attribute most of the deficiency to competition, species at sites where they did occur. If however, both because differences in distances desired, cover values for the environmental from the highway are short and because dis- zone as a whole can be calculated by multiply- persal is a characteristic selected for in oppor- ing cover values presented by the associated tunistic species. constancy value; this will correct the cover 2001] EXOTIC PLANTS OF ROCKY MOUNTAIN ENVIRONMENTS 419

TABLE 1. Environmental types (HTs), locations, and sizes of samples in which exotic distributions were observed. Environmental types are listed in approximate order of altitude, from low to high. Environmental type (HT)a Abbreviationb Locationc Sampled

GRASSLANDS/ SHRUBLANDS Stipa comata/ Bouteloua gracilis STCO/ BOGR Broadwater MT 7 Agropyron spicatum/ Bouteloua gracilis AGSP/ BOGR Broadwater MT 8 Artemisia arbuscula/ Festuca idahoensis ARAR/ FEID Teton WY 10 Artemisia tridentata/ Festuca idahoensis ARTR/ FEID Meagher, Gallatin MT 10 Festuca scabrella/ Festuca idahoensis FESC/ FEID Glacier MT 10

DRY FORESTS Pseudotsuga menziesii/ Symphoricarpos albus PSME/ SYAL Meagher, Gallatin MT 10 Pseudotsuga menziesii/ Physocarpus malvaceus PSME/ PHMA Gallatin MT, Park WY 6

WARM MOIST FORESTS Populus tremuloides/ Calamagrostis rubescens POTR/ CARU Flathead MT 8 Tsuga heterophylla/ Clintonia uniflora TSHE/ CLUN Flathead MT 10 Abies lasiocarpa/ Clintonia uniflora ABLA/ CLUN Flathead MT 9 Abies lasiocarpa/ Xeophyllum tenax ABLA/ XETE Flathead MT 10

COOL CONIFER FORESTS Abies lasiocarpa/ Arnica cordifolia ABLA/ ARCO Teton WY 10 Abies lasiocarpa/ Vaccinium scoparius ABLA/ VASC Teton WY 10

HIGH GRASSLANDS AND TUNDRA Festuca idahoensis/ Agropyron caninum FEID/ AGCA Teton WY 10 Deschampsia caespitosa/ Carex spp. DECA/ CASP Park WY, Carbon MT 11 aEnvironmental types are those of Pfister et al. (1977) and Meuggler and Stewart (1980). bAbbreviations provide a key to Table 3. They represent dominant species by reporting initial letters (2) from genus and species names. cLocations are specified by county. Glacier and Flathead are in the Glacier National Park area. Broadwater and Meagher are adjacent to the Bridger/Big Belt Mountains. Gallatin, Park, and Carbon are at the north edge of Yellowstone. Teton includes Grand Teton National Park. dEach environmental type was sampled at 7–11 sites. At each site 5 environments were sampled with 5 parallel quadrats. Of these, those representing roadcuts and undisturbed vegetation are discussed here. value downward for sites at which the species on disturbed sites to enclose the range on un- did not occur (Table 3). (3) Constancy was cal- disturbed sites because competition is less rig- culated as the percentage of sites in the envi- orous on disturbed sites. ronmental type at which the species occurred. We hypothesized that a strong presence of RESULTS an exotic in roadside samples would result in a strong presence in adjacent undisturbed vege- Our observations of exotic plant presence tation because a strong presence at the road- on roadcuts (outslopes) and adjacent undis- side indicates both good adaptation to the turbed vegetation of 15 environmental types environment and production of many propa- are summarized in Table 3. (1) Vertically, table gules for colonization of nearby sites. We segments list groups of exotic species found, tested this hypothesis, using both constancy according to their ranges on the altitudinal and cover data, by comparing the presence of gradient studied: those with narrow, moderate, each exotic in disturbed vegetation on sites or broad amplitude and those with an inter- adjacent to occupied vs. unoccupied native rupted range. (2) The elevational gradient vegetation. The Mann-Whitney test, a non- ranges from dry steppe, through warm dry parametric t test, was used (Gibbons 1985). An forests, warm moist forests, cool forests, to alternative test, regression/correlation, was mountain meadows and alpine tundra. Fifteen forgone because quantitative data from the segments (environmental types or habitat types) undisturbed sites are currently unavailable. on this gradient are listed horizontally. These are In a companion paper (Despain et al. 2001), named and characterized in Table 1. (3) Entries we map the potential range of an exotic in a in Table 3 specify the presence of exotics, both region by using a map of the environmental on disturbed sites in corridors along which types (HTs) of the region (e.g., Despain 1990b) propagules are expected to move and on adja- as a base and shading ETs invasible by the cent undisturbed sites. Presence on roadsides is species studied. We expect the range mapped indicated by constancy (the percent of occupied 420 WESTERN NORTH AMERICAN NATURALIST [Volume 61

TABLE 2. Comparison of environments in major Rocky Mountain ecosystems. Standard errors, as well as additional data, are available in Weaver (1978, 1980, 1990, 1994).

b c ______Temperatures (C) ______Water availability (mm) d Jan Tgs July Pptn Drt HOH Soil GS Parameter min mean max ann mo deficit WHC mo Alpine tundraa –16 6 12 778 0 0 38 3.6 Abies lasiocarpa –18 12 22 820 0 0 30 4.5 Pseudotsuga menziesii –16 12 27 580 0 1 103 3.6 Festuca idahoensis –12 12 27 380 1 6 101 5.1 Agropyron spicatum –13 12 28 380 1 17 117 4.9 Bouteloua gracilis –15 14 31 350 2 25 117 4.4 aThe ecosystems compared range from alpine down through high forests (ABLA), low forests (PSME), and grasslands (FEID, AGSP, and BOGR). Each is named for its climax dominant vegetation and abbreviated with initial letters from its generic and specific epithets. bTemperatures (Weaver 1980, 1990) include average January minimum, growing season mean, and average July maximum. cWater data include annual precipitation, drought months, and annual water deficit (Weaver 1980, 1990, 1994), and water-holding capacity of the rooted zone (Weaver 1978), all in mm. dGrowing season months are defined as those with moist soils and average air temperature above 0°C (Weaver 1994). sites in the ET). Potential dominance on those wise this pattern would repeat in similar types, sites is indicated by cover (the average cover as it does for Taraxacum. on sites which are occupied); and current real- On disturbed sites the number of high-con- ized success is found by multiplying these stancy (>30%) exotics (Fig. 1, Table 3) was entries. Presence in undisturbed vegetation is 10–11 in grasslands, 9–12 in dry forests, and reported nonquantitatively from plots of the 8–10 in warm moist forests and 7–11 in cool same size and shape. The material in cells hav- forests. Numbers were lower in shrublands ing constancies >30% is in boldface because a (5–7), mountain meadows (5), and alpine (1). higher constancy indicates that the plant has Numbers of low-constancy exotics were 3–6 in established more or less regularly in that envi- grasslands, 5–11 in dry forests, 3–6 in warm ronmental type. moist forests, and 2–6 in cool forests. Low- The 29 exotic plants occuring in >10% of constancy richness was similar in mountain the sites in at least one environmental type are shrublands (2–5) and mountain meadows (7), listed (Table 3, vertically). Ten species have a and low in alpine (2). narrow amplitude; i.e., they have a high con- The number of exotics entering undis- stancy in only 1 or 2 types. Ten species have a turbed sites (Fig. 1, Table 3) decreased from moderate amplitude, that is, range over 4–8 grasslands (9–13) through aspen forests (8) and environmental types, as arranged in Table 3. shrublands (5–7) to conifer forests (0). It Four species have a broad amplitude, ranging increased again in mountain meadows (7) and over 10–14 ETs. The ranges of 5 species of alpine tundra (2). While the richness (average moderate to broad distribution are interrupted; number of species per sample) on undisturbed that is, they occupy low and higher sites, but sites is always lower than on disturbed sites, not the intervening environments. Two types most grassland ETs are occupied by at least of occurrence deserve further comment. First, one species not found on disturbed sites in it. plants with low constancy in a single ET are ignored because they may occupy microsites DISCUSSION in an environmental type; that is, they do not Exotics in the Northern actually occupy the environmental type dis- Rocky Mountains cussed. Alternatively, they could either be new to the region (Forcella and Harvey 1981) or be We found only 29 exotic species (Table 3) in the vanguard of a newly adapted ecotype. Sec- our sample of major upland environmental ond, 7 environmental types contain a species types of Glacier National Park, Grand Teton which occurs on undisturbed, but not on non- National Park, and little disturbed intervening competitive disturbed, sites. Such species could areas including parts of Yellowstone National possibly require a stability not found at road- Park. Our list does not include species that sides, e.g., lack of erosion or frost action. More have invaded since 1986, which occupy heav- likely, these species are “accidentals”; other- ily grazed areas or uncommon substrates. To 2001] EXOTIC PLANTS OF ROCKY MOUNTAIN ENVIRONMENTS 421 ...... 0A X...... 1C .. 1BX .. 2BX .. 4A 4B 4B 3E ...... 1A .. 1A X 1B 2B 3A 5B 3C 4D 3A 3A 8D ...... 1C indicate constancy and cover on disturbed sites tendency ...... 1A ...... 1A 1A 2A 1B 1A .. 2B ...... 5B 5E 8E 6B 7D 4D 8C 4B 8B 7D 9D 8D 6D 8E 8E 6C b ...... 1A .. 1A 1B 2B ...... 1A ...... 6DX 6C 4A 5B 8C 8CX 7BX 7C 5D 3B 3B 1A7C 1A 2B .. 1B2AX 1A .. .. 1A .. 5B 5A 5EX 1AX .. 1A 5BX 5A 3BX occurring in a stand the environment): 0 = 0–9, 1= 10–19, 2 20–29, . 9 90–100%. Cover classes are A .. .. 5DX 1AX 1A 5BX 4C 6C 2A 2A .. spen (8), warm moist forests (9–11), cool (12, 13), mountain meadows (14), and alpine (15). .. 2BX .. 1A ...... 1A ...... 0A ...... 1A ...... 6CX 3CX 4A 4BX 7EX 6BX 3B 5A 9CX 5D 5DX 8B 9DX 7CX 8DX 9EX 9D 9C 9C 8B 8C 7BX of the northern Rocky Mountains. Code digits .. a 3B 5BX 4AX 5BX 3B 8CX 8BX 8BX 9C 8C 5C 9E 8D 9CX 3BX ...... 1AX ...... 1A 1A ...... X ...... 1C ...... 2CX ...... X 9DX 9D 9CX7AX 4BX 5CX 1AX 2B .. X..1A...... 5E 2BX .. 2A .. 1A b 8AX4AX 9BX 8BX 5AX 5AX 6AX 5AX 8BX 8BX 9BX 7DX 7CX 4AX 8AX 5AX 4BX 3AX 1D 4EX 5EX 9EX 9EX 8DX 7CX 6CX 7BX 4B 6C 7B 5C 4C 7DX STCO AGSP ARAR ARTR FESC PSME PSME POTR TSHE ABLA ABLA ABLA ABLA FEID DECA BOGR BOGR FEID FEID FEID SYAL PHMA CARU CLUN CLUN XETE ARCO VACC AGCA CARX 3. Presence of major exotic species in major environmental types 3. Presence of major exotic ABLE T Trifolium repensTrifolium ...... Verbascum thapsusVerbascum Chrysanthemum leucanthemum .. .. 2AX 1AX ...... 2B ...... Festuca pratensisFestuca ...... 1A Dactylis glomerata .. .. 2B Alyssum alyssoides Camelina microcarpa Bromus japonicus Descurania pinnata acetosaRumex .. .. Agrostis alba hybridumTrifolium ...... 2A .. 1AX 1A .. 2C .. Poa compressaPoa pratenseTrifolium 1A .. 1A ...... Cirsium arvense procumbensTrifolium .. .. 1A ...... Madia glomerata .. Medicago lupulina ...... 1A .. .. Melilotus officinalis Bromus tectorum avicularePolygonum .. 1AX Agropyron cristatum dubius Tragopogon Bromus inermisLactuca serriola 2B Centaurea maculosa pratensis Poa officinaleTaraxacum X 2AX X Phleum pratense .. .. 1AX .. PECIES WITH NARROW AMPLITUDE PECIES WITH MODERATE AMPLITUDE PECIES WITH BROAD AMPLITUDE PECIES WITH INTERRUPTED RANGES Codes indicate constancy in roadside sites, cover occupied and invasiveness. Constancy (= the probability of Environmental types are listed from dry to moist, as in Table 1: grasslands (1, 2, 5), shrublands (3, 4), dry forests (6, 7), a Environmental types are listed from dry to moist, as in Table S S S +, B = 0–1%, C 1–2%, D 2–5%, E 5–25%, F >25%. Invasion of undisturbed areas in an environmental type is indicated by X. to invade undisturbed vegetation. Constancies >30% are in boldface emphasize environments where the species is common. S a b 422 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Fig. 1. Number of exotics (richness) by class and environmental type. Classes are total exotics of undisturbed sites (UnD-tot), common exotics of disturbed sites (D-maj), all exotics of disturbed sites (D-tot), and all exotics of disturbed and undisturbed sites (Total). Environmental types are as listed in Tables 1 and 3, that is, grasslands (1, 2, 5), shrublands (3, 4), dry forest (6, 7), aspen (8), warm moist forests (9, 10, 11), cool forests (12, 13), mountain meadows (14), and alpine (15). illustrate this qualifying statement, we give (= late seral) vegetation. And third, presence Chrysanthemum leucanthemum as a species in either disturbed or undisturbed sites demon- underrepresented because its range is expand- strates that the species has dispersed to the ing, Centaurea maculosa as less common than site. While proximity to a highway maximized expected because it expands with grazing the exotic’s likelihood of arrival at the sites we absent on the sites we studied, and Euphorbia studied, sites in environmental types in the esula as a species that is most important in agricultural zone had far more exposure to areas moister than we sampled, e.g., riparian invading propagules than did sites in the high sites. Since volcanic materials are uncommon mountains. in the region, extrapolation to volcanic parts of On disturbed sites the exotic richness Yellowstone National Park must be made with (species number) across the environmental caution. gradient seems remarkably constant (Fig. 1), despite changes in its composition (Table 3). Exotic Richness Across Regulars (constancy >30%) were 8–12 except Environmental Types in shrublands (5–7), one conifer forest (7), The presence of an exotic in an environ- mountain meadows (5), and tundra (1). While mental type depends on at least 3 factors. one might argue that incidentals (constancy First, the environment must be within the <30%) indicate microsite effects, numbers of physiologic niche of the species. A species range incidentals were also rather constant (2–6), on disturbed sites across the broad altitudinal except in the Pseudotsuga/Symphoricarpos envi- gradient suggests the breadth of the physio- ronment (11). Thus, disturbed sites seem to logic niche. Second, the environment/vegeta- have a more or less constant “richness capac- tion must be within the realized niche of the ity,” but filled with species differing among species. Presence in undisturbed vegetation of environments. Shrubland environments are an environmental type demonstrates presence slightly species deficient, for no obvious rea- in the realized niche, with respect to climax son. Pseudotsuga environments have a small 2001] EXOTIC PLANTS OF ROCKY MOUNTAIN ENVIRONMENTS 423 excess, possibly due to their location at eleva- that alpine opportunists are available in the tions supporting floras from Mediterranean/ Old World [Alps, Himalayas, Southeast Asia] steppe environments and northern coniferous flora and that the grazing disturbance has been environments. While we see no corresponding sufficient and environmental rigor slight enough break in environmental rigor (Table 2), the to induce the evolution of opportunistic species.) tundra environment is notably exotic poor. (2) In contrast, while mountain meadows seem In contrast, numbers of species invading almost equally isolated, our culture has pro- undisturbed vegetation of different environ- vided a stepping stone for exotics to them. The mental types vary greatly (Table 3): grasslands exotics have been introduced to environmen- (7–13), aspen (8), dry Pseudotsuga forests (8–10), tally similar foothill sites through commercial other conifer forests (0), and alpine (2). This sug- and agricultural activity, have established, and gests that the undisturbed vegetation of major are being transported upward, especially as environmental types differs greatly in compet- motorized backcountry use increases. For itiveness. In grasslands and dry forests, exotics example, while an experimentally bared por- occupying disturbed sites, i.e., tolerating the tion of a remote mountain meadow (Weaver physical environments, also colonize adjacent and Collins 1977) was not infected by Cirsium undisturbed vegetation. The open structure of arvense in the preceding 2 decades, thistle these vegetation types apparently provides non- appeared soon after loggers entered nearby competitive microsites for these exotics. The forests. exotic deficiency seen in the shrub zone was Distribution of also seen on disturbed sites and is most likely Individual Species induced by the physical environment. In contrast, exotics known to tolerate physical condi- Knowledge of the tolerance range of a tions in the conifer zone (i.e., disturbed sites) species tells us where to look for established rarely invade adjacent forest. These exotics are stands and where to expect establishment. Both probably excluded from forests by heavy com- are useful in planning control. It may also help petition for water/nutrients (Watt and Fraser us estimate a species’ ability to cross stress- 1933) or light. Thus, removal of forest commu- ful—dry or cold—zones without assistance. nities, by fire or harvest, should allow plants The importance of the latter is declining as capable of occupying noncompetitive disturbed human transport becomes the dominant dis- sites to colonize more widely in the forest persal mechanism. environment, where they may inhibit forest DISTRIBUTION AMONG DISTURBED SITES.— establishment but will finally yield when they The physiologic niche of a species is suggested are overtopped by tree species. by its presence in disturbed sites because The low exotic richness of the alpine is open spacing reduces competition. We recog- probably due in part to environments too rig- nize 4 distribution types (niche types). orous for establishment of opportunists (Billings First, species with narrow distributions are and Mooney 1968), but this does not explain most important in lower, warmer environments the sharp decline from the forest and mead- (Table 3). Some occupy dry grasslands (Agropy- ows below. It is likely that failure of dispersal ron cristatum, Alyssum alyssoides, Camelina also contributes. To illustrate, we contrast the microcarpa, and Bromus japonicus), shrublands exotic presence in alpine and mountain meadow (Rumex acetosa), and warm forests (Dactylis vegetation. (1) First, while plants adapted to glomerata, Festuca pratensis, Verbascum thap- disturbed Old World alpine environments may sus). None are important in the moist conifer exist, vectors—crops, animals, machinery— zone, cool conifer zone, or mountain grass- rarely pass directly from these areas to high- lands/tundra. In our data Chrysanthemum leu- altitude areas in the Rockies. Thus, the trans- canthemum seems to have narrow tolerances, fer of potential weeds has been slight. We con- but it is spreading rapidly into drier environ- servatively suggest that as recreational use mental types including those dominated by grows, managers should minimize introduc- Pseudotsuga menziesii and Festuca idahoensis. tions (exchanges) of exotics by increasing both Second, plants with broader tolerances pop- quarantine and efforts to detect and eradicate ulate wider zones in the altitudinal gradient unwanted establishment. (The presumed need (Table 3). Low-site plants (Tragopogon dubius, for this caution might be tested by showing Centaurea maculosa, and Melilotus officinalis) 424 WESTERN NORTH AMERICAN NATURALIST [Volume 61 may prefer grasslands over shrublands. Exotics to lack of precipitation and those of the high dominating near the lower forest margin plants mode occupy sites dry due to the high wind include Cirsium arvense, Poa compressa, and flows near mountain ridges (cf. Weaver 2001). the most drought tolerant (?) of the clovers (Tri- Each altitudinal zone contains species of folium procumbens). Trifolium pratense occurs both narrow and broad environmental ampli- throughout the low/warm conifer zone. Trifo- tudes. This is demonstrated by listing the lium repens, T. hybridum, and Agrostis alba species within an amplitude group according occur in the moist conifer zone, both low/warm to their locations on the altitudinal gradient and high/cool. (Table 3). Thus, among species with narrow Third, 2 plants (Poa pratensis and Tarax- distribution, Agropyron cristatum, important acum officinale) have remarkably wide distri- only in the driest environments, appears first. butions, extending from low grasslands through And among species with broad distribution, forests to mountain meadows and even tundra Poa pratensis appears before Taraxacum offici- (Table 3). Two others (Bromus inermis and nale because it becomes important at lower Phleum pratense) range from moister grass- altitudes. land environments through forest environ- DISTRIBUTION AMONG UNDISTURBED SITES.— ments to mountain meadow environments. All The tendency of exotics to escape from distri- of these species cover 2–5% (D in Table 3) or bution corridors is inversely related to the 5–25% (E) of the ground surface on disturbed penetrability of adjacent vegetation. Thus, while sites in some environments they occupy. establishment on disturbed sites provides an Fourth, 5–6 species representing 2 sub- indication of the physiologic niche, invasion of groups have interrupted or bimodal distribu- natural vegetation provides an indication of tions (Table 3). First, Lactuca serriola and Bro- the realized niche, i.e., performance under mus tectorum were found in dry grasslands competition from natural vegetation. (Bouteloua and Agropyron), were absent from The escape of species of all amplitudes and moister grassland environments, and reappeared gradient segments is proportional to the open- in dry forests (Pseudotsuga). One might specu- ness of the adjacent native vegetation. Species late that these species tolerate arid environ- of narrow to moderate altitudinal ranges often ments, cannot compete in moister grassland escape into relatively open grassland or Dou- environments, and become competitive again glas-fir (Pseudotsuga) vegetation, but they are where precipitation evaporates from treetops unlikely to escape into denser subalpine fir before it becomes available to plants in the forests (Abies; Table 3). Similarly, species with ground layer. This hypothesis would be more broad ranges tend to escape into grassland and convincing if the interruption occurred in the low forests but are unlikely to escape into dense undisturbed zone, but not in the disturbed forest environments (Table 3). Given these zone. The same interruption was reported for observations, we expect bimodal species to 2 native grasses (Stipa viridula and Koeleria escape in their lower, drought-stressed envi- nitida) and 5–10 exotic species (including Bro- ronments, but to be competitively constrained mus inermis, B. tectorum, B. japonicus, 3 annual in their upper, moister environments. This is mustards, and Kochia scoparia) that are pre- true except where the environment in the sent in the dry plains of eastern Montana, dis- upper arm is sufficiently wind-dried to create appear in the foothills and grasslands, and competitive conditions (and escape) similar to reappear in the Pseudotsuga zone to the west that in the low-elevation mode (Table 3). Poly- (Weaver and Meier 1997). Second, 3–4 species gonum aviculare and Madia glomerata are bi- have modes in both a lower-elevation zone modal plants illustrating the last point. and in the Abies/mountain meadow zone. Des- While undisturbed vegetation in the center curania (listed as unimodal) appears in dry of the forest zone may be impenetrable, seg- grasslands and has a weak high mode. Poly- ments of the forest zone that have been logged gonum and Madia appear first in moister grass- or burned are probably more penetrable, either lands and have solid high modes. Medicago because competition for light or water/nutri- appears first in the dry forest zone (Pseudo- ents (Watt and Fraser 1933) is reduced or be- tsuga) and reappears in the Abies/mountain cause wind dispersal is facilitated. Analysis of meadow zone. We speculate (hypothesize) that comparable samples (existing data) will even- plants of the lower mode occupy a site dry due tually test this hypothesis. 2001] EXOTIC PLANTS OF ROCKY MOUNTAIN ENVIRONMENTS 425

TABLE 4. Median constancy of both invading species and noninvading species on disturbed sites. The lower constancy of noninvading species may indicate poorer adaptation or a smaller seed supply. Environmental BOGR AGSP FEID FEID FESC PSME PSME POTR FEID DECA Over- type (HT) STCO BOGR ARAR ARTR FEID SYAL PHMA CARU AGCA CARX all Invaders 5.0 5.0 7.0 5.0 6.0 5.0 5.0 7.5 2.0 1.0 5.0 Noninvaders 1.5 1.0 1.5 2.5 2.5 1.0 5.0 2.5 2.0 0.0 1.0 P = 0.0005a 0.07 0.23 0.08 0.16 0.06 0.06 0.61 0.06 — 0.57 aKruskal-Wallis test (Gibbons 1985)

We expect the dominance of a species on river deposits or landslides) and more com- disturbed sites of an environmental type to mon in ridge sites of the alpine (e.g., unde- indicate its capacity to invade undisturbed composed rock). Although slopes of our road- sites in that environmental type, both because side sites may be steeper than the average dis- a species thriving on the disturbed site must turbed site, our data (Table 3, cover classes D be well adapted to the physical environment and E) probably identify the most problematic it occupies and because, as a well-adapted species of upland sites undergoing primary species, it will produce more seed. Our succession. On disturbed grassland sites (in- hypothesis is, then, that invading species will cluding mountain meadows) the only exotic be more dominant on adjacent disturbed sites with 5–25% cover (E) was Poa pratensis and than noninvaders. In fact, the median constancy exotics having 2–5% cover (D) were Bromus of invaders usually does exceed the median inermis, B. tectorum, Phleum pratense, and constancy of noninvaders, and the difference Rumex acetosa. In dry forests exotics with is significant in 70% of the cases (Table 4). cover 5–25% (E) were Dactylis glomerata and When data are pooled across all except the Phleum pratense, while those with cover 2–5% moist conifer types, which show no escape, (D) were Agrostis alba, Bromus inermis, and the difference is significant (P < 0.0005). The Trifolium repens. In moister conifer forests, moist conifer types, PSME/PHMA, TSHE, and those with cover 5–25% (E) were Bromus iner- ABLA forests are reasonably excluded from mis, Melilotus officinalis, Taraxacum officinale, this analysis because no exotic species have and Trifolium hybridum; and 2–5% (D) were moved from roadside to forested environments. Medicago lupulina, Phleum pratense, Trifolium Evaluating Exotics pratense, and T. procumbens. No exotic covered as much as 5% of either disturbed or undis- If public forest and park vegetation is to be managed for “pre-Columbian” condition (cf. turbed sites in the alpine. Ironically, the most U.S. Congress 1872), exotics should be ex- aggressive exotics are rarely discussed as cluded. If this is impossible, managers should problematic, and none of the exotics desig- strive to prevent exotics from dominating the nated as noxious seem to dominate in the wide vegetation because dominants are most likely range of environments we studied. to affect the success of native plant associates Secondary secession sites—such as recent and, through their influence on vegetation burns, logged areas, or old fields—are more composition, animal associates as well (cf. common on public lands than are primary suc- Clinton 1999). In evaluating species, we mini- cession sites. Here, the performance of exotics mize “breadth of distribution” as a criterion on may be similar to their performance on pri- the assumption that conservationists should mary succession sites. This expectation may equally emphasize preservation of all vegeta- overstate the problem since exotics, mostly tion types important in the region. Vegetation dispersing laterally through space, must com- types rare in the region deserve special atten- pete with natives colonizing both from the tion if they are endemic to it, but they are less propagule bank and dispersing laterally. Thus, critical if they are well represented in other we expect the grasses (Agrostis, Bromus, Dac- regions. Because our project was designed for tylis, Phleum, and Poa), legumes (Melilotus, generality, we studied no rare types. Medicago, and Trifolium), and dandelion, listed Sites undergoing primary succession are above, to be among the most important exotic rare in the forest and grassland zones (e.g., invaders. 426 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Late seral sites may be as common as or sites they occupied in at least one type (Table more common than secondary succession sites 3); they include grasses (Agrostis, Bromus, Dac- in national forests and parks. Later seral vege- tylis, Phleum, and Poa), legumes (Melilotus, tation of moister forests is impenetrable, but Medicago, and Trifolium), Rumex and Tarax- grasslands, shrublands, and dry forests are in- acum species. An additional seven species ex- vaded by many species (Table 3). However, hibited cover of 2–5% on disturbed sites in at because dominance of all species falls from least one environmental type (HT). Most of disturbed to undisturbed sites, we discount these plants were introduced intentionally and most of the species not listed as invaders of none of these stealth plants is normally con- secondary succession sites. This position may sidered a noxious weed. understate the effects of robust (e.g., Agropy- ron cristatum or Melilotus officinalis) or very ACKNOWLEDGMENTS numerous (e.g., Bromus tectorum, B. japonicus, or Alyssum alyssoides) plants of the driest We thank the University of Wyoming/NPS environmental types. (Grant #3-10144) and Montana Department of Transport (FHWA/MT 97-8115) for mone- CONCLUSIONS tary support. B. Wood sampled sites in Glacier and Grand Teton national parks, J. Lichtardt The number of exotics currently common sampled stands in the intervening area, D. in vegetation of the northern Rocky Moun- Gustafson did most of the mathematical analy- tains is relatively few, approximately 29 (Table sis, and T. Weaver did the writing. We thank 3). The altitudinal (temperature/moisture) ampli- many colleagues for critical discussion and D. tude of each of these species is described by Despain, K. Harper, and R. Sheley for review presence in environmental types (HTs) repre- of the manuscript. senting segments of the environmental gradient (Table 3). Knowledge of species ampli- LITERATURE CITED tudes will enable managers to estimate and CLINTON, W. 1994. Environmentally and economically even map potential distributions of exotics, beneficial practices on federally landscaped grounds. both in disturbed (primary succession) and Memo to heads of executive departments and agen- undisturbed (late seral) vegetation. cies. The White House, Office of Press Secretary, The overall invasibility of major environ- Washington, DC. 2 pp. ______. 1999. Invasive species. The White House. www. mental types—in both disturbed and undis- pub.whitehouse.gov. turbed conditions—is indexed by tabulating BILLINGS, W., AND H. MOONEY. 1968. The ecology of arc- exotic species richness across a broad altitudi- tic and alpine plants. Biological Review 43:485–530. nal gradient of types. Grasslands and dry forest DAUBENMIRE, R. AND J. 1968a. Forest vegetation of eastern Washington and northern Idaho. Washington environments harbor the most exotic species, Agricultural Experiment Station Technical Bulletin both in disturbed and undisturbed sites. Moist 60, Pullman. 104 pp. conifer forests have similar species richness on DAUBENMIRE, R. 1968b. Plant communities. Harper and disturbed sites, but no exotics appear on undis- Row, New York. 300 pp. ______. 1970. Steppe vegetation of Washington. Washing- turbed sites. Tundra environments support few ton Agricultural Experiment Station Technical Bul- exotics on either disturbed or undisturbed letin 62, Pullman. 131 pp. sites. ______. 1976. The use of vegetation in assessing the pro- Dominance in vegetation in at least one ductivity of forest lands. Botanical Review 42: environmental type is our criterion for recog- 115–143. DESPAIN, D. 1990a. Yellowstone vegetation. Roberts Rine- nizing an exotic of special concern, because a hardt, Boulder, CO. 239 pp. dominant is most likely to affect the success of ______. 1990b. Habitat types of Yellowstone National Park plant associates and, through its influence on [electronic Arc-Info grid file]. Yellowstone National Park, WY. Available from GIS Specialist, Spatial vegetation composition, the success of animal Analysis Center, Yellowstone National Park, PO Box associates as well. We minimize breadth of 168, Mammoth Hot Springs, WY 82190. distribution as a criterion on the assumption DESPAIN, D., T. WEAVER, AND R. ASPINALL. 2001. Rule- that conservationists should emphasize equally based mapping of potential weed distribution. West- the preservation of all regionally common ern North America Naturalist 61:428–433. FORCELLA, F., AND S. HARVEY. 1981. New and exotic weeds and internationally unique ecosystems. Seven of Montana. Herbarium, Montana State University, species exhibited cover of 5–25% on disturbed Bozeman. 117 pp. 2001] EXOTIC PLANTS OF ROCKY MOUNTAIN ENVIRONMENTS 427

GIBBONS, J. 1985. Non-parametric methods for quantita- ______. 1994. Vegetation distribution and production in tive analysis. American Science Press, Columbus, Rocky Mountain climates—with emphasis on white- OH. 481 pp. bark pine. Pages 142–152 in W. Schmidt and K. Holt- GRIME, J. 1979. Plant strategies and vegetation processes. meier, editors, Stone pines and their environments. J. Wiley, New York. 222 pp. USDA Forest Service General Technical Report 309. HOLDRIDGE, L. 1947. Determination of world plant for- 321 pp. mations from simple climatic data. Science 105: ______. 2001. Whitebark pine (Pinus albicaulis) and its 367–368. environment. Pages 41–73 in D. Tomback, S. Arno, MUEGGLER, W., AND W. S TEWART. 1980 Grassland and and R. Keane, editors, Whitebark pine communities: shrubland habitat types of western Montana. USDA ecology and restoration. Island Press, Washington, Forest Service General Technical Report INT-66. DC. 440 pp. 154 pp. WEAVER T., AND D. COLLINS. 1977. Possible effects of PFISTER R., B. KOVALCHIK, S. ARNO, AND R. PRESBY. 1977. weather modification (increased snowpack) on Fes- Forest habitat types of Montana. USDA Forest Ser- tuca idahoensis meadows. Journal of Range Manage- vice General Technical Report INT-34. 141 pp. ment 30:451–456. STEELE R., S. COOPER, D. ONDOV, D. ROBERTS, AND R. WEAVER, T., AND G. MEIER. 1997 Desirables and weeds PFISTER. 1983. Forest habitat types of eastern Idaho for roadside management—a northern Rocky Moun- and western Wyoming. USDA Forest Service Gen- tain catalogue. Montana Department of Transport eral Technical Report INT-144, Ogden, UT. 144 pp. Report #FHWA/MT-97/8115. 144 pp. U.S. CONGRESS. 1872. An act establishing Yellowstone WEAVER, T., D. GUSTAFSON, AND J. LICHTHARDT. 1993. National Park. 17 Statute 32. Plants colonizing disturbed areas in fifteen Rocky WALTER, H. 1960. Grundlagen der Pflanzen verbreitung. Mountain environments: weeds and reclamation Eugen Ulmer, Stuttgart, Germany. 566 pp. candidates. University of Wyoming/ NPS Research WATT, A., AND G. FRASER 1933. Tree roots in the field layer. Center Reports 17: 0–25. Journal of Ecology 21:404–414. WHIPPLE, J. 2000. Yellowstone’s changing flora: the esca- WEAVER, T. 1978. Changes in soils along a vegetational lation of exotics. Western North American Naturalist (altitudinal) gradient of the northern Rocky Moun- 61:336–346. tains. Pages 14–29 in C. Youngberg, editor, Forest WHITSON, T., EDITOR. 1992. Weeds of the West. Western soils and land use. Soil Science Society of America, Society of Weed Science. Newark, CA. 630 pp. Madison, WI. 623 pp. WHITTAKER, R. 1975. Communities and ecosystems. ______. 1980. Climates of vegetation types of the north- McMillan, New York. 335 pp. ern Rocky Mountains and adjacent plains. American Midland Naturalist 103:392–398. Received 14 March 2000 ______. 1990. Climates of subalpine pine woodlands. Accepted 13 June 2001 Pages 72–79 in W. Schmidt and K. McDonald, com- pilers, Whitebark pine ecosystems. USDA Forest Service General Technical Report INT-270. 386 pp. Western North American Naturalist 61(4), © 2001, pp. 428–433

A RULE-BASED MODEL FOR MAPPING POTENTIAL EXOTIC PLANT DISTRIBUTION1

Don G. Despain2, T. Weaver3, and Richard J. Aspinall4

ABSTRACT.—Wildland managers need a method to predict which portions of the lands under their stewardship are susceptible to invasion by exotic plants. We combined a database listing exotic plant species known to occur in major environmental types (habitat types) throughout the northern Rocky Mountains with a digital vegetation map of environmental types for a major national park in the region (Yellowstone National Park) to produce maps of areas potentially threatened by major exotic species. Such maps should be helpful to managers concerned with monitoring and controlling exotic plants.

Key words: maps, exotics, weeds, GIS, Yellowstone National Park, modeling, Centaurea, Cirsium, Melilotus, Phleum.

More than 100 exotic plant species occur in We use environmental type as a synonym for Yellowstone National Park (Whipple 2001), and Daubenmire’s habitat type, but prefer envi- others will undoubtedly become established in ronmental type because it unambiguously the future. Many of these are likely to under- refers to physical environment and excludes go range expansion. An ability to predict the confusing factors in animal “habitat” such as areas threatened by expanding exotics should characteristics of a community temporarily be of great value to park managers trying to occupying the site (e.g., species composition minimize dispersal to susceptible areas and or structure of a seral community). Dauben- eradicate new colonies of these areas. mire recognized and regretted this confusion Information needed to predict the potential (Weaver et al. 2001). extent of a species includes knowledge of Second, exotic species’ potentials to invade which environments are susceptible to inva- environmental types representing segments of sion by the species and the location and extent the altitudinal gradient of the northern Rocky of susceptible environments. Mountains have been identified by Weaver et Both are available for Yellowstone National al. (2001). In their treatment the environmen- Park. First, we have a map of environmental tal range of a species is expected to be wider types (Despain 1990a). Students of vegetation in disturbed sites (where competition is less) have pointed out that plant communities provide than in late seral communities (where compe- a good indicator for site conditions (Holdridge tition is intense; Daubenmire 1968, Grime 1947, Whittaker 1975, Huschle and Hironaka 1979, Huschle and Hironaka 1980), and this 1980). In our area Daubenmire identified major has been demonstrated (Weaver et al. 2001). environmental types (habitat types) for eastern Thus, we expect geographic ranges of exotic Washington and northern Idaho and demon- species in undisturbed vegetation to be nar- strated the relationship of indicator species rower than, and nested in, ranges of the same to both environmental qualities (Daubenmire species occupying disturbed vegetation. 1952, 1956) and plant performance (Dauben- This paper has 5 objectives: (1) to demon- mire 1976). His environmental types have strate a method for mapping potential plant been extended into southern Idaho, Montana, distribution, (2) to illustrate it with 4 exotic and Wyoming (Pfister et al. 1977, Mueggler et plant species of Yellowstone National Park, (3) al. 1980, Hironaka et al. 1983, Steele et al. to publicize maps of 24 other exotics, (4) to 1983), and their relationships to environment compare the mapped ranges of each species have been reviewed by Weaver et al. (2001). on undisturbed and disturbed sites, and (5) to

1Presented at the 5th Biennial Scientific Conference on the Greater Yellowstone Ecosystem, October 1999. See footnote 1 on p. 409. 2Northern Rocky Mountain Science Center (USGS), PO Box 173492, Bozeman, MT 59717-3492. 3Department of Biology, Montana State University, Bozeman, MT 59717-3460. 4Geographic Information and Analysis Center, Montana State University, Bozeman, MT 59717-3495.

428 2001] MAPPING POTENTIAL WEED DISTRIBUTION 429 evaluate the method by comparing predicted TABLE 1. Environmental types of Yellowstone National distributions with actual distributions recorded Park with Weaver et al. (2001) equivalents. by Yellowstone National Park’s weed manage- Weaver et al.a Yellowstone environmental typeb ment staff. DECA/CARX alpine tundra

METHODS FEID/AGCA FEID/DECA POFR-ARCA/DECA ARCA/FEID Working in Glacier National Park, Yellow- ARTR/FEID-GEVI stone National Park, Grand Teton, and areas FEID/STRI between, Weaver et al. (2001) studied the dis- FEID/AGCA-GEVI tribution of exotic plants in 16 environmental FEID/AGCA types representing altitudinal zones of the ARTR/FEID ARTR/FEID northern Rocky Mountains. To determine ABLA/ARCO PIAL/VASC which exotics invade disturbed and undis- PIAL/CAGE turbed vegetation in these environmental ABLA/VASC ABLA/LIBO-VASC types, they recorded presence at 7–10 sites in ABLA/VAGL-VAGL each environmental type. To test susceptibility, ABLA/THOC they examined sites long exposed to diverse ABLA/VASC-VASC seed sources, i.e., sites near major highways. ABLA/VASC-PIAL ABLA/VASC-CARU Their inspection of each site was concentrated ABLA/CAGE in two 4 × 25-m plots running parallel to the ABLA/CARU road. Entry into disturbed sites was examined ABLA/ARCO ABLA/CARO with a plot on the roadcut (inslope) bordering PICO/CAGE the highway. Entry into near-climax vegeta- PICO/CARO tion was examined with another plot in adja- PICO/PUTR cent undisturbed vegetation. They listed all PSME/PHMA PSME/PHMA species present in the plot (and in similar × PSME/SYAL PSME/SYAL areas around it), recorded presence in five 4 PSME/CARU 5-m segments of the plot as an index of ubi- PSME/SPBE-SPBE quity, and estimated cover with 75 points AGSP/BOGR FEID/AGSP located near the central axis of the plot. They ARTR/AGSP reported both constancy values (the percentage of plots in an environmental type where STCO/AGSP AGSP/POSA-STCO aEnvironmental types are named for 2 species, including a dominant overstory the species occurred) and cover for each species and an indicator species. Names of these species are abbreviated with species. For a more complete description of a 4-letter code including 2 letters from the genus name and 2 from the specific epithet: ABLA = Abies lasiocarpa, AGCA = Agropyron caninum, AGSP their methods, refer to Weaver et al. (2001). = Agropyron spicatum, ARCA = Artemisia cana, ARCO = Arnica cordifolia, We used Weaver et al.’s (2001) constancy ARTR = Artemisia tridentata, BOGR = Bouteloua gracilis, CAGE = Carex geyeri, CARO = Carex rossii, CARU = Calamagrostis rubescens, CARX = value as a measure of a species’ ability to Carex spp., DECA = Deschampsia caespitosa, FEID = Festuca idahoensis, GEVI = Geranium viscosissimum, LIBO = Linnaea borealis, PHMA = establish in an environmental type. Our maps Physocarpus malvaceus, PIAL = Pinus albicaulis, PICO = Pinus contorta, indicate areas where a species was present at POFR = Potentilla fruticosa, POSA = Poa sandbergii, PSME = Pseudotsuga menziesii, PUTR = Purshia tridentata, SPBE = Spirea betulifolia, STCO = more than half the sites, at less than half the Stipa comata, STRI = Stipa richardsonii, SYAL = Symphoricarpos albus, sites, and where they were capable of invad- THOC = Thalictrum occidentale, VAGL = Vaccinium globulare, VASC = Vac- cinium scoparium. ing the climax community. bWeaver et al. (2001) did not encounter all environmental types that occur in Yellowstone National Park. Thus, Yellowstone types (Despain 1998) were Two details require elaboration. First, be- grouped with the Weaver type to which they were most similar. Blocking in cause Weaver et al. (2001) did not encounter this table indicates the correspondences. Yellowstone types for which there are no equivalent Weaver types include hot springs vegetation, sedge bogs, all environmental types that occur in Yellow- willow/sedge, wet forests, talus, and water. stone National Park, we predicted exotic plant species occurrence, in those Yellowstone National Park types for which they had no the known type was assigned to other types in data, from the most similar type for which data its group. were available. A type was judged to be simi- Second, Despain’s (1990a) habitat type map lar if it was in a similar moisture range of the sometimes uses mosaic mapping units that same series. Resultant assignments are shown contain 2 dominant types, such as a matrix of in Table 1. Exotic plant species presence in grasslands with numerous islands of trees or 430 WESTERN NORTH AMERICAN NATURALIST [Volume 61 vice versa. In these cases we averaged the Centaurea maculosa constancy values of the 2 component types to derive a value for the mosaic units. If a species could invade the climax vegetation of either of the types, the entire map unit was considered to be susceptible to that species. The resultant database was combined with the vegetation map using GIS to create 28 maps, one for each species studied. Four species are used as illustrations. Canadian thistle (Cirsium arvense [L.] Scop.) and spotted knapweed (Centaurea maculosa Lam.) are classed as noxious weeds by the surrounding states. Yellow sweetclover (Melilotus officinalis [L.] Lam.) and timothy (Phleum pratense L.) are crop plants that have become widely established in nonagricultural areas of the region. All 4 are of special concern to Yellowstone National Park managers. To evaluate the success of our model, we compared locations we mapped for 3 species with actual locations mapped by Yellowstone National Park’s staff: Canadian thistle, spotted knapweed, and yellow sweetclover (data for Fig. 1. Potential distribution of Centaurea maculosa timothy were not available). (spotted knapweed) in Yellowstone National Park. Gray areas show the distribution of disturbed areas where it is RESULTS expected to occur in less than half a series of study plots. No areas occurred where it would be capable of occurring Potential ranges of 28 exotic species found in more than half the study plots. Black areas show distribution of those sites where it is capable of invading climax repeatedly in northern Rocky Mountain vege- vegetation. Roads are indicated (solid line) for reference. tation (Weaver et al. 2001) were mapped. Actual locations recorded by Yellowstone’s weed manage- Maps are available from the Geographic Infor- ment staff are shown by triangles. mation and Analysis Center, Montana State University, Bozeman, website (http://www.giac. montana.edu) in raster format at 50-m resolu- or an indication that knapweed enters environ- tion, which should be useful for field purposes. ments not predicted by this model and is thus Centaurea maculosa is classified as a nox- a serious threat over a much larger area than ious weed in the Greater Yellowstone Ecosys- that mapped. While it does not appear to pose tem. The potential range of spotted knapweed a serious threat to the majority of the park, mapped for disturbed sites (Fig. 1) includes it should be closely monitored as a potential the drier portions of the park, i.e., dry grass- threat especially in the Yellowstone River val- lands/shrublands and drier Douglas-fir forests. ley along the north boundary. We mapped no areas where knapweed would Cirsium arvense is a 2nd noxious weed of have an expected constancy >50%. It is ex- the Greater Yellowstone Ecosystem. The poten- pected to invade climax vegetation only in dry tial range mapped for it (Fig. 2) includes dis- grasslands predominantly at low elevations. In turbed areas primarily in sparsely vegetated contrast to our predictions, actual Yellowstone forest types and montane and subalpine grass- National Park data showed many locations lands/shrublands. We map no potential for entry along park roads outside our predicted areas. into dry grasslands/shrublands. Because it does Thus, more data are required to determine not invade climax vegetation, colonies estab- how threatening this species is in Yellowstone lished on disturbed sites are expected to die National Park. Unpredicted locations may be out as succession progresses to climax. No areas either transient occurrences that would disap- occurred where Canadian thistle would have pear without constant seeding from the outside an expected constancy >50%. Our map is 2001] MAPPING POTENTIAL WEED DISTRIBUTION 431

Cirsium arvense Melilotus officinalis

Fig. 2. Potential distribution of Cirsium arvense (Cana- Fig. 3. Potential distribution of Melilotus officinalis (yel- dian thistle) in Yellowstone National Park. Gray areas low sweetclover) in Yellowstone National Park. Gray areas show the distribution of disturbed areas where it is show the distribution of disturbed areas where it is expected to occur in less than half a series of study plots. expected to occur in less than half a series of study plots. No areas occurred where it would be capable of occurring No areas occurred where it would be capable of occurring in more than half the study plots. Black areas show distri- in more than half the study plots. Black areas show distribution of those sites where it is capable of invading climax bution of those sites where it is capable of invading climax vegetation. Roads are indicated (solid line) for reference. vegetation. Roads are indicated (solid line) for reference. Actual locations recorded by Yellowstone’s weed manage- Actual locations recorded by Yellowstone’s weed management staff are shown by triangles. ment staff are shown by triangles. validated by noting that the majority of loca- staff did not correspond to predicted locations. tions mapped by the Yellowstone National Park More data must be gathered to determine the weed management staff are within the areas threat posed by yellow sweetclover. This species we mapped as potential habitat. Anomalous could become a serious problem if it displaces colonies in dry grassland/shrubland units may native climax species in sites to which it is be located within inclusions of wetter environ- well adapted. mental types. While Phleum pratense is less obvious than Melilotus officinalis is a plant of special the forbs just discussed, it has a significant concern because it tends to dominate grass- tendency to dominate Yellowstone National lands. The potential range mapped for yellow Park vegetation (Weaver et al. 2001). The poten- sweetclover on disturbed sites (Fig. 3) includes tial range mapped for timothy (Fig. 4) includes areas from drier grassland/shrubland sites in the disturbed areas in most of the park. The map northern part of the park to moist subalpine indicates that it can invade climax communi- meadows. Our map predicts that Melilotus is ties in a smaller range of environmental types, capable of invading open climax communities i.e., moister grasslands/shrublands and lower across the same range. In the higher-elevation forest communities. It is more common than forest zone it can invade disturbed areas. No spotted knapweed, Canadian thistle, and yellow areas were mapped where yellow sweetclover sweetclover on disturbed sites; i.e., it had would have an expected constancy >50%. Most a constancy >50% over large portions of locations recorded by the weed management the park. Because weed management staff 432 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Phleum pratense are usually highest in or near the potential range predicted. We attribute near-misses to interfingering of environmental types in eco- tonal areas, unmapped islands of one type in a matrix of another, or inexact records of invader colony locations made by the weed team. The certainty of our maps could be increased by adding more observations, particularly in those types where Weaver et al. (2001) have no data. The most extensive of these in Yellow- stone are the wetland types and high-elevation forests. For simplicity we have mapped ranges in successional extremes of severely disturbed roadside cuts and near-climax conditions. Road- cuts are typically void of developed soil and are usually in the early stages of primary succession. Gathering more data relating to more moderate disturbances, such as wildland fire, could usefully extend the work. For example, while Canadian thistle has been shown to increase after forest fire (Turner et al. 1997), this is not reflected (Fig. 2). The persistence of this species as the community succeeds, after Fig. 4. Potential distribution of Phleum pratense (timo- fire, to climax vegetation deserves study. Thus, thy) in Yellowstone National Park. Light gray areas indicate where it is expected to occur in less than half a series it would be useful to gather exotic species dis- of study plots. Dark gray indicates areas where it is tribution data across successional stages within expected to occur in more than half the study plots. Black each of the environmental types (Despain areas show distribution of those sites where it is capable 1990a) to allow a broader and more accurate of invading climax vegetation. Roads are indicated (solid line) for reference. evaluation of the threat posed by a particular species. We recommend that managers gather the members do not map it, actual location data data necessary to use this method to further for this widespread species are not available their efforts in monitoring and controlling the for validation of our maps. While it is of little establishment and spread of these exotic plants, concern in the forested types because it would especially those that are most likely to cause be greatly reduced at canopy closure, it could extensive ecological and economic problems. be of major concern in moister grassland/shrubland environments, which are the major source ACKNOWLEDGMENTS of forage for native ungulates. National Park Service/University of Wyom- DISCUSSION ing Research Center and U.S./Montana Depart- ment of Transport supported measurement of This exercise has demonstrated a method exotic presence in major environmental types. for producing maps showing the potential Yellowstone National Park supported mapping ranges of exotic plant species in disturbed and of the environmental types of the park and undisturbed environments. Some general pat- provided information about the distribution of terns are seen in the maps: (1) potential ranges exotic plants in the park. of some species are limited while others are extensive; (2) most invader species are adapted LITERATURE CITED to colonize disturbed sites, and thus species DAUBENMIRE, R. 1952. Forest vegetation of northern ranges are broader on disturbed than undis- Idaho and adjacent Washington, and its bearing on turbed segments of an environmental type; (3) concepts of vegetation classification. Ecological Mono- where colony locations are known, constancies graphs 22:301–330. 2001] MAPPING POTENTIAL WEED DISTRIBUTION 433

______. 1956. Climate as a determinate of vegetation dis- Forest Service, General Technical Report INT-66, tribution in eastern Washington and northern Idaho. Ogden, UT. Ecological Monographs 26:131–154. PFISTER, R.D., B.L. KOVALCHIK, S.F. ARNO, AND R.C. ______. 1976. The use of vegetation in assessing the pro- PRESBY. 1977. Forest habitat types of Montana. USDA ductivity of forest lands. Botanical Review 42: Forest Service, General Technical Report INT-34, 115–143. Ogden, UT. DESPAIN, D.G. 1990a. Habitat types of Yellowstone Nation- STEELE, R., S.V. COOPER, D.M. ONDOV, D.W. ROBERTS, al Park [electronic Arc-Info grid file]. Yellowstone AND R.D. PFISTER. 1983. Forest habitat types of east- National Park, WY. Available from: GIS Specialist, ern Idaho–western Wyoming. USDA Forest Service, Spatial Analysis Center, Yellowstone National Park, General Technical Report INT-144, Odgen, UT. PO Box 168, Mammoth Hot Springs, WY 82190. TURNER, M.G., W.H. ROMME, R.H. GARDNER, AND W. W. http://www.nps.gov/yell/technical/gis/download.htm. HARGROVE. 1997. Effects of fire size and pattern on ______. 1990b. Yellowstone vegetation: consequences of early succession in Yellowstone National Park. Eco- environment and history in a natural setting. Roberts logical Monographs 67:411–433. Rinehart, Inc, Boulder, CO. 239 pp. WEAVER, T. 2001. Whitebark pine (Pinus albicaulis) and GRIME, J. 1979. Plant strategies and vegetation processes. its environment. Pages 41–73 in D. Tombeck, S. Arno, John Wiley, New York. 222 pp. and R. Keane, editors, Whitebark pine communities: HIRONAKA, M., M.A. FOSBERG, AND A.H. WINWARD. 1983. ecology and restoration. Island Press, Washington, Sagebrush-grass habitat types of southern Idaho. DC. 440 pp. Forest, Wildlife and Range Experiment Station Bul- WEAVER, T., D. GUSTAFSON, AND J. LICHTHARDT. 2001. letin 35, University of Idaho, Moscow. 44 pp. Exotic plants in early and late seral vegetation of fif- HOLDRIDGE, L.R. 1947. Determination of world plant for- teen northern Rocky Mountain environments (HTs). mations from simple climatic data. Science 105: Western North America Naturalist 61:417–427. 367–368. WHIPPLE, J. 2001. Annotated checklist of exotic vascular HUSCHLE, G., AND M. HIRONAKA. 1980. Classification and plants in Yellowstone National Park. Western North ordination of seral plant communities. Journal of American Naturalist 61:336–346. Range Management 33:179–182. WHITTAKER, R.H. 1975. Communities and ecosystems. KUCHLER, A.W. 1967. Vegetation mapping. Ronald Press Macmillan, New York. 385 pp. Company, New York. 470 pp. MUEGGLER, W.F., AND W.L. STEWART. 1980. Grassland and Received 22 February 2000 shrubland habitat types of western Montana. USDA Accepted 15 June 2001 Western North American Naturalist 61(4), © 2001, pp. 434–444

COMPARATIVE TOLERANCE OF FOUR STOCKS OF CUTTHROAT TROUT TO EXTREMES IN TEMPERATURE, SALINITY, AND HYPOXIA

Eric J. Wagner1,2, Ronney E. Arndt1, and Mark Brough1

ABSTRACT.—Four stocks of cutthroat trout (Oncorhynchus clarki) were exposed to high temperature, high salinity, and low dissolved oxygen to determine inherent differences. The fish tested included 2 stocks of Bonneville cutthroat trout (O. c. utah), a lacustrine stock derived from Bear Lake and a fluvial-origin stock from southern Utah (Manning Meadow Reservoir). The other 2 stocks tested were from Electric Lake (largely Yellowstone cutthroat trout, O. c. bouvieri) and Jackson Hole, Wyoming (fine-spotted Snake River cutthroat trout, O. c. subsp.). Temperature tests were either critical thermal maximum (CTM) or 96-hour trials using juveniles acclimated between 12.5°C and 18.0°C. Two CTM end points were temperature at first loss of equilibrium (CTMeq) and onset of spasms (CTMs). There were no significant differences ° ° in CTMeq among test fish acclimated to 18.0 C, but CTMs was significantly higher for Bear Lake Bonneville (30.0 C) than for Snake River (29.6°C) or southern Bonneville (29.7°C) stocks. With fish acclimated at 13.0°C, there were no significant differences among the stocks in CTMeq or CTMs. Differences among stocks varied significantly among nine 96- hour tests. Overall, it appeared that the southern Bonneville stock had slightly better survival at warmer temperatures than other stocks. In 24-hour survival tests at high salinities, the Snake River stock had the lowest tolerance, with significant mortality occuring at 18‰ (29.5 mS ⋅ cm–1 conductivity). The southern Bonneville stock had the highest tolerance, with no mortality until 22‰ (38 mS ⋅ cm–1). Bear Lake Bonneville and Electric Lake stocks had 60% and 30% mortality, respectively, at 21‰ (36 mS ⋅ cm–1). Hypoxia tolerance measured by resistance time, 24-hour mortality, or probit analysis ⋅ –1 (LEC50) did not differ among stocks. The 24-hour LEC50 was 2.34 mg O2 L for all stocks combined.

Key words: temperature, oxygen, conductivity, critical thermal maximum, Bonneville cutthroat trout, Yellowstone cutthroat trout, Snake River cutthroat trout.

Cutthroat trout (Oncorhynchus clarki) have throat trout have undergone some degree of been divided into 4 major and 10 minor sub- natural selection, creating adaptations to par- species by Behnke (1988), based on phyloge- ticular environments. Genetic differences, e.g., netic divergence. Behnke (1992) has described as measured by mitochondrial DNA or enzyme phenotypic differences among these subspecies proteins, among subspecies of cutthroat trout as well as their geographic distribution, status, have been noted and used to differentiate be- life history, and ecology. tween them (Loudenslager and Gall 1980, Differences among subspecies are of impor- Leary et al. 1987, Shiozawa and Evans 1994). tance to fisheries managers and anglers. For Several stocks of cutthroat trout are avail- example, Lahontan cutthroat trout (O. clarki able from certified pathogen-free sources for henshawi) may reach a maximum size twice fisheries management in Utah. These include that of other subspecies (Behnke 1992). In the Bear Lake Bonneville (BL), Bear Lake, Utah- rainbow trout (O. mykiss) group, redband trout Idaho; southern Bonneville (BV) from Man- (O. m. subsp.) display an ability to adapt to high ning Meadow Reservoir, Monroe Mountains, temperatures, feeding at temperatures lethal Utah; and Electric Lake (EL), Emery County, to other subspecies (Behnke 1992). Researchers Utah, stocks. The EL stock is primarily Yel- have also noted differences in feeding habits, lowstone cutthroat trout (O. c. bouvieri), with susceptibility to various methods of angling, potential introgression by rainbow trout and return to the creel, and condition among cut- Bear Lake Bonneville cutthroat trout while in throat trout stocks (Trojnar and Behnke 1974, Strawberry Reservoir before transfer to Elec- Nielson and Lentsch 1988, Dwyer 1990, Hep- tric Lake (Martin et al. 1985). The BL and BV worth et al. 1999). are recognized as O. clarki utah (Behnke 1992). These genetically based differences among The history of these broodstocks was reviewed subspecies and stocks are indicators that cut- by Wagner (1996) and Hepworth et al. (1997).

1Fisheries Experiment Station, Utah Division of Wildlife Resources, 1465 West 200 North, Logan, UT 84321. 2Corresponding author.

434 2001] COMPARATIVE TOLERANCE OF CUTTHROAT TROUT 435

The Snake River stock (O. c. subsp.; SN; increased at a constant rate of 0.2°C ⋅ min–1 Behnke 1992), originating from wild stock in on a laboratory hot plate, similar to the rate Wyoming via Jackson National Fish Hatchery, recommended by Becker and Genoway (1979). was also included in this study for comparison. Temperature was recorded at 10-minute inter- Water quality is a critical aspect of the fish’s vals from a digital probe suspended off the environment, but how these subspecies differ bottom, and change in temperature was calcu- in water quality tolerances has not been evalu- lated for each interval. Time and temperature ated. Water quality in Utah is as varied as the were recorded at the first loss of equilibrium landscape, ranging from nearly distilled water and at the onset of spasms, whereupon the test in granite watersheds of the Uinta Mountains was concluded. DO was recorded at the end to over 150,000 mg ⋅ L–1 salinity in Great Salt of the test. We conducted these tests between Lake (Gliwicz et al. 1995). Some reservoirs 25 September and 31 October 1996 using fish experience low dissolved oxygen during win- of 0.7–2.9 g average weight. ter ice cover and summer stratification, and others experience high temperature during 96-hour Temperature summer months and irrigation drawdown. Tolerance Tests Summertime stream temperatures can exceed Eight 96-hour mortality tests (3 February to lethal limits. This study was conducted to 7 April 1997) were used to compare the toler- determine inherent differences among avail- ance of the 4 cutthroat trout stocks to high able cutthroat trout stocks in the ability to temperature. All tests were conducted in insu- withstand water quality extremes, possibly lated 800-L indoor circular tanks. Airstones in making some stocks more suitable for stocking each tank insured that low DO did not com- in harsher environments. promise survival. For each 1997 test, we transferred fish METHODS directly from acclimation temperatures (13.6 ± 0.6°C or 18.0 ± 0.5°C) into test tanks. We used We reared 4 cutthroat trout stocks from 4 cylindrical cages (56-cm diameter) per tank, wild broodstocks from the sources mentioned placing 10 fish of one stock per cage. This above from eggs at the Fisheries Experiment insured that all stocks within a tank were Station, Logan, Utah. Stocks were tested for exposed to the same water temperature. Test survival differences at extremes of tempera- temperatures ranged from 23.0°C to 24.5°C ture, dissolved oxygen (DO), and salinity. We among the 8 tests. Two or 3 tanks per temper- divided test fish into 2 groups, one acclimated ature treatment provided the replication for to 13.6 ± 0.6°C and the other to 18.0 ± 0.5°C. each test. Fish acclimated to 13.6°C had been reared at Three additional 96-hour tests were con- that temperature since hatching. Fish at 18.0°C ducted in January 2000, one with the 4 stocks were transferred to that temperature on 10 acclimated to 12.5°C and the other two with September 1996 and remained there for at fish acclimated to 16.0°C for at least 30 days. least 30 (CTM tests) to 90 days (96-hour tests) The study fish were freeze-branded 1 or 2 prior to tests. Additional 96-hour temperature weeks prior to testing to identify the stocks. tests were conducted in January 2000 with the For each test, 20 fish from each stock were put 4 stocks acclimated to either 12.5°C or 16.0°C. into each of 5 circular tanks, 3 of which had The temperature difference from previous tri- recirculating heaters to control temperature als was the result of variation among years in and 2 that served as controls. the well water. Hatchery well water had a total Water exchange of half volume was per- ⋅ –1 hardness as CaCO3 of 222 mg L , pH of 7.5 formed once or more (first 2 tests at 18.0°C to 7.6, and total alkalinity of 222 mg ⋅ L–1. acclimation temperature), halfway through each test using water of the appropriate tem- Critical Thermal Maximum perature. DO was monitored daily with an Fish were selected at random and tested oxygen meter calibrated with replicate Win- individually; 10 fish were sampled from each kler titrations (APHA et al. 1989) and did not ⋅ –1 stock and acclimation temperature. We placed drop below 5.2 mg L . NH3–N was deter- each fish in a 4.0-L Erlenmeyer flask contain- mined by Nesslerization (APHA et al. 1989) at ing 3.0 L of 13°C water. Temperature was the end of each test, and levels did not exceed 436 WESTERN NORTH AMERICAN NATURALIST [Volume 61

0.019 mg ⋅ L–1. We weighed each fish after it effects on resistance time. The temperature had died or at the end of the test. Mortality was 14.7 ± 0.8°C for each test. was recorded after 96 hours. Mean weights We performed 2 tests in which dissolved during the temperature tests were as follows: oxygen (DO) was gradually decreased over 4.9–23.7 g, BL; 4.0–23.6 g, BV; 2.4–19.8 g, SN; 6–7 hours; resistance time (time at which equi- and 4.5–17.6 g, EL. librium was lost) and DO at the time of loss of equilibrium were noted. Ten fish from each Salinity Tolerance stock were maintained in plastic-mesh cages Seven salinities were tested in 2 separate for each test, 4 cages per 800-L tank. Mean 24-hour tests. Salinities of 0.4 (control), 29.6, weights for these tests were BL, 21.9 g; BV, 32.3, and 33.5 mS ⋅ cm–1 (0, 18.0, 19.0, and 15.7 g; SN, 10.0 g; and EL, 22.1 g. Test-1 fish 20.0‰, respectively) were evaluated in test 1; were acclimated to the cages for 40 hours and and salinities of 0, 36.1, 38.2, and 41.6 mS ⋅ not fed for 96 hours prior to the test. Test-2 cm–1 (0, 21.0, 22.0, 23.5‰) were evaluated in fish were fed until test time and given no test 2. We used 2 replicate 800-L circular acclimation time in the cages. Of interest was tanks per treatment, stocking 10 fish per stock the comparative response of the 4 stocks to into cylindrical cages in each tank. Salinities hypoxia under the varied conditions of the 2 were adjusted by adding noniodized rock salt tests, avoiding possible biases due to handling to the water and dissolving it prior to testing. stress and oxygen demand for digestion. Water between replicate tanks was exchanged Remaining tests were 24-hour challenges at a given DO; average levels ranged from 1.85 prior to adding fish to minimize salinity differ- ⋅ –1 ences between replicates. Salinity was mea- to 3.34 mg L among the 6 tests. The 4 cutthroat trout stocks were freeze-branded one sured by a specific conductivity probe (Hydro- week prior to tests to identify stocks upon lab, Austin, TX). Other variables measured removal from a tank. Fish were transferred concurrently included temperature (14.4 ± from outdoor raceways in which DO ranged 0.4°C) and dissolved oxygen (5.7–7.3 mg ⋅ L–1, from 6.6 to 7.8 mg ⋅ L–1. For each test we put maintained with airstones bubbling com- 10 fish from each stock into each of 2 circular pressed air). Un-ionized ammonia nitrogen at tanks holding 400 L of water. During each test the end of the test did not exceed 0.008 mg ⋅ –1 we monitored DO and removed the fish upon L . Mortality was recorded after 24 hours. loss of equilibrium, weighed them in water, Total body weights of mortalities were and noted the time. Mean weights during these recorded and compared to live weights sepa- tests were 25.2–30.7 g, BL; 16.1–21.6 g, BV; rately for each stock using a t test for paired 8.8–11.6 g, SN; and 20.2–32.3 g, EL. samples. Statistical Analysis Low Dissolved Oxygen Tolerance LEC50, i.e., the lower limit of dissolved oxygen causing loss of equilibrium within 24 hours, Between 15 April and 15 May 1997, we or LT50, the maximum temperature in 96-hour conducted 8 tests using fish acclimated to tests that killed half the fish, was calculated for 13°C. By adding nitrogen gas to the water in a each stock by probit analysis (Newman 1994, flow-through system similar to that described SPSS 1994). For DO data we used natural log by Cochran and Babcock (1974), we manipu- transformation for probit analysis; we also con- lated dissolved oxygen. Two 800-L circular ducted analysis with stocks pooled. Probit tanks were used for each test, and each re- analysis for each acclimation temperature was ceived flows of 11 ± 1 L ⋅ min–1. Un-ionized not possible due to insufficient data, and so ammonia determined by Nesslerization at the data were pooled. Percent mortality was arc- end of 2 tests did not exceed 0.001 mg ⋅ L–1. sine transformed prior to 1-way analysis of Fish were allowed access to the surface to variance (ANOVA) within each test. Normality gulp for air. Loss of equilibrium was the end tests (Kolmogorov-Smirnov) were conducted point (resistance time) for each fish in all tests, for each continuous variable, and the data and most recovered when returned to normoxic were subsequently analyzed by 1-way ANOVA water. Weight of each fish was measured after if the data were normally distributed and, if loss of equilibrium to evaluate possible size not, by the Kruskal-Wallis 1-way ANOVA (SPSS 2001] COMPARATIVE TOLERANCE OF CUTTHROAT TROUT 437

1993). Duncan’s test was used for mean com- was significantly higher for SN (75%) than the parisons. Using 1-way ANOVA, we analyzed other 3 stocks (10–15%). After raising the test separately transformed mortality data from the temperature half a degree, mortality reached salinity trials for each salt concentration. For 100% in SN and BL groups, but was less for EL DO data analysis, median resistance time and BV (75% and 65%, respectively). Probably (average elapsed time of 5th and 6th fish) and due to having only 2 replicates, these stock average DO at that time were used to com- differences were not significant (P = 0.21, pare stocks in 1-way ANOVA. Simple least- Kruskal-Wallis test). A repeat of this test squares regression was used to test the resulted in significantly lower mortality for BV strength of the relationship between total body (27%) than for SN (87%). Repetition of the weight and DO resistance time. To assess total tests with another year-class of fish acclimated body weight effects on mortality in the tem- to 16.0°C provided similar results; BV had sig- perature tests, we compared average live and nificantly lower mortality (1.7%) than EL dead weights by Wilcoxon matched-pairs and (30.0%) or SN (68.3%) when challenged at signed-ranks tests for each temperature test 24.0°C (Table 2). Probit analysis resulted in ° ° (stocks pooled), stock (tests pooled), and for LT50 values of 23.5 C for SN, 23.9 C for BL, combined data. Differences were considered 24.0°C for EL, and 24.3°C for BV (Fig. 1). significant at P ≤ 0.05 for all statistical tests. Overall, BV had slightly better survival at high temperatures than the other stocks, but the RESULTS difference was small. Mean weight of mortalities was not signifi- Critical Thermal cantly different from survivor weight for each Maximum Tests stock when all tests were pooled for analysis. Mean CTMeq and CTMs were significantly Analysis of pooled data combining tests and higher for fish acclimated at 18.0°C (29.5°C stocks (49 pairs) similarly resulted in no signif- and 29.7°C, respectively) than those acclimated icant difference. at 13.6°C (28.1°C, 28.6°C). There were no sig- Salinity Tolerance Tests nificant differences in CTMeq among stocks ° acclimated to 18.0 C, but CTMs was signifi- There were significant differences among cantly higher (P = 0.03) for BL (30.0) than for stocks in 24-hour survival at high salinities SN (29.6) or BV (29.7°C; Table 1). For tests (Table 3). SN had the lowest salinity tolerance, with fish acclimated at 13.6°C, there were no with significant mortality occurring at 29.6 mS significant differences among the 4 stocks in ⋅ cm–1 conductivity (18‰); BV had the highest –1 CTMeq or CTMs. tolerance (<5% mortality up to 38 mS ⋅ cm [22‰]). BL and EL were intermediate, with 96-hour Temperature 60% or 30% mortality, respectively, at 36 mS ⋅ Tolerance Tests cm–1 (21‰). Size appeared to have little influ- In 96-hour tests, differences among stocks ence on mortality; for only BV, the mean varied among the 11 tests (Table 2). For fish weight of dead fish was smaller than that of acclimated to 13.6°C and challenged at 24.0°C, survivors (1.8 versus 2.4 g, respectively, P = mortality for BL and SN was significantly higher 0.032, t test for paired samples). (100%) than for EL (73%) and BV (13%). At 23.0°C and 23.5°C, results were much differ- Dissolved Oxygen ent; BV had significantly higher mortality than Tolerance Tests the other stocks or did not significantly differ Using the method of gradually decreasing (Table 2). When cold-acclimated fish were DO over time, resistance times (elapsed time challenged at 24.0°C with a larger sample size until loss of equilibrium) among the 4 stocks per tank, BV had the lowest mortality (3.3%) did not significantly differ for either of the 2 and SN the highest (86.7%; Table 2). trials (Table 4). Similarly, mean DO at the Five 96-hour tests were conducted with median resistance time did not differ signifi- fish acclimated to 18.0°C. At 23.2°C, only the cantly among stocks. A t test for differences in EL experienced mortality (3%). At 23.7°C, BV DO between tests (stocks pooled) was not sig- had significantly lower mortality (21%) than nificant, indicating that acclimation to tanks or SN (95%) or BL (90%). Mortality at 23.9°C time off feed had little impact on the lower 438 WESTERN NORTH AMERICAN NATURALIST [Volume 61

TABLE 1. Mean (n = 10, ±s) critical thermal maximum values (°C) for 4 cutthroat trout stocks (BL, Bear Lake Bon- neville; BV, southern Bonneville from Manning Meadow Reservoir; SN, fine-spotted Snake River; and EL, Electric ° ° Lake) acclimated to either 13.6 C or 18.0 C. Critical thermal maximum was at first loss of equilibrium (CTMeq) and at ≥ the onset of spasms (CTMs). Means not significantly different (1-way ANOVA, P 0.05) among stocks within a column are followed by a common letter or no letter.

______Acclimation temperature ° ° ______13.6 C ______18.0 C

Stock CTMeq CTMs CTMeq CTMs BL 27.9 ± 0.68 28.6 ± 0.55 29.7 ± 0.20 30.0 ± 0.16 a BV 28.3 ± 0.41 28.7 ± 0.34 29.5 ± 0.39 29.7 ± 0.32 b SN 28.2 ± 0.31 28.6 ± 0.32 29.4 ± 0.41 29.6 ± 0.39 b EL 28.1 ± 0.23 28.5 ± 0.25 29.6 ± 0.24 29.8 ± 0.19 ab Average 28.1 ± 0.45 28.6 ± 0.38 29.5 ± 0.34 29.7 ± 0.31

TABLE 2. Summary of mean percent mortality (n = 2 or 3 tanks within a test) recorded for 4 cutthroat trout stocks exposed to high temperatures in a series of tests. Stock abbreviations: BL = Bear Lake, BV = Bonneville, SN = Snake River fine-spotted, and EL = Electric Lake. A common subscript letter or no letter among means within a test indicates no significant difference (1-way ANOVA, Kruskal-Wallis, P ≥ 0.05). Number of Test fish per Percent mortality Acclimation temperature stock per ______temperature (°C) tank BL BV SN EL 12.5 ± 0.1°C 24.0 ± 0.5 20 41.7 ab 3.3 a 86.7 b 21.7 a 13.6 ± 0.6°C 23.0 ± 0.9 10 20.0 a 100.0 b 13.3 a 20.0 a 23.5 ± 0.5 10 36.7 33.3 3.3 23.3 24.0 ± 0.5 10 100.0 a 13.3 c 100.0 a 73.3 b 16.0 ± 0.2°C 23.5 ± 0.5 20 0.0 0.0 0.0 6.7 24.0 ± 0.3 20 15.0 ab 1.7 a 30.0 b 68.3 c 18.0 ± 0.4°C 23.2 ± 0.8 10 0.0 0.0 0.0 3.3 23.7 ± 1.3 10 90.0 a 21.2 b 95.0 a 51.4 ab 23.9 ± 0.7 10 15.0 a 15.0 a 75.0 b 10.0 a 24.2 ± 0.5 10 73.3 ab 27.0 b 86.7 a 85.2 b 24.5 ± 0.6 10 100.0 65.0 100.0 75.0 lethal limit of DO. Mortality began when DO was not significantly different among stocks in dropped below 1.9 mg ⋅ L–1 (test 1) or 1.7 mg ⋅ most tests, except 1 trial (DO = 2.18 mg ⋅ L–1) L–1 (test 2). The last fish survived until DO in which BL survived significantly longer than dropped to 1.2 to 1.3 mg ⋅ L–1. the other stocks (Table 4). Probit analysis of In the 24-hour tests, mortality began when mortality in the 24-hour tests indicated little DO dropped below 2.6 mg ⋅ L–1. Least-squares difference in hypoxia tolerance among stocks regression for each of these tests indicated no (Fig. 1). When stocks were pooled, the LEC50 significant relationship or a poor relationship of DO was 2.34 mg ⋅ L–1 (27% of saturation or 2 (e.g., r = 0.254, DO test 8) between individ- PO2 = 35.3 mm Hg). ual total body weight and the resistance time to low DO levels for any of the 4 stocks. There DISCUSSION were no significant differences in mortality among stocks except for 1 test in which SN Inherent differences in thermal tolerance experienced higher mortality (30%) than the among the 4 stocks of cutthroat trout were not other stocks (≤5%; Table 5). Resistance time evident based upon critical thermal maximum 2001] COMPARATIVE TOLERANCE OF CUTTHROAT TROUT 439

TABLE 3. Comparison of percent mortality of 4 cutthroat trout stocks after 24-hour exposure to various salinities (presented as specific conductivity and parts per thousand [in parentheses]). Percent mortality is the average of 2 tanks with 10 fish per stock per tank. Stock abbreviations: BL = Bear Lake, BV = Bonneville, SN = Snake River fine-spotted, and EL = Electric Lake stock. A common subscript letter or no letter among means within a given salinity indicates no significant difference (1-way ANOVA, P ≥ 0.05). Percent mortality Salinity ______mS ⋅ cm–1 (‰) BL BV SN EL 0.4 (0.0) 0 0 0 0 29.6 (18.0) 0 0 10 0 32.3 (19.0) 0 0 40 5 33.5 (20.0) 0 a 5 a 66 b 0 a 36.1 (21.0) 60 c 0 a 95 d 30 b 38.2 (22.0) 80 ab 45 a 100 b 84 a 41.6 (23.5) 100 a 75 b 100 a 100 a

Basin, this stock would likely have been exposed to selective pressures from high summer temperatures. Duff (1988) reported Bon- neville cutthroat trout found in “small headwater streams with degraded habitat and warm summer water temperature (21°C).” Differ- ences in LEC50 values were minor, but stock differences may be accentuated if fish are exposed to daily fluctuating high temperatures ± Fig. 1. Lethal temperature (LT50 95% confidence instead of constant high temperatures, where interval) and dissolved oxygen concentration inducing loss ± fish can recover overnight. For example, Otto of equilibrium in 50% (LEC50 95% confidence interval) of each of 4 cutthroat trout stocks (Bear Lake Bonneville, (1974) observed higher thermal tolerance in BL; southern Bonneville, BV; Snake River fine-spotted, western mosquitofish (Gambusia affinis affinis) SN; and Electric Lake, EL). exposed to cyclic high temperatures than in those exposed to constant temperatures. Fem- inella and Matthews (1984) reported similar (CTM) tests. CTM tests conducted by Lee and findings for the orangethroat darter (Etheo- Rinne (1980) indicated little difference between stoma spectabile). The biological significance the native Gila (Oncorhynchus gilae) and Apache of our results still requires field testing, but trout (O. apache) and introduced trout species warm-temperature adaptation might be a useful such as rainbow, brown (Salmo trutta), and trait for maintenance of trout populations in brook trout (Salvelinus fontinalis). Similar to shallow reservoirs and streams with high sum- results of this study, Lee and Rinne (1980) and mer temperatures. Intraspecific differences in Lohr et al. (1996) also noted an increase in CTM thermal tolerance have been observed by McCauley (1958) for arctic char (Salvelinus at higher acclimation temperatures. CTM val- alpinus), but not for brook trout. Bidgood and ues for cutthroat trout in this study (28.1– ° Berst (1969) did not detect any difference in 29.8 C) were similar to those reported for thermal tolerance among rainbow trout from 4 other salmonids at similar acclimation temper- different Great Lakes stocks. atures (Lee and Rinne 1980, Lohr et al. 1996). The upper incipient lethal temperature Contrary to CTM tests, 96-hour tests gen- (UILT) is defined as the upper temperature at erally indicated that BV had a greater thermal which 50% mortality is observed at a given tolerance than the other stocks. Given the nat- acclimation temperature (Amour 1991). In this ural habitats found in the southern Bonneville study cutthroat trout had UILT limits ranging 440 WESTERN NORTH AMERICAN NATURALIST [Volume 61

TABLE 4. Average (n = 2 tanks, 10 fish per stock per tank) of the median time to loss of equilibrium (resistance time) for each of 4 cutthroat trout stocks (Bear Lake Bonneville, BL; Manning Meadow Bonneville, BV; Snake River fine-spotted, SN; and Electric Lake, EL), and average dissolved oxygen (DO ± range) level at the median time for each of 6 tests in which fish were exposed to low dissolved oxygen levels for 24 hours or to gradually decreasing levels. Means within a test that are not significantly different share a common letter or no letters (1-way ANOVA, P ≥ 0.05).

Dissolved oxygen ______Resistance time (minutes) (mg ⋅ L–1) BL BV SN EL Test

DECREASING DO 1.34 ± 0.13 335 273 242 235 2 1.39 ± 0.07 340 360 275 293 1

CONSTANT DO 1.85 ± 0.16 125 63 108 126 4 1.99 ± 0.18 289 116 295 154 5 2.18 ± 0.11 634 a 310 b 401 b 371 b 6 2.29 ± 0.10 518 803 457 314 8

from 23.6°C to 24.3°C. Other researchers have 2 species of darters (Ingersoll and Claussen indicated that rainbow trout tolerate slightly 1984). The 2 methods measure different re- higher values at similar acclimation tempera- sponses and thus should both be conducted tures; Bidgood and Berst (1969) and Charlon for future temperature studies. et al. (1970) both noted UILT values ranging As in this study, size differences in thermal from 25°C to 26°C. Arctic grayling (Thymallus tolerance were not significant for studies with arcticus) in Montana had UILTs of 23.0°C for brook trout and arctic char (McCauley 1958), fish acclimated to 8.4°C or 16.0°C, or 25.0°C nor for perch (Perca flavescens; Hathaway 1927, for those acclimated to 20.0°C (Lohr et al. Hart 1952). Smale and Rabeni (1995) similarly 1996). found no difference in tolerance related to size CTM tests resulted in higher lethal tem- in tests of 34 species found in Missouri head- perature values than 96-hour tests, similar to water streams. tests with arctic grayling (Lohr et al. 1996). The Salinity tolerance among salmonids seems 2 methods differ in that CTM is a progressive to vary considerably with species and age change in temperature resulting in a “physical (Clarke 1982, Blackburn and Clarke 1987). For disorganization response,” whereas the UILT example, salinities causing 50% mortality of method is an abrupt change resulting in a lethal coho salmon increased from 33‰ in early response (Becker and Genoway 1979). The February to 41‰ in mid-April as the fish pro- CTM method has been useful for comparative gressed through the parr-to-smolt transforma- studies, such as that of Kowalski et al. (1978) tion (Blackburn and Clarke 1987). Varnavskiy of New York stream fishes. Also, Matthews and Varnavskaya (1984) reported significant and Maness (1979) observed significant differ- mortality in 28‰ seawater for sockeye and ences in CTM values among cyprinid species coho salmon. Tolerances are also complicated that correspond to fluctuations in fish popula- by the fact that salt in the diet can improve tions in the South Canadian River in Okla- salt tolerance (Basulto 1976). Rainbow trout homa. However, as the 96-hour tests indi- have been reared in water with salinities of cated, CTM values should not be regarded as 30‰ after 9 days of acclimation (Murai and the maximum “safe” temperature for a fish at a Andrews 1973). Salt-tolerant strains of rain- given acclimation temperature. Mortality at a bow trout have been developed for marine net given temperature can occur over a much pen rearing in Finland (Stickney 1991). Tatum longer period than that allowed in the CTM (1973) successfully reared rainbow trout in tests. In this study the 96-hour tests were cages in salinities of 20‰ after a 24-hour more sensitive indicators of differences in acclimation period. We found cutthroat trout temperature tolerance, detecting differences tolerated salinities approaching 18–22‰. where CTM tests did not. Incongruous results Salinity tolerance was quite different among were not limited to this study. CTM tests were the 4 stocks, with highest to lowest tolerance inconsistent with thermal preference tests for following the order BV > EL > BL > SN; i.e., 2001] COMPARATIVE TOLERANCE OF CUTTHROAT TROUT 441

TABLE 5. Mean (n = 2 tanks with 10 fish per stock per tank) mortality of 4 cutthroat trout stocks (Bear Lake Bon- neville, BL; southern Bonneville, BV; Snake River fine-spotted, SN; and Electric Lake, EL) exposed to low levels of dissolved oxygen (mean ± range) over a 24-hour period in 6 separate tests. Means within a test not significantly different share a common letter or no letters (1-way ANOVA, P ≥ 0.05).

Dissolved oxygen ______Mortality (%) (mg ⋅ L–1) BL BV SN EL Test 1.85 ± 0.16 100 100 100 100 4 1.99 ± 0.18 95 100 95 100 5 2.18 ± 0.11 90 95 90 100 6 2.29 ± 0.10 67 82 75 80 8 2.40 ± 0.15 5 a 0 a 30 b 5 a 7 3.34 ± 0.71 0000 3

the southern stock was most tolerant and the also found differences between the southern northern stocks the least. The Great Basin and and northern forms of the Bonneville cut- Intermountain West have experienced several throat trout, particularly in temperature and fluctuations of wetter and drier climatic peri- salt-tolerance limits. ods over the last 25,000 years (Bright 1963, DO limits have been reviewed by Barton Smith 1978), which may have altered ancient and Taylor (1996), Davis (1975), and Doudoroff lake levels and salinity. Lake Bonneville is and Shumway (1970), who summarized that reported to have undergone 4 periods of low mortality for most fish occurs at concentrations water levels between 8,000 and 22,000 years between 1 and 3 mg ⋅ L–1. Cutthroat trout in ago, including one period of complete dessica- this study had lower DO threshold values (1.9 tion followed by refilling that occurred 11,000 mg ⋅ L–1 or 22% of saturation or 28.6 mm Hg, years ago (Hickman 1978). Such conditions PO2) when DO levels were steadily dropped may have exerted selective pressures on the compared to 24-hour tests (2.3–2.5 mg ⋅ L–1 or cutthroat trout. Galat et al. (1985) noted that 26–29% saturation). Differences in tolerance Lahontan cutthroat trout, which evolved in to hypoxia were not evident among the 4 stocks, Pleistocene Lake Lahontan that underwent indicating little difference in natural selection similar water level fluctuations, have adapted for that trait. For rainbow trout in 24-hour to salinities approaching 11.9‰ (15.1 mS ⋅ tests, Alabaster et al. (1957) reported a median cm–1). tolerance level of 2.6–2.7 mg ⋅ L–1. Steelhead Cutthroat trout subspecies can be divided (O. mykiss) at 16–20°C died at 1.6–1.7 mg ⋅ into 3 major groups: coastal (O. c. clarki) char- L–1 in tests by McNeil (1956, cited by Warren acterized by 68 chromosomes, westslope (O. c. and Bouck 1973). The discrepancy between lewisi) characterized by 66 chromosomes, and these 2 studies may be stock related since interior cutthroat trout characterized by 64 intraspecific differences in DO tolerance have chromosomes (Loudenslager and Gall 1980, been observed in rainbow trout (Klar et al. Behnke 1981, 1988). Using starch gel elec- 1979). The 4 cutthroat trout stocks in this study trophoresis, Loudenslager and Gall (1980) fur- had incipient lethal limits that fell between ther subdivided the interior cutthroat trout these 2 reported ranges. Hepworth et al. into 2 additional groups: (1) Lahontan cutthroat (1999) indicated that cutthroat trout survived trout and (2) those cutthroat trout inhabiting overwinter in a reservoir whereas rainbow the Colorado, Yellowstone, and Upper Snake trout did not. Controlled studies comparing rivers, and the Bonneville Basin. Further study rainbow trout and cutthroat trout stocks are of this latter group by Martin et al. (1985) using needed to better define these DO tolerance the same techniques indicated that differences differences. were evident between northern (Bear River In DO studies with other salmonids, Katz drainage) and southern (Sevier River drainage) et al. (1959) noted that resting juvenile chinook forms of the Bonneville cutthroat trout; they salmon (Oncorhynchus tschawytscha) mortality also suggested that fish from the Wasatch Front at 20°C occurred at 1.4–1.9 mg ⋅ L–1 in 24- streams of Utah may form a 3rd group. We hour tests. At 10°C, Klyashtorin (1975) reported 442 WESTERN NORTH AMERICAN NATURALIST [Volume 61 lower threshold values for chinook salmon (1.6 concentrations of dissolved oxygen and carbon diox- mg ⋅ L–1, 19.5 mm Hg), sockeye salmon (O. ide. Annals of Applied Biology 45:177–188. ⋅ –1 AMOUR, C.L. 1991. Guidance for evaluating and recom- nerka; 1.4 mg L , 19.0 mm Hg), coho salmon mending temperature regimes to protect fish. U.S. (O. kisutch; 1.4 mg ⋅ L–1), and arctic char (Sal- Fish and Wildlife Service Biological Report 90(22). velinus alpinus; 1.4 mg ⋅ L–1). Burdick et al. U.S. Fish and Wildlife Service, Washington, DC. (1954) noted mortality of individual brook APHA (AMERICAN PUBLIC HEALTH ASSOCIATION), AMERI- ⋅ –1 CAN WATER WORKS ASSOCIATION, AND WATER POL- trout occurred at 1.15–3.40 mg O2 L and LUTION CONTROL FEDERATION. 1989. Standard meth- individual rainbow trout occurred at 0.81–2.47 ods for the examination of water and wastewater. ⋅ –1 17th edition. American Public Health Association, mg O2 L , depending upon the temperature. Tests by Burdick et al. (1954) and Klyashtorin Washington, DC. 1543 pp. BARTON, B.A., AND B.R. TAYLOR. 1996. Oxygen require- (1975) were conducted using the sealed-vessel ments of fishes in northern Alberta rivers with a method, which can lead to underestimation of general review of the adverse effects of low dis- actual lethal levels (Doudoroff and Shumway solved oxygen. Water Quality Research Journal of 1970). Shepard (1955) noted that brook trout Canada 31:361–409. ° ⋅ –1 BASULTO, S. 1976. Induced saltwater tolerance in connec- held at 9 C all survived 1.9 mg L (17% sat- tion with inorganic salts in the feeding of Atlantic uration) for 5 days, but died at concentrations salmon (Salmo salar). Aquaculture 8:45–55. below this. BECKER, C.D., AND R.G. GENOWAY. 1979. Evaluation of The literature on the effect of size upon tol- the critical thermal maximum for determining thermal tolerance of freshwater fish. Environmental Biology erance to low DO has been contradictory. of Fishes 4:245–256. Shepard (1955) reported that small fish died BEHNKE, R.J. 1981. Systematic and zoogeographical inter- more quickly than large fish. Wells (1913) and pretation of Great Basin trouts. Pages 95–124 in R.J. Naiman and D.L. Soltz, editors, Fishes in North Keys (1931) observed similar results, but it American deserts. Wiley and Sons, New York. was not evident in our study. Similarly, no cor- ______. 1988. Phylogeny and classification of cutthroat relation between size and resistance time to trout. Pages 1–7 in R.E.Gresswell, editor, Status and low DO was observed by Alabaster et al. (1957) management of interior stocks of cutthroat trout. American Fisheries Society Symposium 4, Bethesda, in studies with rainbow trout and perch (Perca MD. fluviatilis). Doudoroff and Shumway (1970) re- ______. 1992. Native trout of western North America. viewed several studies and also found inconsis- American Fisheries Society, Monograph 6, Bethesda, tencies in the effect of size upon DO tolerance. MD. 275 pp. BIDGOOD, B.F., AND A.H. BERST. 1969. Lethal tempera- Overall, results indicated differences in tures for Great Lakes rainbow trout. Journal of the water quality tolerance among the cutthroat Fisheries Research Board of Canada 26:456–459. trout stocks tested, especially to high tempera- BLACKBURN, J., AND W.C. CLARKE. 1987. Revised proce- ture and salinity. These differences accentuate dure for the 24 hour seawater challenge test to measure seawater adaptability of juvenile salmonids. the importance of preserving the genetic diver- Canadian Technical Report of Fisheries and Aquatic sity of the individual stocks for greater flexibil- Sciences 1515. ity in fisheries management and stock survival. BRIGHT, R.C. 1963. Pleistocene lakes Thatcher and Bon- These adaptations may make the southern neville, southeastern Idaho. Doctoral dissertation, University of Minnesota, St. Paul. Bonneville stock a better candidate for stock- BURDICK, G.E., M. LIPSCHUETZ, H.J. DEAN, AND E.J. ing in waters of marginal temperature or salin- HARRIS. 1954. Lethal oxygen concentrations for trout ity than the other stocks tested. and smallmouth bass. New York Fish and Game Journal 1:84–97. CHARLON, N., B. BARBIER, AND L. BONNET. 1970. Résis- ACKNOWLEDGMENTS tance de la truite arc-en-ciel (Salmo gairdneri Richardson) a des variations brusques de tempéra- Thanks to M.D. Routledge, R. Mellenthin, ture. Anales d’Hydrobiologie 1:73–89. and Q. Bradwisch for culturing fish for the study CLARKE, W.C. 1982. Evaluation of the seawater challenge test as an index of marine survival. Aquaculture 28: and to M. Smith for assistance in laboratory 177–183. tests. The project was supported by the Utah COCHRAN, J.O., AND W.H. BABCOCK. 1974. A device for Division of Wildlife Resources and the Sport maintaining constant oxygen concentration in flow- Fish Restoration Program, Project F-96-R. ing water. Progressive Fish-Culturist 36:177–178. DAVIS, J.C. 1975. Minimal dissolved oxygen requirements of aquatic life with emphasis on Canadian species: a LITERATURE CITED review. Journal of the Fisheries Research Board of Canada 32:2295–2332. ALABASTER, J.S., D.W.M. HERBERT, AND J. HEMENS. 1957. DOUDOROFF, P., AND D.L. SHUMWAY. 1970. Dissolved oxy- The survival of rainbow trout (Salmo gairdnerii gen requirements of freshwater fishes. FAO Fish- Richardson) and perch (Perca fluviatilis L.) at various eries Technical Paper 86. 2001] COMPARATIVE TOLERANCE OF CUTTHROAT TROUT 443

DUFF, D.A. 1988. Bonneville cutthroat trout: current sta- LEARY, R.F., F.W. ALLENDORF, S.R. PHELPS, AND K.L. tus and management. Pages 121–127 in R.E. Gress- KNUDSEN. 1987. Genetic divergence and identifica- well, editor, Status and management of interior stocks tion of seven cutthroat trout subspecies and rainbow of cutthroat trout. American Fisheries Society Sym- trout. Transactions of the American Fisheries Society posium 4, Bethesda, MD. 116:580–587. DWYER, W.P. 1990. Catchability of three strains of cut- LEE, R.M., AND J.N. RINNE. 1980. Critical thermal max- throat trout. North American Journal of Fisheries ima of five trout species in the southwestern United Management 10:458–461. States. Transactions of the American Fisheries Soci- FEMINELLA, J.W., AND W. J. M ATTHEWS. 1984. Intraspecific ety 109:632–635. differences in thermal tolerance of Etheostoma spec- LOHR, S.C., P.A. BYORTH, C.M. KAYA, AND W. P. D WYER. tabile (Agassiz) in constant versus fluctuating envi- 1996. High-temperature tolerances of fluvial Arctic ronments. Journal of Fish Biology 25:455–461. grayling and comparisons with summer river tem- GALAT, D., G. POST, T.J. KEEFE, AND G.R. BOUCK. 1985. peratures of the Big Hole River, Montana. Transac- Histological changes in the gill, kidney and liver of tions of the American Fisheries Society 125:933–939. Lahontan cutthroat trout, Salmo clarki henshawi, liv- LOUDENSLAGER, E.J., AND G.A.E. GALL. 1980. Geographic ing in lakes of different salinity-alkalinity. Journal of patterns of protein variation and subspeciation in Fish Biology 27:533–552. cutthroat trout, Salmo clarki. Systematic Zoology 29: GLIWICZ, Z.M., W.A. WURTSBAUGH, AND A. WARD. 1995. 27–42. Brine shrimp ecology in the Great Salt Lake, Utah. MARTIN, M.A., D.K. SHIOZAWA, E.J. LOUDENSLAGER, AND Performance report to the Utah Division of Wildlife J.N. JENSEN. 1985. Electrophoretic study of cutthroat Resources, Salt Lake City, UT. trout populations in Utah. Great Basin Naturalist HART, J.S. 1952. Geographic variations of some physiologi- 45:677–687. cal and morphological characters in certain freshwa- MATTHEWS, W.J., AND J.D. MANESS. 1979. Critical thermal ter fish. University of Toronto Studies, Biological maxima, oxygen tolerances and success of cyprinid Series 60. fishes in a southwestern river. American Midland Naturalist 102:374–377. HATHAWAY, E.S. 1927. Quantitative study of the changes produced by acclimation in the tolerance of high MCCAULEY, R.W. 1958. Thermal relations of geographic temperatures by fishes and amphibians. Bulletin of races of Salvelinus. Canadian Journal of Zoology 36: the U.S. Bureau of Fisheries 43:169–192. 655–662. MCNEIL, W.J. 1956. The influence of carbon dioxide and HEPWORTH, D.K., C.B. CHAMBERLAIN, AND M.J. OTTEN- pH on the dissolved oxygen requirements of some BACHER. 1999. Comparative sport fish performance freshwater fish. Master’s thesis, Oregon State Uni- of Bonneville cutthroat trout in three small put- versity, Corvallis. grow-and-take reservoirs. North American Journal MURAI, T., AND J.W. ANDREWS. 1973. Growth and food of Fisheries Management 19:774–785. conversion of rainbow trout reared in brackish and HEPWORTH, D.K., M.J. OTTENBACHER, AND L.N. BERG. fresh water. Fishery Bulletin 70:1293–1295. 1997. Distribution and abundance of native Bon- NEWMAN, M.C. 1994. Quantitative methods in aquatic neville cutthroat trout (Oncorhynchus clarki utah) in ecotoxicology. CRC Press, Boca Raton, FL. southwestern Utah. Great Basin Naturalist 57:11–20. NIELSON, B.R., AND L. LENTSCH. 1988. Bonneville cut- HICKMAN, T.J. 1978. Systematic study of the native trout throat trout in Bear Lake: status and management. of the Bonneville basin. Master’s thesis, Colorado Pages 128–133 in R.E. Gresswell, editor, Status and State University, Fort Collins. management of interior stocks of cutthroat trout. INGERSOLL, C.G., AND D.L. CLAUSSEN. 1984. Temperature American Fisheries Society Symposium 4, Bethesda, selection and critical thermal maxima of the fantail MD. darter, Etheostoma flabellare, and johnny darter, E. OTTO, R.G. 1974. The effects of acclimation to cyclic ther- nigrum, related to habitat and season. Environmen- mal regimes on heat tolerance of the western mos- tal Biology of Fishes 11:131–138. quitofish. Transactions of the American Fisheries KATZ, M., A. PRITCHARD, AND C.E. WARREN. 1959. The Society103:331–335. ability of some salmonids and a centrarchid to swim SHEPARD, M.P. 1955. Resistance and tolerance of young in water of reduced oxygen content. Transactions of speckled trout (Salvelinus fontinalis) to oxygen lack, the American Fisheries Society 88:88–95. with special reference to low oxygen acclimation. KEYS, A.B. 1931. A study of the selective action of decreased Journal of the Fisheries Research Board of Canada salinity and of asphyxiation on the Pacific killifish, 12:387–433. Fundulus parvipinnis. Bulletin of the Scripps Insti- SHIOZAWA, D.K., AND R.P. EVANS. 1994. Relationships be- tute of Oceanography, Technical Series 2:417–490. tween cutthroat trout populations from thirteen Utah KLAR, G.T., C.B. STALNAKER, AND T.M. FARLEY. 1979. Com- streams in the Colorado River and Bonneville drain- parative physical and physiological performance of ages. Final report to the Utah Division of Wildlife rainbow trout, Salmo gairdneri, of distinct lactate Resources, Salt Lake City, UT. 2 dehydrogenase B phenotypes. Comparative Bio- SMALE, M.A., AND C.F. RABENI. 1995. Hypoxia and hyper- chemistry and Physiology 63A:229–235. thermia tolerances of headwater stream fishes. Trans- KLYASHTORIN, L.B. 1975. The sensitivity of young Salmon- actions of the American Fisheries Society 124:698–710. idae to oxygen deficiency. Journal of Ichthyology SMITH, G.R. 1978. Biogeography of intermountain fishes. 15:337–341. Great Basin Naturalist Memoirs 2:17–42. KOWALSKI, K.T., J.P. SCHUBAUER, C.L. SCOTT, AND J.R. SPSS. 1993. SPSS® for Windows™ base system user’s SPOTILA. 1978. Interspecific and seasonal differences guide, release 6.0. SPSS Inc., Chicago, IL. in the temperature tolerance of stream fish. Journal ______. 1994. SPSS advanced statistics 6.1. SPSS Inc., of Thermal Biology 3:105–108. Chicago, IL. 444 WESTERN NORTH AMERICAN NATURALIST [Volume 61

STICKNEY, R.R. 1991. Effects of salinity on aquaculture during natural smoltification and in geothermal production. Pages 105–132 in D.E. Brune and J.R. water culture. Journal of Ichthyology 24:106–111. Tomasso, editors, Aquaculture and water quality. WAGNER, E.J. 1996. History and fluctuating asymmetry of World Aquaculture Society, Baton Rouge, LA. Utah salmonid broodstocks. Progressive Fish-Cul- TATUM, W.M. 1973. Brackish water cage culture of rainbow turist 58:92–103. trout (Salmo gairdneri) in south Alabama. Transac- WARREN, C.E., AND G.R. BOUCK. 1973. Development of tions of the American Fisheries Society 102:826–828. dissolved oxygen criteria for freshwater fish. Report TROJNAR, J.R., AND R.J. BEHNKE. 1974. Management impli- EPA-R3-73-019, U.S. Environmental Protection cations of ecological segregation between two intro- Agency, Washington, DC. duced populations of cutthroat trout in a small Colo- WELLS, M.M. 1913. The resistance of fishes to different rado lake. Transactions of the American Fisheries concentrations and combinations of oxygen and car- Society 103:423–430. bon dioxide. Biological Bulletin 25:323–427. VARNAVSKIY, N.S., AND N. VARNAVSKAYA. 1984. Assessment of some parameters characterizing the ion regulating Received 17 February 2000 system in coho salmon, Oncorhynchus kisutch, and Accepted 28 July 2000 sockeye salmon, Oncorhynchus nerka (Salmonidae) Western North American Naturalist 61(4), © 2001, pp. 445–451

DRUMMING BEHAVIOR AND LIFE HISTORY NOTES OF A HIGH-ALTITUDE COLORADO POPULATION OF THE STONEFLY ISOPERLA PETERSONI NEEDHAM & CHRISTENSON (PLECOPTERA: PERLODIDAE)

John B. Sandberg1 and Kenneth W. Stewart1

ABSTRACT.—Late instar nymphs and adults of a Colorado Hudsonian zone population of Isoperla petersoni Needham and Christenson were studied during the summer and early fall months of 1998, when stream temperature ranged from 3.3°C to 8.9°C. Early and middle instar nymphs were absent from July to October, and nymphs attained maximum size in mid-August during the 2nd week of adult field presence, suggesting a univoltine-slow life cycle. Nymphs were carnivorous and fed primarily on chironomid larvae. Adults were present from August to mid-October, with peak numbers of adult males and females occurring in mid-September and late September, respectively. Fecundity of field-collected females averaged 94.1 ± 45.15 eggs per female in September (N = 81) and decreased to 85.2 ± 57.6 in October (N = 12). Drumming duets were 2-way (male-female), and female answers followed male calls (nonoverlapped) or began before completion of the male call (overlapped). Male calls averaged 11.1 ± 2.09 beats and female answers averaged 6.2 ± 2.99, with mean intervals of 118 and 58 ms, respectively. The average number of female answer beats increased from 5.3 ± 2.56 when duets were nonoverlapped, to 6.6 ± 3.08 when overlapped.

Key words: stonefly, drumming, life history, Perlodidae, behavior.

The stonefly genus Isoperla is widely dis- generally biological indicators of good water tributed throughout the Holarctic region and quality, relatively little is known about their contains approximately 131 species worldwide behavior, life history, and ecology. Their drum- (Illies 1966, Zwick 1973). It represents the ming behavior, which can be a useful line of largest stonefly genus in North America, with evidence for delineating species (Stewart et al. 57 documented species. Szczytko and Stewart 1988), has been described for only 19 North (1979a) provided a revision of 20 western North American Isoperla species (Szczytko and Stew- American species (except I. decolorata Ricker), art 1979b, Maketon and Stewart 1984, Stewart including descriptions of the eggs and nymphs et al. 1988), and detailed life histories have of 19 and 12 species, respectively. Three west- been published for only 13 of them (Stewart ern species have subsequently been discov- and Stark 1988, Sandberg and Szczytko 1997) ered (Szczytko and Stewart 1984, Bottorff et largely because nymphs have not been corre- al. 1990a). lated with adults and therefore are unknown The large eastern fauna has remained a for most species. problematic group taxonomically and biologi- Isoperla petersoni is a widely distributed cally because of the cryptic external morphol- western species, from Alaska where it is com- ogy of adults and poorly known eggs and mon in most clearwater streams (Stewart et al. nymphs; it is currently under revision by S.W. 1990) southward to Utah. It was included for Szczytko (personal correspondence) and con- Colorado in a later electronic list of Stark et al. tains substantially more than the currently (1973), but that record has been deleted since recognized 33 species. Definitive separation of no literature record can be substantiated. Its most Isoperla species requires aedeagal ever- adult emergence and nymphal growth have sion and its description in males, and descrip- been described only from eastern Alaskan North tion of associated eggs, nymphs, and females. Slope populations (Stewart et al. 1990), and its Despite the diversity and importance of this drumming behavior has not been studied. In large group as lotic food web components and its southernmost range in Colorado, it occurs

1University of North Texas, Department of Biological Sciences, Denton, TX 76203.

445 446 WESTERN NORTH AMERICAN NATURALIST [Volume 61 at high elevations between 3109 and 3444 m Kogotus modestus (Banks). Mayflies collected and is one of the last regional Isoperla species include Drunella coloradensis (Dodds), Cinyg- to emerge (Baumann et al. 1977). This research mula ramaleyi (Dodds), and Cinygmula par was conducted on a population in Boulder (Eaton). County, Colorado, which we discovered in 1998. METHODS AND MATERIALS

SITE DESCRIPTION We collected late instar nymphs qualita- tively at approximately weekly intervals from This study was conducted on an I. petersoni 26 July to 26 September 1998 with a triangu- population from a northern unnamed tributary lar-frame kicknet fitted with a 600-µm-mesh of the North Fork Middle Boulder Creek, bag. Stream temperature during this period Boulder County, Colorado. The study reach is ranged from 3.3° to 8.9°C. Access by foot to a 1st-order stream located approximately the study site was delayed until July due to 40°00′27″N, 105°39′56″W, and is about 2 km extended presence of deep snow well into east of Arapaho Pass (3629 m) on the Arapaho June 1998 that limited ability to sample through Glacier Trail near the abandoned Fourth of the nymphal growth period. Senior author July Mine (elevation 3414 m). The study site is Sandberg traveled to Colorado in mid-Decem- in the Hudsonian life zone dominated by trees ber 1999 to attempt collection of nymphs for with stunted growth. The alpine zone begins winter growth determination. Avalanches in approximately 100 m north and 50 m west of the area caused trail closure and prevented the study site. The stream flows through a wet access to the study tributary. Typically, <33 meadow near scattered clusters of blue spruce nymphs were sampled from this sparse popu- (Picea pungens Engelman), Douglas-fir (Pseudo- lation and preserved weekly; they were hand- tsuga menziesii var. glauca (Beissn.), and Engel- picked from each kicknet sample that also man spruce (Picea engelmanni Parry ex Engel- contained moss and debris. Additional late man), most of which have been stunted by instar nymphs were transported live to the high winds and snow. The riparian vegetation laboratory for rearing. These we maintained in consists of short stands of willow (Salix sp.) styrofoam cups containing stream water from and alder (Alnus sp.), which are more preva- the North Fork Middle Boulder Creek, Buck- lent at the lower tributary reaches, and herba- ingham Campground (elevation 2896 m). Tem- ceous annuals and perennials including ele- perature and photoperiod were adjusted in a phants head (Pedicularis sp.), gentian (Gen- refrigerator to simulate stream environmental tiana sp.), Indian paintbrush (Castilleja sp.), conditions, and cups were checked twice daily fireweed (Epiobium sp.), and short grasses and for emergence. Reared adults were preserved sedges along the stream margins. in 80% ethanol. The tributary flows 1.5 km before joining We determined nymphal growth by inter- with the North Fork Middle Boulder Creek. ocular distance (IOD) measurements made The water source of the seeps and flowing with a calibrated ocular micrometer fitted to a springs forming the tributary consists of nearby stereo-dissection microscope. The IOD is the permanent snowfields and the South Arapaho shortest distance between the eyes. Sex was Glacier. Typical stream substrates include gravel, determined using the presence (female) or cobble, and boulders that form alternating rif- absence (male) of a gap in the posterior setal fles, small waterfalls, and pools. Areas that row of the 8th abdominal sternite (Stewart and include cobble substrate are tightly embedded Stark 1988). Food habit was determined from and limit the effectiveness of kicknet sam- midgut and hindgut contents, removed by dis- pling. However, much of the study reach subsection and identified to the lowest practical strate is covered with aquatic moss (Hygro- taxonomic level. hypnum sp.) that provides cover for nymphs Adult presence data and sex ratios were de- and emerging adults. Other stoneflies col- termined from weekly beating sheet samples lected in the tributary include Sweltsa borealis of adults taken from riparian vegetation within (Banks), Plumiperla diversa (Frison), Zapada 1 m of the stream margins. Collecting effort on haysi (Ricker), Megarcys signata (Hagen), and each sampling date was duplicated as closely 2001] COLORADO STONEFLY POPULATION 447 as possible and consisted of beating riparian RESULTS AND DISCUSSION vegetation for about 3 hours within 1 m of the stream margin. Emergence timing and sex Emergence was not observed in the field; ratios of laboratory-reared adults were com- however, we collected several exuviae using pared with presence and sex ratios of field-col- the beating sheet or hand-picked them from lected adults. exposed, moss-covered rocks. Adults reared in On 21 August 1999 we collected late instar the laboratory emerged both during the day nymphs at the study site for rearing. Eight vir- (N = 33) and evening (N = 27) throughout the gin males and 6 virgin females were success- emergence period and ecdysis required 20–30 fully reared in a Fridgid Units Living Stream™, minutes. In the field adults were found in low and drumming signals from pairings of these vegetation along the stream margin where were recorded at room temperatures of 20– they had crawled up to seek cover as has been 24°C, under fluorescent room light of about demonstrated for other Isoperla (Hynes 1967, 70 foot candles. Digital sound recordings were Jop and Szczytko 1984, DeWalt and Stewart made with a Sony MiniDisc (model MZ-R37) 1995). and Optimus Electret omnidirectional con- The presence of adults at North Fork Mid- denser microphones in a sound-dampened, dle Boulder Creek from early August through partitioned recording chamber described by the 1st half of October indicated that I. peter- Stewart and Zeigler (1984). We made digital, soni has an extended emergence. Isoperla computer-generated graphic facsimiles from petersoni adults (N = 18) from eastern Alaska recorded signals played into a computer and North Slope streams also emerged over an translated with the sound editing and analyz- extended period, from June to mid-August ing software Audiowave (Voyetra-Turtle Beach (Stewart et al. 1990). Therefore, in terms of Inc.) and Acid WAV (Polhedric Software). The length of adult presence, this southern-lati- latter program displayed each stereo channel tude Hudsonian zone population is similar to (male-left, female-right) with unique colors, the Alaskan population, but the onset of pres- assisting in the determination of signal charac- ence is later by 2 months (June vs. August). teristics. Clear, well-defined signals of both The 1st adult male was collected on 8 August males and females were measured to deter- 1998 when stream temperature was 7°C (Fig. mine number of drum beats, beat intervals to 1). The field sex ratio (male:female) of 225 nearest 1 ms, duet duration, and for overlapped adults was 1.1:1. A substantially lower sex duets the interval between the 1st female re- ratio of 0.4:1 was observed for the 67 individu- sponse beat and the male beat immediately als reared in the laboratory. Adult field pres- before it. For nonoverlapped duets, the male- ence lasted 68 days and was assumed com- female interval (interval between end of male plete when the last adult was observed on 10 call to start of female answer) was also mea- October. Although 0.6 m of snow had fallen on sured. 1 October, several adults were collected from Drumming amplitude of all recorded sig- willow brush after snow was carefully removed. nals was low; this complicated identification of The last 3 nymphs were collected in the field specific male call and female answer drum- on 26 September (Fig. 2), 2 weeks before the beats. We attempted audiocassette recordings end of adult field presence. made with a Marantz (model PMD 340) re- Mean fecundity of field-collected adult corder early in the recording sessions but females was 92.9 ± 46.44 eggs (N = 94), but abandoned them due to noise obliteration of decreased slightly over time from 94.1 ± 45.15 the low-amplitude drum beats, and therefore in September (N = 81) to 85.2 ± 57.63 in poor generation of graphic facsimiles. Digital October (N =12). This may have been due to recordings provided increased discrimination the small number of females remaining late in between drumbeats and noise interference. In the emergence cycle for analysis, or to the some instances the beat(s) of the overlapped possibility that later-emerging females were female answer fell at the exact time of male smaller. Smaller females late in the emergence call beat(s). These duets were not included in cycle have been noted for other stoneflies the analysis because the resolution power of (Khoo 1968, Schwarz 1970, Sheldon, 1972, the signal analysis software did not allow defin- Cather and Gaufin 1975, Orberndorfer and itive measurement of these “masked” events. Stewart 1977, Snellen and Stewart 1979), even 448 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Fig. 1. Field (N = 225) and laboratory (N = 67) adult presence and emergence of Isoperla petersoni from the North Fork Middle Boulder Creek, 1998. Field data indi- Fig. 2. Late instar growth of 127 Isoperla petersoni cated by dark symbols, laboratory data by light; males nymphs in North Fork Middle Boulder Creek, 1998. indicated by solid lines, females by dashed lines. Black circles = average female IOD, open circles = average male IOD, vertical lines = standard deviation, arrows = adult presence, N = male and female sample size. though we did not measure I. petersoni females. Laboratory-reared females produced 6 egg masses with average egg counts increasing both sexes then decreased for the remaining slightly from 92.5 ± 58.69 in late September sample dates. (N = 2) to 119.5 ± 14.25 in early October (N = Isoperla petersoni was carnivorous from 4). late July to mid-September. Chironomidae Winter growth of nymphs was not docu- (91.4%) and Ostracoda (5.6%) formed the mented due to inaccessibility of the study site majority of food items ingested (Table 1). The from November to June. However, the absence highest frequency of empty midguts and of early instars at the end of the snow-free sea- hindguts was observed during the onset of son in mid-October and again in early July emergence between late August and early when nymphs had attained maximum size September. The predominance of chironomid (Fig. 2) suggested a univoltine cycle. We assume larvae as the prey of I. petersoni is consistent the eggs hatched over a short period and that with observations for other Isoperla (Frison the majority of nymphal growth occurred over 1935, Minshall and Minshall 1966, Richardson winter, presumably under an insulative snow and Gaufin 1971, Fuller and Stewart 1977, 1979, cover. Our nymph collections late in the Sandberg and Szczytko 1997). growth cycle did not include a wide size Vibrational communication signals were range; however, the Alaskan population (Stew- recorded from 8 males and 6 females from 3 to art et al. 1990) did have a wide range in 13 September 1999. Duets consisted of simple monthly size of nymphs from fall through monophasic male calls (5–19 beats) and female spring, suggesting a flexible life cycle that responses (1–14 beats). Females varied their ranged from univoltine-slow to semivoltine. responses from nonoverlapped (Fig. 3A) to The latter was represented by a slow, winter- overlapped (Fig. 3B); when overlapped, they growing portion of the cohort that was recruited began answering after the 7th male call beat in late summer, or possibly by an egg diapause. (Fig. 3C). We analyzed a total of 228 signals; Sexual size dimorphism of 127 nymphs was 134 were overlapped, 55 nonoverlapped, and observed throughout the sampling period, 39 were lone male calls (Table 2). Male calls with average IOD for males and females of had 11 mode beats (x– = 11.1 ± 2.09) with 1.11 ± 0.06 and 1.36 ± 0.09, respectively. Max- average beat intervals of 118 ± 9 ms. The imum average size was attained first by male mean number of call beats increased slightly nymphs (1.16 ± 0.04 mm) on 8 August (N = 7), for overlapped duets (11.6 ± 1.32) and de- followed by female nymphs (1.40 ± 0.09 mm, creased slightly to 11.0 ± 1.48 for nonover- N = 15) on 22 August (Fig. 2). Mean IOD for lapped duets. The average number of answer 2001] COLORADO STONEFLY POPULATION 449

TABLE 1. Taxa found in the midgut and hindgut of 124 Isoperla petersoni nymphs collected July to September 1998 from a high-altitude stream in Colorado. Sample date (1998): 07/26 08/08 08/16 08/22 08/29 09/06 09/19 % of total Number of organisms Number dissected : 11 19 16 33 22 13 10 124 Plecoptera 1 2 0.63 Zapada sp. 1 1 0.42 Baetidae 1 0.21 Chironomidae 5 1 1 1 1.67 Orthocladiinae 10 2 2 1 3.13 Cricotopus/Orthocladius sp. 56 17 49 86 14 1 46.55 Diplocladius sp. 1 0.21 Paraorthocladius sp. 6 1 2 7 8 5.01 Parametriocnemus sp. 1 0.21 Rheocricotopus sp. 1 0.21 Tanytarsini 1 2 0.63 Diamesinae 9 2 2 3 1 3.55 Diamesa sp. 12 53 34 30 10 29.02 Psuedodiamesa sp. 1 1 1 2 1.04 Pagastia sp. 3 0.63 Prosimulium sp. 3 1 1 1.04 Ostracoda 14 7 2 2 2 5.64 Hydrachnidia 1 0.21 TOTAL 117 85 95 137 42 0 3 100% Number of midguts/hindguts containing Chironomidae parts 4/4 1/6 6/6 11/9 4/4 1/0 Unidentified insect parts 1/0 5/3 1/0 3/4 1/1 Unidentified insect eggs 3/0 0/1 1/0 Unidentifiable material 3/10 2/13 1/12 4/16 4/11 3/4 3/2 Sand 1/1 2/0 1/3 0/3 Empty 1/2 1/0 6/9 2/4 10/9 6/8

TABLE 2. Drumming signal statistical characteristics of Isoperla petersoni recorded from 8 males and 6 females between 3 and 13 September 1999 at 20–24°C. Total calls, answers, duets and intervals Nonoverlapped duets Overlapped duets Number of individuals ______8 6 5 6 8 6 BEATS Number 228 189 55 55 134 134 Mean ± s 11.1 ± 2.09 6.2 ± 2.99 11.0 ± 1.48 5.3 ± 2.57 11.6 ± 1.32 6.6 ± 3.08 Mode 11 4 11 4 11 4 Range 5–19 1–14 8–15 1–11 9–16 2–14

DUET DURATION (MS) Mean ± s 1458 ± 163 (N = 189) 1483 ± 170 1448 ± 159 Mode 1490 1490 1610 Range 1070–1930 1080–1930 1070–1800

BEAT INTERVALS (MS) Mean ± s 118 ± 9 58 ± 22 118 ± 9 59 ± 27 117 ± 8 58 ± 20 (N = 2307) (N = 1178) (N = 551) (N = 292) (N = 1417) (N = 886) Mode 120 50 110 50 110 50 Range 90–160 10–290 100–160 10–230 9–160 10–290 450 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Fig. 3. Duets of Isoperla peterson: A, nonoverlapped male 11-beat, female 4-beat, monophasic exchange, duet duration = 1440 ms; B, overlapped male 13-beat, female 7-beat exchange, duet duration = 1520 ms; C, 4× horizontal, 2× vertical enlargement of duet 3B, segment duration = 396 ms. Numbers above signal indicate female beat number and those below indicate male. beats increased from 5.3 ± 2.56 when duets (Stewart and Maketon 1991) in being mono- were nonoverlapped to 6.6 ± 3.08 when over- phasic and involving 2-way vibrational com- lapped. The rhythm of male calls was uneven munication with or without overlap of calls- and had gradual interval transitions, which answers. The drumming evolution paradigm began long (131 ± 9 ms), then decreased to of Stewart (2001) indicates that all of these 113 ± 5 ms for beats 7 and 8, and then Isoperla have achieved specificity from ances- returned up to 130 ms for the final intervals. tral Plecopteran signals by slight modification Isoperla petersoni belongs to the I. sordida of numbers of beats and beat intervals (rhythm). Banks species complex (Szczytko and Stewart Isoperla petersoni average call beat intervals are 1979b, 1984, Bottorff et al. 1990a) that includes about median to those of the 24.5 to 356.8 ms 8 western Nearctic species. Drumming is known range of other known western Isoperla species for only 3 of them (Bottorff et al. 1990b): I. signals. Only a few Isoperla species have de- adunca Jewett, I. bifurcata Szczytko and Stew- rived, and more complex, grouped or phased art, and I. miwoc Bottorff and Szczytko. Isop- calls (Stewart and Maketon 1991). erla petersoni male calls had nearly twice as many mode and average number of beats as other species within the complex. The average LITERATURE CITED male call interval was closest to I. adunca (138 BAUMANN, R.W., A.R. GAUFIN, AND R.F. SURDICK. 1977. ms). Isoperla petersoni male and female sig- The stoneflies (Plecoptera) of the Rocky Mountains. nals generally fit the ancestral Isoperla pattern Memoirs of the American Entomological Society 31. 2001] COLORADO STONEFLY POPULATION 451

BOTTORFF, R.L., S.W. SZCZYTKO, AND A.W. KNIGHT. 1990a. SHELDON, A.L. 1972. Comparative ecology of Arcynopteryx Descriptions of a new species and three incom- and Diura (Plecoptera) in a California stream. Archiv pletely known species of western Nearctic Isoperla für Hydrobiologie 69:521–546. (Plecoptera: Perlodidae). Proceedings of the Ento- SNELLEN, R.K., AND K.W. STEWART. 1979. The life cycle of mological Society of Washington 92:286–303. Perlesta placida (Plecoptera: Perlidae) in an intermit- BOTTORFF, R.L., S.W. SZCZYTKO, A.W. KNIGHT, AND J.J. tent stream in northern Texas. Annals of the Ento- DIMICK. 1990b. Drumming behavior of four western mological Society of America 72:659–666. Nearctic Isoperla species (Plecoptera: Perlodidae). STARK, B.P., B.R. OBLAD, AND A.R. GAUFIN. 1973. An anno- Annals of the Entomological Society of America 83: tated list of the stoneflies (Plecoptera) of Colorado 991–997. Part I. Entomological News 84:269–277. CATHER, M.R., AND A.R. GAUFIN. 1975. Life history and STEWART, K.W. 2001. Vibrational communication (drum- ecology of Megarcys signata (Plecoptera: Perlodidae), ming) and mate-searching behavior of stoneflies Mill Creek, Wasatch Mountains, Utah. Great Basin (Plecoptera); evolutionary implications. Proceedings Naturalist 35:39–48. of the XIII International Symposium on Plecoptera. DEWALT, R.E., AND K.W. STEWART. 1995. Life histories of In press. stoneflies (Plecoptera) in the Rio Conejos of southern STEWART, K.W., AND M. MAKETON. 1991. Structures used Colorado. Great Basin Naturalist 55:1–18. by Nearctic stoneflies (Plecoptera) for drumming FRISON, T.H. 1935. The stoneflies, or Plecoptera, of Illi- and their relationship to behavioral pattern diversity. nois. Bulletin of the Illinois Natural History Survey Aquatic Insects 13:33–53. 20:281–471. STEWART, K.W., AND B.P. STARK. 1988. Nymphs of North FULLER, R.L., AND K.W. STEWART. 1977. The food habits American stonefly genera (Plecoptera). Entomological of stoneflies (Plecoptera) in the Upper Gunnison Society of America, Thomas Say Foundation 12:1–460. River, Colorado. Environmental Entomology 6: STEWART, K.W., AND D.D. ZEIGLER. 1984. Drumming be- 293–302. havior of twelve North American stonefly (Plecop- ______. 1979. Stonefly (Plecoptera) food habits and prey tera) species: first descriptions in Peltoperlidae, Tae- preferences in the Delores River, Colorado. American niopterygidae and Chloroperlidae. Aquatic Insects Midland Naturalist 101:170–181. 6:49–61. HYNES, H.B.N. 1976. Biology of Plecoptera. Annual Re- STEWART, K,W., R.L. HASSAGE, S.J. HOLDER, AND M.W. views in Entomology 21:135–153. OSWOOD. 1990. Life cycles of six stonefly species ILLIES, J. 1966. Katalog der rezenten Plecoptera. Das (Plecoptera) in subarctic and arctic Alaska streams. Tierreich, 82. Walter de Gruyter, Berlin. Annals Entomological Society of America 83:207–214. JOP, K., AND S.W. SZCZYTKO. 1984. Life cycle and produc- STEWART, K.W., S.W. SZCZYTKO, AND M. MAKETON. 1988. tion of Isoperla signata (Banks) in a central Wiscon- Drumming as a behavioral line of evidence for delin- sin stream. Aquatic Insects 6:81–100. eating species in the genera Isoperla, Pteronarcys, KHOO, S.G. 1968. Experimental studies on diapause in and Taeniopteryx (Plecoptera). Annals of the Ento- stoneflies. I. Nymphs of Capnia bifrons (Newman). mological Society of America 81:689–699. Proceedings of the Royal Entomological Society of SZCZYTKO, S.W., AND K.W. STEWART. 1984. Descriptions of London (A) 43:40–48. Calliperla Banks, Rickera Jewett, and two new west- MAKETON, M., AND K.W. STEWART. 1984. Drumming behav- ern Nearctic Isoperla species (Plecoptera: Perlodi- ior in four North American Perlodidae (Plecoptera) dae). Annals of the Entomological Society of Amer- species. Annals of the Entomological Society of ica 77:251–263. America 77:621–626. ______. 1979a. The genus Isoperla (Plecoptera) of western MINSHALL, G.W., AND J.N. MINSHALL. 1966. Notes on the North America; holomorphology and systematics, life history and ecology of Isoperla clio (Newman) and a new stonefly genus Cascadoperla. Memoirs of and Isogenus decisus Walker (Plecoptera: Perlodi- the American Entomological Society 32. 120 pp. dae). American Midland Naturalist 76: 340–350. ______. 1979b. Drumming behavior of four western Nearc- ORBERNDORFER, R.Y., AND K.W. STEWART. 1977. The life tic Isoperla (Plecoptera) species. Annals of the Ento- cycle of Hydroperla crosbyi (Plecoptera: Perlodidae). mological Society of America 72:781–786. Great Basin Naturalist 37:260–273. ZWICK, P. 1973. Insecta: Plecoptera. Phylogenetisches Sys- RICHARDSON, J.W., AND A.R. GAUFIN. 1971. Food habits of tem und Katalog. Das Tierreich 94. Walter de Gruyter, some western stonefly nymphs. Transactions of the Berlin. American Entomological Society 97:91–121. SANDBERG, J.B., AND S.W. SZCZYTKO. 1997. Life cycle of Received 31 January 2000 Isoperla lata (Plecoptera: Perlodidae) in a central Accepted 31 May 2000 Wisconsin trout stream. Great Lakes Entomologist 30(4):143–159. SCHWARZ, P. 1970. Autokologische Untersuchungen zum Lubenszyklus von Setipalpia-Arten (Plecoptera). Archiv für Hydrobiologie 67:103–140. Western North American Naturalist 61(4), © 2001, pp. 452–462

PHYSIOLOGICAL, MORPHOLOGICAL, AND ENVIRONMENTAL VARIATION AMONG GEOGRAPHICALLY ISOLATED COTTONWOOD (POPULUS DELTOIDES) POPULATIONS IN NEW MEXICO

Diane L. Rowland1, Lucille Beals2, Amina A. Chaudhry2, Ann S. Evans2, and Larry S. Grodeska2

ABSTRACT.—The ability of a plant population to respond and eventually adapt to environmental stress ultimately determines that population’s survival. This becomes especially significant in environments where important plant resource levels have radically decreased. Southwestern riparian areas have numerous plant species that are experiencing radical changes in water availability due to construction of dams, and thus their ability to respond to such changes is critical. One such species likely to be greatly affected by these hydrological changes is Populus deltoides var. wislizenii (cottonwood) because it relies heavily on both groundwater and river surface volume as primary water sources. Both water sources have been extremely impacted by impoundments along southwestern rivers. To understand how New Mexico populations of cottonwood may respond to environmental changes, we quantified environmental differences and characterized physiological and morphological variation among 4 cottonwood populations. Significant differences among study sites in water availability were indicated by both soil and groundwater salinity. The northernmost site, at Abiquiu, had the highest salinity levels in both soil and groundwater, followed by Bernardo, while San Antonio and Corrales sites had the lowest soil salinity. As expected, variation in physiological and leaf morphological characters existed among and within the tree populations, most likely in response to environmental factors. Midday xylem pressure potentials indicated that Abiquiu individuals suffered the greatest water stress and they also had the highest transpiration levels. Because of high specific leaf weights and high photosynthetic levels, cottonwoods at Corrales may better mitigate lower water availability. Such physiological and morphological trait variability among populations is ecologically important and may be of use in present reclamation and conservation efforts in these areas.

Key words: Rio Grande, Populus deltoides, ecophysiology, soil salinity, groundwater, cottonwood.

Studies in plant physiological ecology strive Mexico is an ideal model for studying physio- to provide an understanding of the diversity of logical plant response to environmental change. plant physiological functions and how such In the last 80 years, this ecosystem has experi- diversity allows plants to interact favorably with enced dramatic hydrologic perturbations that, their changing environment (Jacquard and in turn, may have induced physiological and Urbanska 1988). Many of these environmental morphological variation among these riparian changes eventually impose selection that acts on populations. Construction of dams and exten- plants functioning as physiological units (Chapin sive channelization of this river system have and Oechel 1983). The end result of selection increased salinization of soil and groundwater, varies depending upon the nature of environ- eradicated most flood events, and decreased mental change and the concomitant plant re- water availability (Crawford et al. 1993). With sponse. Variation among populations in their the control of flooding and the construction of abilities to respond to environmental change is a system of drains paralleling the river through important in understanding the result of ecosys- most of the middle Rio Grande basin in the tem perturbations and in our ability to predict 1930s, the water table was lowered by 1.5 m a population’s ability to adapt and survive. and the periodic, evenly distributed recharge Therefore, it is ecologically and evolutionarily of groundwater by flood events was virtually important to study the amount and direction of eliminated (Crawford et al. 1993). Even now, physiological variation in natural plant popula- the storm-water conveyance system along a tions that experience environmental changes. large portion of the Rio Grande surrounding The riparian ecosystem along the middle Albuquerque, New Mexico, delivers most of Rio Grande and its major tributaries in New the water runoff directly to the river channel,

1USDA-ARS, National Peanut Research Lab, 1011 Forester Dr. SE, Dawson, GA 31742. 2Department of Biology, University of New Mexico, Albuquerque, NM 87131.

452 2001] VARIATION AMONG COTTONWOOD POPULATIONS 453 thereby effectively bypassing the recharge of morphological characters on trees at these groundwater (Crawford et al. 1993). In addition, sites. We chose physiological traits that can be the low mean annual rainfall and high summer indicators of water stress including photosyn- temperatures of the desert Southwest likely thesis, water vapor exchange, and water exacerbate these hydrologic impacts. All 3 likely potential (Buxton et al. 1985, Baldocchi et al. sources of water in these riparian ecosystems, 1987, Iacobelli and McCaughey 1993). We i.e., surface rainfall, channel flow, and ground- also examined several morphological traits water stores, can be highly variable and are that are indicators of changes in water avail- often relatively scarce. The effects of low pre- ability, i.e., traits that significantly affect gas cipitation and flow are exaggerated by the exchange characteristics including leaf area, sandy alluvium in the middle Rio Grande val- specific leaf weight, and chlorophyll content. ley, which cannot provide a suitable substrate for groundwater storage (Crawford et al. 1993). METHODS One might expect, therefore, that geographi- Study Sites cally separated plant populations within this system are likely to experience differing degrees Four study sites were established to sample of water availability. If so, separation by dis- the relatively contiguous cottonwood forest tance combined with changes in the hydrolog- along the middle Rio Grande basin of New ical environment may be driving differential Mexico. In New Mexico cottonwood forests physiological and morphological plant re- are found from just below Taos Gorge south to sponses to lowered water availability in New Elephant Butte Reservoir. Populations south Mexico riparian ecosystems. of the reservoir have been reduced to isolated One species likely to be acutely affected by patches due to a variety of anthropogenic these hydrologic changes is New Mexico’s impacts. The study sites are located along an native cottonwood, Populus deltoides var. wis- approximately 280-km expanse of the Rio lizenii. Although primarily phreatophytic, P. Grande watershed in New Mexico (Fig. 1). deltoides relies secondarily on precipitation The northernmost site is Abiquiu (36°12′30″N, and stream flow for its water resources (Lef- 106°19′06″W, elevation 1807 m) on the Rio fler and Evans 1999). Dams have not only Chama (a major tributary of the Rio Grande). impacted stream flow and groundwater levels, Along the Rio Grande there are 3 additional but they have also virtually eliminated germi- study sites: 1 each at Corrales (35°14′16″N, nation sites in most P. deltoides populations 106°36′22″W, elevation 1552 m), Bernardo along the Rio Grande. It is therefore expected (34°25′06″N, 106°50′06″W, elevation 1444 m), that these populations will be replaced by and San Antonio (33°55′06″N, 106°52′06″W, exotic plant species within the next 50 years elevation 1380 m). (Howe and Knopf 1991). This is not an uncom- For a previous study, ten 10-m-wide ran- mon phenomenon; in parts of western Canada, domly located study plots were established cottonwood forests downstream from dams within an approximately 0.5-km section of for- have virtually been eliminated due to imposed est at each study site (Rowland et al. 2000). water stress (Rood and Heinze-Milne 1989, Each plot is perpendicular to the river and Rood and Mahoney 1990). Therefore, studies extends from the river’s edge to the end of the that examine the physiological-morphological riparian forest. At 3 of the sites, the forest edge responses of riparian tree populations to im- coincides with a levee road and water diver- posed changes in water regime provide new sion channel that both parallel the river. At insights and predictions about how dams con- Abiquiu the river and an abandoned agricul- tribute to the decline of downstream ecosys- tural field border the forest. Thirty trees at tems (Tyree et al. 1994). each site were randomly chosen for measure- To gauge the degree of variation among P. ments. Cottonwood stands sampled tended to deltoides cottonwood populations in New be continuously mature throughout each site, Mexico in both environmental conditions and and tree age averaged 30–40 years (unpub- possible physiological responses to differences lished data). However, minor recruitment was in hydrology, we quantified variability among observed at Corrales and San Antonio. At these 4 study sites in both soil and groundwater latter sites the gallery forest ended abruptly at conditions and examined physiological and a 50-m (Corrales) or 200-m (San Antonio) stretch 454 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Following procedures outlined by Black et al. (1965), we prepared saturation paste extracts and analyzed them for salinity by the electrical conductivity (EC) method using an Accumet model 50 pH/ion/conductivity meter. We determined total nitrogen and phosphorus on 10 random samples per site using the Kjeldahl digestion method (Black et al. 1965). Soil texture from 8 randomly chosen samples per site was analyzed by the hydrometer method (Black et al. 1965). After percentages of sand, silt, and clay were calculated, we used a Soil Survey Staff chart to apply textural classifications (e.g., clay, loam, sandy loam). GROUNDWATER SITE CHARACTERISTICS.— After taking physiological and morphological measurements, we established 3 groundwater wells at each site in a stratified random manner in December 1995. At each site 3 plots were chosen randomly and single wells were Fig. 1. Study sites within the state of New Mexico. Study installed 20 m from the edge of the river, 20 m populations of Populus deltoides are located at Abiquiu along the Rio Chama, and at Corrales, Bernardo, and San from the edge of the forest near the levee road Antonio along the Rio Grande. (or near the agricultural field at Abiquiu), and at the center of the plot. All wells were hand- augered to groundwater. Three-inch (7.62-cm) of floodplain inhabited by younger trees be- PVC pipe was used for the well body, and tween the forest edge and the river. These approximately 1 m of screened PVC tube was trees were sampled at Corrales and averaged inserted into each well at groundwater. Mea- 10–15 years of age, but trees at San Antonio surements were taken monthly during the were not measured because they were too cottonwood growing season from May 1996 small to tag without causing damage and pos- until December 1996. Depth to groundwater sible mortality. (m) from the soil surface was measured using a Measurements meter tape to the nearest cm. Dissolved oxygen was measured with a YSI dissolved oxygen SOIL SITE CHARACTERISTICS.—Soil, physiological, and morphological measurements were meter (Yellow Springs Instrument Co., Inc.), made during the same sampling period, June and water samples were taken for electrical and July 1995. At each of the 4 sites, we ran- conductivity measurements (samples were ° domly chose 30 trees for soil and physiological stored in a refrigerator at 4 C until conductiv- measurements. Soil samples were taken from ity measurements were made). Conductivity, a the north side of each tree at either 2/3 of the measure of salinity, was determined with an canopy width from the trunk or 1.6 m from the Accumet model 50 pH/ion/conductivity meter. trunk if the canopy was not even; this distance MORPHOLOGICAL AND PHYSIOLOGICAL MEA- ensured that samples were taken from the SUREMENTS.—In mid-June 1995 we took mor- active root zone. Samples were taken with a 2- phological and physiological measurements at inch (5.08-cm) PVC pipe sharpened at one end. the southernmost population (San Antonio). After removing overlying organic material, the Because of seasonal changes in flower produc- PVC pipe was driven into the ground to a tion, the delay in leaf development between depth of 20 cm and withdrawn. We then re- successively more northern sites was approxi- moved core samples, placed them in plastic mately 1 week. To compensate for this sea- bags, and refrigerated them at 4°C until pro- sonal difference, each successive northern site cessed. Soils were air-dried and homogenized was sampled 1 week later than the previous using a #10 sieve (2.00-mm sieve opening) site. At each of the 4 sites, the same 30 indi- before analysis. viduals chosen randomly for soil samples were 2001] VARIATION AMONG COTTONWOOD POPULATIONS 455 used for gas exchange, morphological, and chlorophyll (µg ⋅ cm–2) = (SPAD – 8.1095)/0.855 water potential measurements. To minimize light-induced developmental differences, we developed for Populus deltoides trees in New collected 3 shade leaves near the tree bole Mexico (J. Leffler personal communication). from the lower canopy of each tree. Branch After field measurements of gas exchange apices were excised with an extendable pole- and chlorophyll, leaves were placed in plastic cutter and leaf samples analyzed immediately bags and kept on ice or refrigerated at 4°C for gas exchange. Although measuring gas ex- until leaf area and stomatal density measure- change, especially photosynthesis, on excised ments were completed. Hence, leaf morpho- tissue may be problematic (Slavik 1974, Lakso logical characters were measured on the same 1982), others have found no appreciable effects leaves measured for gas exchange. Leaf area was (Barden et al. 1980). In this study, measurement measured to the nearest square centimeter of intact leaf tissue was not possible due to with a model CI-201 leaf area meter (CID, inaccessibility of the canopy. Therefore, mea- Inc.). For each leaf, 3 area measurements were surements were standardized across all sites taken and averaged. Leaves were then dried ° by equalizing time between excision and gas in a drying oven at 60 C for 5 days and, after exchange analysis. One fully expanded leaf per drying was complete, leaf mass (g) was deter- cut branch was selected for measurement; mined using an analytical balance. Specific selected leaves had blade lengths between 6 leaf weight (SLW) was calculated for each leaf and 7 cm, which standardized leaf develop- (i.e., mean leaf area/leaf mass). mental stage based on the plastochron index Leaf xylem pressure potentials were mea- for Populus (Dickmann 1971, Larson and sured in the field with a pressure bomb (PMS Isebrands 1971, Isebrands and Larson 1973, Instrument Company, Corvallis, OR). Using a Lamoreaux et al. 1978). Gas exchange mea- polecutter, we excised 1 branch from each of the same 30 trees measured for gas exchange. surements were made in full sun. Photosyn- Branches were placed in plastic bags and mea- thesis, stomatal conductance, and transpira- sured within 15 minutes after excision in a tion were measured with an ADC infrared gas Scholander-style pressure chamber (Scholan- analyzer (The Analytical Development Com- der et al. 1965). Water potential measurements pany Ltd., England; LCA model 3). Water-use were taken between 1200 and 1500 hours MDT, efficiency (WUE) was calculated as the ratio of the hours of greatest water loss. Predawn mea- photosynthesis:transpiration. To ensure that gas surements were not taken because previous exchange was measured at maximum levels, results at the Corrales site indicated no signifi- all such measurements were taken between cant differences among individuals for predawn 0900 and 1130 hours MDT at all sites. This water potentials; therefore, we concluded that time interval was previously determined to be midday water potentials would provide a rela- within the period of peak gas exchange for P. tive index of water stress among sites. deltoides at the New Mexico sites (unpub- At each site we measured stomatal density lished results). on 10 trees randomly selected from the same Immediately after the gas exchange mea- 30 trees measured for physiology. One branch surements, chlorophyll content was estimated per tree was collected and frozen at 0°C. One with a Minolta SPAD chlorophyll meter leaf was removed from each branch for stom- (Minolta Corp., Ramsey, NJ). The SPAD chloro- atal density measurement. For each leaf, we phyll meter measures absorbance by plant tis- cut 2 samples from the center of the leaf blade, sues of a particular range of wavelengths in one on either side of the midrib, and mounted the visible spectrum; this is a relative measure them on glass microscope slides with distilled of the internal concentration of chlorophyll a water. Because Populus is amphistomatous, and b. Three SPAD measurements were taken stomatal density was counted on both the adax- per leaf and then averaged to correct for possi- ial and abaxial surfaces of each leaf sample. ble non-homogeneous distribution of chloro- Samples were observed and stomata counted phyll throughout the leaf (Monje and Bugbee with a Zeiss Axioskop light microscope at 20× 1992). SPAD measurements were then con- magnification. For each mounted sample, stom- verted to chlorophyll content using the equa- ata were counted in 4 arbitrary fields (right tion: adaxial, right abaxial, left adaxial, left abaxial). 456 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Statistical Analyses RESULTS For environmental variation among sites, 1- Environmental Variation way fixed model ANOVAs (SAS Institute 1989), Significant variation was evident among the with SITE as a fixed factor, were used to assess 4 sites for all soil characteristics (Table 1). Soil possible differences in salinity, total nitrogen, salinity showed a wide range among sites. total phosphorus, ratio of nitrogen to phospho- Abiquiu had the highest mean salinity, fol- rus, and soil texture among sites. Repeated- lowed by Bernardo; whereas San Antonio and measures ANOVAs (SAS Institute 1989) were Corrales had similarly low soil salinity. Total used to assess differences in groundwater depth, nitrogen and soil phosphorus content was dissolved oxygen, and salinity among sites. highest at Abiquiu and lowest at Corrales Physiological and morphological variation both (Table 1). The nitrogen-to-phosphorus ratio among and within sites was examined with was highest at Bernardo and lowest at Cor- individual mixed-model nested univariate rales, while San Antonio had an intermediate ANOVAs with SITE as a fixed factor and value. Soil textures ranged widely among sites TAG(SITE) or tree nested within site as a ran- as well (Table 1). Analysis of variance showed dom factor (Sokal and Rohlf 1995). Scheffe’s significant differences among sites for both multiple range tests were used to determine sand and clay. Mean percentages for sand where differences existed among sites. Pear- ranged from 11.8 at San Antonio to 57.2 at son correlations (SYSTAT 1996) were used to Corrales; for clay, mean percentages ranged relate the environmental parameters to physi- from 19.9 at Corrales to 56.3 at San Antonio. ological and morphological characteristics with- However, silt content did not differ among in a given site. Within each of the 4 sites, we sites; mean percentages ranged from 22.9 at correlated soil salinity and distance to the river Corrales to 36.5 at Abiquiu. Corrales samples of a measured tree with its average measured were extremely sandy (57%), San Antonio was physiological characters of photosynthesis, mostly clay (57%), and Bernardo soil was 43% stomatal conductance, transpiration, water-use clay. Abiquiu soils were more equally parti- efficiency, chlorophyll content, and water tioned among 3 soil particle classes (31% clay, potential. 36% silt, 32% sand), i.e., a texture of loam.

TABLE 1. ANOVA results and means for variation in soil characteristics among 4 populations (AB = Abiquiu, CO = Corrales, BE = Bernardo, SA = San Antonio). Values for soil texture are in percentage of each soil type. Means reported with standard error in parentheses; different letters denote significant differences among populations at a 0.05 significance level (Scheffe grouping).

______Site mean Trait AB CO BE SA Salinity (dSiemens ⋅ m–1) (df = 3, MS = 10543755, F = 86.7, P-value = 0.0001 1.33 (0.11) a 0.10 (0.02) c 0.92 (0.05) b 0.18 (0.02) c Nitrogen (µg ⋅ g–1) (df = 3, MS = 1243372, F = 7.7, P-value = 0.0004) 1132.9 (142.7) a 331.4 (138.8) b 1010.3 (141.8) a 808.3 (71.0) ab Phosphorus (µg ⋅ g–1) (df = 3, MS =38448.1, F = 7.8, P-value = 0.0004) 350.0 (22.8) a 208.9 (29.1) b 292.4 (18.2) ab 328.1 (16.5) a Nitrogen / phosphorus ratio (df = 3, MS = 9.2584, F = 10.5, P-value = 0.0001) 3.2 (0.3) a 1.2 (0.3) b 3.3 (0.3) a 2.5 (0.2) ab Sand (df = 3, MS = 2574, F = 5.93, P-value = 0.0034) 32.3 (5.2) ab 57.2 (8.2) a 33.0 (9.5) ab 11.8 (7.5) b Clay (df = 3, MS = 1816, F = 8.44, P-value = 0.0005) 31.2 (2.3) b 19.9 (1.5) b 42.6 (8.1) ab 56.3 (6.5) a Silt (df = 3, MS = 291, F = 1.77, P-value = 0.18) 36.5 (3.1) a 22.9 (6.9) a 24.4 (3.8) a 31.9 (4.4) a 2001] VARIATION AMONG COTTONWOOD POPULATIONS 457

Fig. 2. Groundwater measurements for 3 New Mexico Populus deltoides populations during the 1996 growing season (A = Abiquiu, B = Bernardo, C = Corrales). Depth to groundwater (2a) is the depth of the groundwater below the soil surface; conductivity (2b) is a measure of salinity; and dissolved oxygen content (2c) is the percentage of oxygen present in groundwater samples.

Groundwater measures varied significantly dissolved oxygen content than Bernardo and among sites. Distance to groundwater (df = 2, Corrales for most months during the measure- SS = 11.317, MS = 5.658, F = 12.65, P < ment period (Figs. 2b, 2c). Bernardo ground- 0.0186), conductivity (df = 2, SS = 1054096, water was higher in dissolved oxygen than the MS = 527048, F = 394.82, P < 0.0002), and other 2 sites for every month except June 1996 dissolved oxygen content (df = 2, SS = 1.39, (Fig. 2c). MS = 0.69, F = 15.84, P < 0.0126) were all Physiological and Leaf significantly different among sites. Depth to Morphological Variation groundwater was greatest at Corrales and least at Abiquiu throughout the sampling period Individual nested ANOVAs, using a signifi- (May–December; Fig. 2a). However, Abiquiu cance level of P < 0.0125, adjusted for multiple groundwater was highly saline and lower in tests (Sokal and Rohlf 1995), showed signifi- 458 WESTERN NORTH AMERICAN NATURALIST [Volume 61 cant variation among and within populations significantly different between Corrales and for specific physiological and morphological Bernardo. Leaf area also showed significant characters. Photosynthesis was significantly variation among and within populations. Leaf different among and within populations and area was largest at the 2 southernmost popula- ranged from 7.2 µmol C ⋅ m–2 s at San Antonio tions, Bernardo (31.7 cm2) and San Antonio to 12.2 µmol C ⋅ m–2 s at Corrales (Table 2). (29.5 cm2), while leaves were smallest at Cor- However, there was no clear geographical (i.e., rales (20.1 cm2; Table 4). Specific leaf weight north–south) pattern for photosynthetic levels, (SLW) differed among and within sites; it was with the northernmost site (Abiquiu) not sig- highest at Corrales (100.0 g ⋅ m–2) and lowest nificantly different from the southernmost site at San Antonio (62.9 g ⋅ m–2; Table 4). Although (San Antonio). Nonetheless, stomatal conduc- leaves were smaller at Corrales, they were tance, transpiration, and midday xylem pres- thicker, thus contributing to high SLW. Adaxial sure potentials did show geographical trends. (upper) stomatal density did not significantly Stomatal conductance was significantly differ- differ among sites; however, abaxial (lower) ent among and within populations (Table 2) density approached significance (Table 4). and decreased with each site from north to Abiquiu had the lowest mean values for both south; Abiquiu had the highest stomatal con- adaxial stomatal density (67.2 ⋅ mm–2) and ductance, while San Antonio had the lowest abaxial (53.4 ⋅ mm–2). A paired t test across value. Transpiration also showed significant populations showed that the adaxial surface differences among and within populations and had a significantly higher stomatal density a similar decreasing trend to the south (Table than the abaxial surface (df = 39, tcrit = 2.02, 2); however, the 2 middle geographic popula- P < 0.0001). tions (Corrales and Bernardo) were not significantly different from one another. Midday DISCUSSION xylem pressure potentials were significantly different among populations and showed a Our study provides evidence of consider- geographic trend of increasing xylem pressure able variation in environmental, physiological, potentials to the south (Table 2). The greatest and morphological characteristics within and water stress was noted at Abiquiu (–1.9 MPa), among natural riparian populations of Populus the lowest at San Antonio and Bernardo (–1.7 deltoides in the New Mexico Rio Grande basin. MPa). Although water-use efficiency showed While many studies have addressed physio- no strong geographic trend, differences among logical variation among and within Populus and within populations were significant. All species, most were conducted on limited, populations were significantly different from clonal material (e.g., Ceulemans and Impens one another, with the highest WUE occurring 1980, 1983, 1987, Blake et al. 1984, Liu and ⋅ –1 at Corrales (1.8 mmol C mol H2O ) and the Dickmann 1992, Dunlap et al. 1993, Donahue ⋅ –1 lowest at Abiquiu (0.8 mmol C mol H2O ; et al. 1994). Those studies comparing the degree Table 2). Water-use efficiency was also signifi- and type of physiological differences among cantly negatively correlated with soil salinity natural Populus populations of contrasting en- at Abiquiu, the site with highest soil salinity vironments, as in our study, are quite scarce (Table 3), but it was not correlated with dis- (see McGee et al. 1981). Our study is ecologi- tance to the river. No other physiological or cally important for detecting and describing morphological variables were significantly the partitioning of physiological and morpho- correlated with these 2 environmental para- logical variation among populations. These meters within individual sites. descriptions assist us in predicting future pop- Leaf morphological traits were also signifi- ulation survival because variability among cantly different among and, in most cases, with- populations in physiological and morphologi- in populations (Table 4). Mean chlorophyll cal characteristics will eventually impact their content (converted values, µg ⋅ cm–2) showed ability to respond to these environmental dif- significant differences among and within pop- ferences. In addition, studies that examine ulations and ranged from 28.1 at San Antonio variation within natural populations are a to 23.6 at Abiquiu. Chlorophyll levels showed mainstay of silviculture research for screening a geographic trend of increasing to the south; and identifying new candidate genetic stock like transpiration, chlorophyll levels were not (Bassman and Zwier 1991). Because Populus is 2001] VARIATION AMONG COTTONWOOD POPULATIONS 459

TABLE 2. Variation in physiological characteristics among 4 populations (AB = Abiquiu, CO = Corrales, BE = Bernardo, SA = San Antonio). Mixed-model nested ANOVAs were performed to determine both among- (SITE) and within-population (among individuals) (TAG[SITE]) differences for all traits except water potential. A fixed-model ANOVA was performed to determine among-population differences in water potential. Means reported with standard error in parentheses; different letters denote significant differences among populations at a 0.05 significance level (Scheffe grouping).

______Source of variation Trait AB CO BE SA Photosynthesis (µmol C ⋅ m–2 s) Site: df = 3, MS = 455.3, F = 42.7, P-value = 0.0001; Tag(Site): df = 116, MS = 10.7, F = 2.3, P-value = 0.0001 7.8 (0.3) c 12.2 (0.3) a 9.0 (0.2) b 7.2 (0.2) c ⋅ –2 Stomatal conductance (mmol H2O m s) Site: df = 3, MS = 468899.6, F = 68.5, P-value =0.0001 Tag(Site): df = 116, MS = 6869.4, F = 3.8, P-value = 0.0001 265.8 (7.8) a 225.0 (6.9) b 165.3 (4.2) c 102.6 (5.0) d ⋅ –2 Transpiration (mmol H2O m s) Site: df = 3, MS = 430.0, F = 123.7, P-value = 0.0001 Tag(Site): df = 116, MS = 3.5, F = 4.0, P-value = 0.0001 9.8 (0.1) a 6.7 (0.1) b 6.8 (0.2) b 4.5 (0.1) c ⋅ –1 Water-use efficiency (mmol C mol H2O ) Site: df = 3, MS = 18.4, F = 62.6, P-value = 0.0001 Tag(Site): df = 116, MS = 0.3, F = 6.2, P-value = 0.0001 0.8 (0.0) d 1.8 (0.0) a 1.4 (0.0) c 1.7 (0.0) b Xylem pressure potential (MPa) Site: df = 3, MS = 30.3, F = 6.7, P-value = 0.0003 –1.9 (0.0) c –1.8 (0.0) bc –1.7 (0.0) ab –1.7 (0.0) a

being studied for energy plantations (Schulte and nitrogen levels at this site; abandoned et al. 1987) and for wood and fiber production agricultural fields can be extremely saline (Ceulemans and Impens 1980, 1987), identify- (Hendrickx et al. 1992). The Bernardo site also ing genotypes in natural populations with had high salt levels, but these may be due to drought-adaptive traits would be extremely increased abundance of salt cedar. Salt cedar beneficial (Bassman and Zwier 1991). We know was much more abundant at Bernardo than at that plants coming from arid environments the other 3 sites; and, as Busch and Smith generally are better adapted for survival under (1995) have shown, salt cedar, because of its low water conditions (Gurevitch et al. 1986); high salt content in leaf litter, tends to salinize therefore, genotypes from these populations soil. Corrales showed a relative deficiency in might be useful for introduction to plantations soil nitrogen and very low soil salinity, both of under arid and low watering regimes. which might be a result of its extremely sandy Water availability, as reflected in soil and soil texture. These environmental differences, groundwater parameters, differed significantly both in soil and groundwater, may be factors among separate, cottonwood-dominated sites contributing to the physiological and morpho- in the Rio Grande basin. At the northern site logical variation we found among these popu- at Abiquiu, there were high levels of salt in lations. both the soil and groundwater, presumably Large variation in many physiological traits reducing the water available to P. deltoides and leaf morphological traits related to water trees at this site. High salinity and nitrogen use was evident among the New Mexico pop- levels at Abiquiu may be due to the extensive ulations and appears to be somewhat influenced agricultural fields that are directly adjacent to by environmental differences among popula- the forest. Leaching and runoff of salts and tions. With high soil and groundwater salinity fertilizer into the cottonwood forest from agri- levels at Abiquiu, we expected cottonwoods cultural fields could cause elevated salinity there to be water stressed. That is precisely 460 WESTERN NORTH AMERICAN NATURALIST [Volume 61

TABLE 3. Pearson correlations between average water- from southeast to northwest in British Colum- ⋅ –1 use efficiency (mmol C mol H2O ) and the environmen- bia, Canada. tal characteristics of soil salinity and distance to river. Correlations are within a single population (Abiquiu, N = Morphological variation among sites suggests 28; Corrales, N = 30; Bernardo, N = 27; San Antonio, N that some populations are better able than = 30). others to respond favorably to water stress Salinity Distance conditions. For example, at Corrales, cotton- Population (dSiemens ⋅ cm–1) to river (m) woods had higher specific leaf weights than Abiquiu –0.523* –0.272 did cottonwoods at other sites, indicating that Corrales 0.146 0.284 Corrales leaves were smaller but thicker (Busch Bernardo 0.088 0.080 and Smith 1995). This type of leaf morphology San Antonio 0.019 –0.005 is adaptive in relatively dry habitats. Leaf mor- *P ≤ 0.05 phology may explain the ability of Corrales trees to maintain higher water-use efficiencies than trees in the other populations. Geber and what we found, with Abiquiu cottonwoods Dawson (1990) provided evidence that small- experiencing higher transpiration levels and leafed populations in certain plants have high lower midday water potentials than cotton- gas exchange rates, low water-use efficiencies, woods at the other sites. In fact, midday water and maximum vegetative yields, all leading to potentials of –1.8 MPa suggest that cavitation a highly cost-efficient photosynthetic system. stress may be high in these trees because cavi- This appears to be the case for cottonwoods in tation is known to occur in Populus at water Corrales, since they have relatively small leaves potentials of –1.7 MPa (Tyree et al. 1994). The and high photosynthetic rates. Other morpho- negative correlation of water-use efficiency logical traits, such as stomatal frequency, can and soil salinity at Abiquiu appears to support be linked to gas exchange characteristics that the conclusion that environment is influencing are positively correlated with photosynthesis cottonwood physiology to some extent. (Paul and Eagles 1988). Although we found no Environmental differences among cotton- significant variation in stomatal density among wood populations related to the geography of studied populations, a trend toward high sto- the Rio Grande basin are indicated by the matal density was noted in Corrales trees, north–south trends in physiological and mor- which may be another factor explaining their phological variation. The 2 middle sites, Ber- high photosynthetic rates. nardo and Corrales, are separated by 60 miles In addition to the variation among popula- and are isolated from the other 2 sites by nat- tions, we also found significant variation with- ural constrictions in the river north of Cor- in populations of P. deltoides for several physi- rales and south of Bernardo. Hence, these 2 ological and morphological traits. This is not middle populations may be more environmentally similar. This is certainly true for our mea- an uncommon finding since individual plants surements of soil nitrogen and nitrogen:phos- within populations can respond differently to phorus ratio. Further, we found that transpira- stress, and recent studies have shown consid- tion, water potential, and chlorophyll content erable microscale environmental variation did not differ between these 2 sites. Nonethe- within populations. What is important about less, there are important environmental para- this intrapopulation variation in physiological meters, such as soil salinity, which did differ response is that it may lead to small-scale between them and may explain tree-response genetic variation within populations (Perry and differences in other physiological traits (e.g., Knowles 1991, Young and Merriam 1994, Loisell water-use efficiency) and morphological traits et al. 1995). Because our study spanned only a (e.g., specific leaf weight). We found a strong single growth season, long-term predictions geographic trend in stomatal conductance that about the continued direction of inter- and was greatest at Abiquiu and decreased to the intrapopulation variation are limited. Long-term south. Geographic trends in physiological traits monitoring of these populations is necessary are not uncommon. For example, Dang et al. as well as determining the genetic basis of the (1994) found that photosynthesis, midday physiological and morphological traits for a water potentials, and transpiration in red alders full understanding of any possible adaptive (Alnus rubra) have a geographic trend increasing responses to environmental stress. 2001] VARIATION AMONG COTTONWOOD POPULATIONS 461

TABLE 4. Variation in leaf morphological characteristics among 4 populations (AB = Abiquiu, CO = Corrales, BE = Bernardo, SA = San Antonio). Mixed-model nested ANOVAs were performed to determine both variation among populations (SITE) and within-population differences (among individuals) (TAG[SITE]) for all traits except stomatal density. A fixed-model ANOVA was performed to determine among-population differences in stomatal density. Means reported with standard error in parentheses; different letters denote significant differences among populations at a 0.05 significance level (Scheffe grouping).

______Source of variation Trait AB CO BE SA Chlorophyll (µg ⋅ cm–2) Site: df = 3, MS = 217.0, F = 6.4, P-value = 0.0005 Tag(Site): df = 116, MS = 34.1, F = 2.9, P-value = 0.0001 23.6 (0.5) c 26.1 (0.6) b 25.9 (0.5) b 28.1 (0.5) a Leaf area (cm2) Site: df = 3, MS = 2379.5, F = 27.2, P-value = 0.0001 Tag(Site): df = 116, MS = 87.8, F = 3.2, P-value = 0.0001 24.9 (0.7) b 20.1 (0.5) c 31.7 (0.9) a 29.5 (0.7) a Specific leaf weight (g ⋅ m–2) Site: df = 3, MS = 22709.7, F = 45.4, P-value = 0.0001 Tag(Site): df = 116, MS = 501.5, F = 6.6, P-value = 0.0001 75.5 (1.6) b 100.0 (2.1) a 71.6 (1.3) c 62.9 (1.1) d Stomatal density (stomates ⋅ mm–2) Adaxial density Site: df = 3, MS = 330.0, F = 2.0, P-value = 0.1287 67.2 (2.9) a 79.8 (2.7) a 77.6 (3.4) a 71.6 (2.9) a Abaxial density Site: df = 3, MS = 445.6, F = 2.6, P-value = 0.0656 53.4 (2.7) a 68.3 (2.5) a 66.0 (2.7) a 65.1 (2.8) a

ACKNOWLEDGMENTS deciduous forest. Journal of Applied Ecology 24: 251–260. The Research Experience for Undergradu- BARDEN, J.A., J.M. LOVE, P.J. PORPIGLIA, R.P. MARINI, AND J.D. CALDWELL. 1980. Photosynthesis and dark res- ates Program provided field and monetary piration of apple leaves are not affected by shoot assistance for this study; LB, AC, and LG were detachment. HortScience 15:595–597. supported by NSF Grant #BIR-9424121. The BASSMAN, J.H., AND J.C. ZWIER. 1991. Gas exchange characteristics of Populus trichocarpa, Populus deltoides Graduate Research Allocations Committee at and Populus trichocarpa × P. deltoides clones. Tree UNM and the LTER Sevilleta provided mone- Physiology 8:145–159. tary support for DR. We thank P. Kelley, J. BLACK, C.A., D.D. EVANS, L.E. ENSMINGER, J.L. WHITE, Leffler, M. Healey, L. LaBong, D. Wheeler, and AND F.E. CLARK. 1965. Methods of soil analysis. American Society of Agronomy Inc., Madison, WI. R. Cabin for field assistance. We especially 1572 pp. thank D.P. Rowland, Dr. D.T. Jennings, and N. BLAKE, T.J., T.J TSCHAPLINSKI, AND A. EASTHAM. 1984. Jennings for field support and design consulta- Stomatal control of water use efficiency in poplar tion. We gratefully thank M. Mason for the use clones and hybrids. Canadian Journal of Botany of the property and trees at Abiquiu. We also 62:1344–1351. BUSCH, D.E., AND S.D. SMITH. 1995. Mechanisms associ- thank Dr. E. Bedrick of UNM and Dr. I. Harris ated with decline of woody species in riparian eco- of Northern Arizona University for statistical systems of the southwestern U.S. Ecological Mono- consultation, and Dr. D. Marshall for use of the graphs 65:347–370. BUXTON, G.F., D.R. CYR, E.B. DUMBROFF, AND D.P. WEBB. Zeiss microscope. Drs. D. Natvig, G. Johnson, 1985. Physiological responses of three northern C. Crawford, M. Lechowicz, N. Johnson, and conifers to rapid and slow induction of moisture D. Jennings provided helpful and constructive stress. Canadian Journal of Botany 63:1171–1176. comments on an earlier draft. CEULEMANS, R., AND I. IMPENS. 1980. Leaf gas exchange processes and related characteristics of seven poplar clones under laboratory conditions. Canadian Jour- LITERATURE CITED nal of Forest Research 10:429–435. _____. 1983. Net CO2 exchange rate and shoot growth of BALDOCCHI, D.D., S.B. VERMA, AND D.E. ANDERSON. 1987. young poplar (Populus) clones. Journal of Experi- Canopy photosynthesis and water-use efficiency in a mental Botany 34:866–870. 462 WESTERN NORTH AMERICAN NATURALIST [Volume 61

_____. 1987. Variations in photosynthetic, anatomical, and LEFFLER, A.J., AND A.S. EVANS. 1999. Variation in carbon enzymatic leaf traits and correlations with growth in isotope composition among years in the riparian tree recently selected Populus hybrids. Canadian Journal Populus fremontii. Oecologia 119:311–319. of Forest Research 17:273–283. LIU, Z., AND D.I. DICKMANN. 1992. Responses of two CHAPIN, F.S., III, AND W.C. OECHEL. 1983. Photosynthesis, hybrid Populus clones to flooding, drought, and nitro- respiration, and phosphate absorption by Carex gen availability. I. Morphology and growth. Cana- aquatilis ecotypes along latitudinal and local envi- dian Journal of Botany 70:2265–2270. ronmental gradients. Ecology 64:743–751. LOISELLE, B.A., V.L. SORK, J. NASON, AND C. GRAHAM. 1995. CRAWFORD, C.S., S.C. CULLY, R. LEUTHEUSER, M.S. SIFUEN- Spatial genetic structure of a tropical understory TES, L.H. WHITE, AND J.P. WILBUR. 1993. Middle shrub, Psychotria officinalis (Rubiaceae). American Rio Grande Ecosystem: Bosque Biological Manage- Journal of Botany 82:1420–1425. ment Plan. U.S. Department of Interior, Fish and MCGEE, A.B., M.R. SCHMIERBACH, AND F.A. BAZZAZ. 1981. Wildlife Service. Photosynthesis and growth in populations of Populus DANG, Q.L., C.Y. XIE, C. YING, AND R.D. GUY. 1994. Genetic deltoides from contrasting habitats. American Mid- variation of ecophysiological traits in red alder (Alnus land Naturalist 105:305–311. rubra Bong.). Canadian Journal of Forest Research MONJE, O.A., AND B. BUGBEE. 1992. Inherent limitations 24:2150–2156. of nondestructive chlorophyll meters: a comparison DICKMANN, D.I. 1971. Photosynthesis and respiration by of two types of meters. Hortscience 27:69–71. developing leaves of cottonwood (Populus deltoides PAUL, N.K., AND C.F. EAGLES. 1988. Cultivar differences Bartr.). Botanical Gazette 132:253–259. in net photosynthetic rates and their relationship DONAHUE, R.A., T.D. DAVIS, C.H. MICHLER, D.E. RIEMEN- with leaf anatomical characteristics in Brassica. Pho- SCHNEIDER, D.R. CARTER, P.E. MARQUARDT, N. tosynthetica 22:320–327. SANKHLA, ET AL. 1994. Growth, photosynthesis, and PERRY, D.J., AND P. K NOWLES. 1991. Spatial genetic struc- herbicide tolerance of genetically modified poplar. ture within three sugar maple (Acer saccharum Canadian Journal of Forest Research 24:2377–2383. Marsh.) stands. Heredity 66:137–142. DUNLAP, J.M., J.H. BRAATNE, T.M. HINCKLEY, AND R.F. ROOD, S.B., AND S. HEINZE-MILNE. 1989. Abrupt down- STETTLER. 1993. Intraspecific variation in photosyn- stream forest decline following river damming in thetic traits of Populus trichocarpa. Canadian Journal southern Alberta. Canadian Journal of Botany of Botany 71:1304–1311. 67:1744–1749. GEBER, M.A., AND T.E. DAWSON. 1990. Genetic variation ROOD, S.B., AND J.M. MAHONEY. 1990. Collapse of riparian in and covariation between leaf gas exchange, mor- poplar forests downstream from dams in western phology, and development in Polygonum arenastrum, prairies: probable causes and prospects for mitiga- an annual plant. Oecologia 85:153–158. tion. Environmental Management 14: 451–464. GUREVITCH, J., J.A. TEERI, AND A.M. WOOD. 1986. Differ- ROWLAND, D.L., B. BIAGINI, AND A.S. EVANS. 2000. Vari- entiation among populations of Sedum wrightii ability among five riparian cottonwood (Populus fre- (Crassulaceae) in response to limited water availabil- montii Wats.) populations: an examination of size, ity: water relations, CO2 assimilation, growth and density, and spatial distribution. Western North Amer- survivorship. Oecologia 70:198–204. ican Naturalist 60:384–393. HENDRICKX, J.M.H., B. BAERENDS, Z.I. RAZA, M. SADIG, SAS INSTITUTE. 1989. SAS Institute Inc., Cary, NC. AND M.A. CHAUDHRY. 1992. Soil salinity assessment SCHOLANDER, P.F., H.T. HAMMEL, E.D. BRADSTREET, AND by electromagnetic induction of irrigated land. Soil E.A. HEMMINGSEN. 1965. Sap pressure in vascular Science Society of America Journal 56:1933–1941. plants. Science 148:339–346. HOWE, W.H., AND F.L. KNOPF. 1991. On the imminent SCHULTE, P.J., T.M. HINCKLEY, AND R.F. STETTLER. 1987. decline of Rio Grande cottonwoods in central New Stomatal responses of Populus to leaf water potential. Mexico. Southwestern Naturalist 36:218–224. Canadian Journal of Botany 65:255–260. IACOBELLI, A., AND J.H. MCCAUGHEY. 1993. Stomatal con- SLAVIK, B. 1974. Methods of studying plant water relations. ductance in a northern temperate deciduous forest: Springer-Verlag, New York. 449 pp. temporal and spatial patterns. Canadian Journal of SOKAL, R.R., AND F. J . R OHLF. 1995. Biometry. W.H. Free- Forest Research 23:245–252. man and Company, New York. 887 pp. ISEBRANDS, J.G., AND P.R. LARSON. 1973. Anatomical SYSTAT. 1996. SYSTAT: statistics. SPSS Inc., Chicago, IL. changes during leaf ontogeny in Populus deltoides. TYREE, M.T., K.J. KOLB, S.B. ROOD, AND S. PATIÑO. 1994. American Journal of Botany 60:199–208. Vulnerability to drought-induced cavitation of ripar- JACQUARD, P., AND K. URBANSKA. 1988. Population genetics, ian cottonwoods in Alberta: a possible factor in the whole-plant physiology, and population biology: inter- decline of the ecosystem. Tree Physiology 14:455–466. actions with ecology. Acta Oecologia 9:3–10. YOUNG, A.G., AND H.G. MERRIAM. 1994. Effects of forest LAKSO, A.N. 1982. Precautions on the use of excised shoots fragmentation on the spatial genetic structure of Acer for photosynthesis and water relations measurements saccharum Marsh. (sugar maple) populations. Hered- of apple and grape leaves. HortScience 17:368–370. ity 72:201–208. LAMOREAUX, R.J., W.R. CHANEY, AND K.M. BROWN. 1978. The plastochron index: a review after two decades of Received 31 August 1999 use. American Journal of Botany 65:586–593. Accepted 14 June 2000 LARSON, P.R., AND J.G. ISEBRANDS. 1971. The plastochron index as applied to developmental studies of cottonwood. Canadian Journal of Forest Research 1:1–11. Western North American Naturalist 61(4), © 2001, pp. 463–472

EXPERIMENTAL MANIPULATIONS OF PRECIPITATION SEASONALITY: EFFECTS ON OAK (QUERCUS) SEEDLING DEMOGRAPHY AND PHYSIOLOGY

Jake F. Weltzin1,2, Keirith A. Snyder1, and David G. Williams1

ABSTRACT.—Predicted changes in regional precipitation patterns and soil moisture caused by anthropogenic trace gas emissions may affect the distribution and abundance of woody plants in arid and semiarid regions. To test the response of woody plants to potential changes in precipitation regimes, we manipulated summer and winter precipitation on plots that contained seedlings of Quercus emoryi Torr. (Emory oak), the dominant tree in oak savannas of the southwestern United States. Throughout the growing season, we monitored seedling survival and physiology (predawn leaf water potential, midday instantaneous gas exchange, and leaf carbon isotope discrimination). Seedling survival and physiological performance differed little between treatments, which embodied 50% changes to quantities of summer and winter precipitation, and encompassed a continuum of precipitation from 359 mm ⋅ year–1 to 846 mm ⋅ year–1. How- ever, survival and physiological performance of seedlings were negatively impacted by seasonal environmental conditions common to all treatments, especially during the annual pre-‘monsoon’ drought. Seedling predawn leaf water potentials, net CO2 assimilation, and stomatal conductance indicate that growing conditions for Q. emoryi seedlings at this site are generally restricted to periods with adequate soil moisture (i.e., April and August). Results contrast with an assumption implicit to the “two-layer” soil water resource partitioning hypothesis that woody plants in all life history stages are more dependent upon winter than summer precipitation. In fact, summer precipitation appears more important than winter precipitation for Q. emoryi seedling recruitment and growth.

Key words: Quercus emoryi, precipitation seasonality, seedling recruitment, population demographics, carbon isotope discrimination, leaf gas exchange.

Increasing atmospheric carbon dioxide changes in seasonal precipitation and soil mois- concentration is expected to increase global ture regimes may cause major shifts in plant temperatures and thereby alter the amount, species composition, distribution, and abun- seasonality, and intensity of precipitation on dance (Stephenson 1990). For example, poten- global to regional scales (Houghton et al. 1996, tial increases in winter soil moisture content Mahlman 1997, Giorgi et al. 1998). Although may favor deeply rooted woody plants with considerable research has described the effects the C3 photosynthetic pathway and facilitate of increasing atmospheric carbon dioxide con- their recruitment within arid and semiarid centration (e.g., Koch and Mooney 1996, grasslands dominated by C4 grasses (Neilson Körner and Bazzaz 1996) and expected increase 1986, 1993, Melillo et al. 1996). Alternatively, in temperature (e.g., Chapin et al. 1995, Harte increases in summer precipitation may favor and Shaw 1995, Beerling and Woodward 1996) shallow-rooted species and C4 grasses (e.g., on ecosystems, little research has focused on Walter 1954, 1979, Knoop and Walker 1985, potential effects of changes in the amount Ehleringer et al. 1991, Lauenroth et al. 1993, or seasonality of precipitation anticipated in Burgess 1995, Neilson and Drapek 1998). the next few decades (Weltzin and McPherson However, recent research suggests that as 1995). However, changes in precipitation the ratio of summer to winter precipitation regimes are expected to have important rami- increases, long-lived (woody) perennials will fications for the structure, composition, and shift from preferential development of deep diversity of ecosystems (e.g., VEMAP Mem- roots to a dimorphic root system with active bers 1995, Neilson and Drapek 1998). roots in both deep and shallow soil layers In arid and semiarid regions where vegeta- (Ehleringer and Dawson 1992, Dawson and tion is highly dependent upon precipitation, Pate 1996, Weltzin and McPherson 1997,

1School of Renewable Natural Resources, University of Arizona, Tucson, AZ 85721. 2Corresponding author. Present address: Department of Ecology and Evolutionary Biology, 569 Dabney Hall, University of Tennessee, Knoxville, TN 37996-1610.

463 464 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Williams and Ehleringer 2000). Similarly, both nated by C4 perennial bunchgrasses (Brown the physiologic performance and demographic 1982, McClaran and McPherson 1999). performance of woody plants in western North The study site is located in lower Garden America are often coupled to the quantity and Canyon (31°29′N, 110°20′W) on Fort Huachuca timing of summer precipitation (Neilson and Military Reservation (FHMR) near Sierra Vista, Wullstein 1983, 1985, Ehleringer et al. 1991, Arizona. During the study period overstory tree Lin et al. 1996, Weltzin and McPherson 2000, cover within the savanna was 11%, as estimated Williams and Ehleringer 2000). Physiologic from aerial photographs (Haworth and McPher- and demographic studies of adult woody son 1994). Herbaceous vegetation was domi- plants are constrained by the fact that environ- nated by the perennial bunchgrass Trachypo- mental conditions sufficient for survival of gon montufari (H.B.K.) Nees. The site is 1550 adult plants are often insufficient for recruit- m in elevation with a 5% slope on a northeast- ment of seedlings (sensu Grubb 1977). ern aspect. Soils developed from gravelly allu- We investigated how amount and seasonal- vium. Climate is semiarid, with an average ity of precipitation affect the physiologic and annual temperature of 20°C. Average annual demographic performance of oak (Quercus L.) precipitation of 602 mm is bimodally distrib- seedlings in oak savannas of the southwestern uted, with peaks during the summer ‘monsoon’ United States (Brown 1982, McClaran and (July–September) and during winter (Decem- McPherson 1999). This region is characterized ber–February; NOAA 1996). Weltzin and by a bimodal precipitation regime, with peaks McPherson (2000) provide further details on in amount of precipitation in both summer climate, vegetation, and soils at this site. (52% of annual precipitation) and winter (29% Experimental Design of annual precipitation). This regional precipitation regime is likely to change within the In June 1994 we initiated a field experi- next century as atmospheric CO2 concentra- ment consisting of 5 simulated precipitation tion increases (Houghton et al. 1996, Giorgi et treatments applied to plots isolated from al. 1998), although the extent and direction of ambient precipitation and soil moisture. The these changes are difficult to predict (Mahlman 1st treatment received simulated precipitation 1997). equivalent to the long-term (30-year) mean We used a manipulative field experiment to annual precipitation for the site (602 mm ⋅ simulate potential scenarios of precipitation year–1; Table 1). The other 4 treatments re- redistribution that southwestern oak savannas ceived all possible combinations of 50% addi- may experience by the mid- to late 21st cen- tions and reductions of summer (July–Sep- tury. Recruitment and production of seedlings tember) and winter (December–February) pre- of the dominant savanna tree, Quercus emoryi cipitation relative to the long-term seasonal Torr. (Emory oak), are described in Weltzin mean. Treatments received equal amounts of and McPherson (2000). The objectives of this precipitation in spring (defined herein as study were to (1) assess physiological perfor- March–June) and autumn (October–November). mance of these seedlings and (2) compare This experimental design incorporated measures of seedling physiology with seedling changes to both seasonal and total precipita- demographic responses. tion because changes in atmospheric circulation that will accompany climate change will MATERIALS AND METHODS produce regional changes in both the amount and seasonality of precipitation (Houghton et Study Site al. 1996, Mahlman 1997, Giorgi et al. 1998). We conducted research between 1994 and Because effects of interannual variation in 1996 at the lower (and drier) margin of tem- total precipitation on plant communities have perate, evergreen oak woodland at the base of been well studied (e.g., Weaver and Clements the Huachuca Mountains in southeastern Ari- 1929, Stephenson 1990), we focused on the lit- zona, USA. The ecotone between oak wood- tle-studied component of precipitation season- land and adjacent semidesert grassland is ality (see Neilson 1986, Neilson et al. 1992). characterized by Q. emoryi–dominated savan- Treatments were arranged within a ran- nas bordered by semidesert grassland domi- domized complete block design (n = 4). Blocks 2001] CHANGING PRECIPITATION AND OAK RECRUITMENT 465

TABLE 1. Season, frequency of application, and amount (mm) of 5 precipitation treatments (n = 4) applied to plots isolated from ambient precipitation and soil moisture at an oak savanna site in southeastern Arizona, USA. Long-term mean represents the 30-year average seasonal precipitation for the site, and seasonal wet and dry treatments represent 50% additions and reductions, respectively, of the long-term seasonal mean.

______Treatment Long-term Summer dry/ Summer dry/ Summer wet/ Summer wet/ Season Months Frequency mean winter wet winter dry winter wet winter dry Spring MAMJ 7 62 62 62 62 62 Summer JAS 29 315 158 158 473 473 Autumn ON 7 53 53 53 53 53 Winter DJF 14 172 258 86 258 86 TOTAL 57 602 531 359 846 674 were established within homogeneous stands visual examination and flotation (Nyandiga of perennial bunchgrasses. Within each block, and McPherson 1992), and planted 49 acorns we linearly arranged five 1.2 m × 1.5 m-plots at 10-cm spacing into each plot. Survival of at 1.5-m spacing. The perimeter of each plot emerged seedlings was monitored throughout was trenched to 1-m depth and lined with the experiment. When seedlings were about 8 polyethylene film to prevent lateral movement months old, we started our assessments of leaf of soil water. The edge of each plot was bor- water potential and leaf gas exchange. Quercus dered to prevent lateral movement of surface emoryi are evergreen and accumulate little water. Vegetation in each plot was left intact. A aboveground biomass during the first several permanent precipitation shelter (16 m × 4 m) years after germination (Weltzin and McPher- constructed of steel tubing, clear polyethylene son 2000). Because seedlings in this experi- film, and fence posts was erected over each ment developed few, if any, new leaves in block to exclude ambient precipitation (Fig. 1). 1996, we sampled leaves initiated in 1995. The pitched roof of each shelter was 2.2 m We determined Q. emoryi predawn leaf aboveground at its apex and 1.5 m high along water potential (Ψ) with a Scholander-type the sides and ends. Poultry netting (2.5-cm pressure chamber (PMS Instrument Com- mesh) was wired to fence posts and rebar pany, Corvallis, OR) on 20 April, 30 June, 22 stakes around each block to form a 60-cm-tall August, and 17 October 1996. In particular, on vertebrate exclosure. each date we selected one seedling at random We kept the shelters open-sided to minimize from each plot. During a period of 1–3 hours microclimatic impact. Shelters reduced photo- before the beginning of the daily photoperiod, synthetically active photon flux density by 29% we collected one leaf at random from near the ± 10% (mean ± 1s–) at solar noon on a clear, x top of the seedling canopy for determination midsummer day. Although shelters likely al- Ψ tered other, unquantified microenvironmental of (n = 4). On each date we also used a variables (e.g., ambient temperature, relative portable open-loop photosynthesis system humidity), experimental units were affected (CIRAS-1 CO2/H2O Infrared Gas Analysis equally. System, PP Systems, Haverhill, MA) to deter- Precipitation collected and stored on-site mine midday net CO2 assimilation (A), the was applied to plots according to a randomly ratio of leaf intercellular to ambient CO2 con- generated precipitation regime that simulated centration (ci/ca), and stomatal conductance natural precipitation patterns (Nicks and Lane (gs) (n = 4) of randomly selected seedlings 1989; CLIGEN, USDA-ARS Southwestern other than those used for assessment of Ψ. Watershed Research Center, J. Stone, personal Quercus emoryi leaves used for Ψ were communication). Simulated precipitation events, retained for carbon isotope (δ13C) analysis ranging from 1 mm to 120 mm, were applied (Brugnoli and Farquhar 2000). We measured by hand-watering 57 times annually (Table 1). δ13C on finely ground, oven-dried (70°C for Additional details of the experimental design 48 hours) samples using an isotope ratio mass are in Weltzin and McPherson (2000). spectrometer (delta S, Finnigan MAT, San Jose, On 17 July 1995 we collected Q. emoryi CA) at the University of Utah Stable Isotope acorns from trees on-site, sorted them by Ratio Facility for Environmental Research 466 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Fig. 1. Experimental system for capturing and redistributing precipitation. Each of 4 precipitation shelters (16 m × 4 m) was covered with clear polyethylene film to exclude ambient precipitation from experimental plots. Ambient precipitation was stored in on-site tanks for later application to plots.

(SIRFER). δ13C values of leaves were used to demography by assessing correlations between calculate discrimination as ∆ and seedling survival rates. Seedling survival rates were calculated as the change in ∆ δ δ δ = ( a – p)/(1 + p) (1) percentage survival divided by the number of days within a given time period, and are δ δ13 δ –1 where p is C of leaves and a is that of the expressed as % day . Seedling survival rates air (–8‰) according to Farquhar et al. (1989). were determined for 4 time periods: 60 and 30 ∆ during photosynthesis is directly related to days before determination of ∆ (PRE60 and ci/ca as PRE30, respectively), and 30 and 60 days after determination of ∆ (POST30 and POST60, ∆ = a + (b – a)ci/ca (2) respectively). These 30- and 60-day time periods do not correspond exactly with monthly where a and b represent fractionations associ- and bimonthly assessment of seedling survival ∆ ∆ ated with CO2 diffusion into the leaf (4.4‰) and , respectively, because survival and and carboxylation (27‰), respectively. The ∆ were not necessarily determined on the same value forms the basis of a flux integrated esti- date of the month. POST30 and POST60 were mate of ci/ca in C3 plants (Farquhar et al. not determined for the October sample date 1989) and reflects the balance between bio- because the experiment was terminated shortly chemical demand for CO2 by the chloroplasts thereafter. and CO supply through stomata. ∆ integrates 2 Statistical Analyses ci/ca over the active periods of photosynthesis and leaf formation and is frequently correlated For each sample date we used analysis of to stomatal conductance and drought stress variance (ANOVA; SAS procedure GLM, SAS (Ehleringer 1990, Meinzer et al. 1992). Institute 1989) to evaluate random and fixed To assess relationships between seedling effects of block and treatment, respectively, on Ψ ∆ physiology and demography on each sample seedling survival, A, ci/ca, gs, , and . We date, and for all dates combined, we examined used Fisher’s protected LSD (Fisher 1960) a correlations between seedling survival and A, posteriori mean separation tests for significant Ψ ∆ ci/ca, gs, , and . We further investigated treatment effects (P < 0.05 unless otherwise relationships between seedling physiology and indicated). In addition, we compared summer- 2001] CHANGING PRECIPITATION AND OAK RECRUITMENT 467 wet vs. summer-dry and winter-wet vs. winter- Net CO2 assimilation (A) and ci/ca at mid- dry treatments using single-degree-of-free- day did not differ between treatments on any dom contrasts (Zar 1996; P-values are for date (Table 2). Stomatal conductance (gs) dif- ANOVAs unless otherwise indicated). fered between treatments only in April, when We used Pearson and Spearman rank corre- gs was about 2.5 times higher in winter-wet lation procedures (SAS procedure CORR, SAS than winter-dry treatments (Table 2). On all Institute 1989) to assess correlations between sample dates carbon isotope discrimination (∆) Ψ percent seedling survival and A, ci/ca, gs, , was greater (contrast P < 0.06) for seedlings in and ∆ for each sample date (n = 20) and for all wet summer treatments than seedlings in dry sample dates combined (n = 80). Spearman summer treatments (Table 2). rank correlation coefficients for ∆ vs. PRE60, Percent seedling survival was positively PRE30, POST30, and POST60 were deter- correlated with ∆ when all dates were consid- mined for all treatments at each sample date ered collectively (r = 0.28, P = 0.01, n = 79) (n = 20 except POST30 and POST60 in Octo- and in August (r = 0.50, P = 0.02, n = 20). ber), and for all sample dates combined (n = Percent survival and ci/ca were negatively cor- 80 except POST30 and POST60 where n = 60). related only when all dates were considered (r For each correlation analysis, we performed = –0.29, P = 0.01, n = 77). Percent survival Ψ sequential Bonferroni corrections to control was not correlated with A, gs, or for any the group-wide type I error rate (Rice 1989). given sample date or when all dates were considered (P > 0.16, data not shown). We used least-squares regression analysis ∆ (SAS procedure REG, SAS Institute 1989) to Carbon isotope discrimination ( ) was not correlated with rates of seedling survival investigate the relationship between c /c and i a either 30 or 60 days before or after assessment ∆ for all treatments for each date (n = 20), and of ∆ when sample dates were considered col- for all dates combined (n = 80). lectively (Table 3). In contrast, PRE60 and Prior to analysis, all data were tested for POST60 were positively correlated with ∆ normality with the Shapiro-Wilk W-statistic determined in August and June, respectively. (Shapiro and Wilk 1965). Data not normally Although variations in ∆ explained only about distributed (P < 0.05) were transformed or 38% of the variation in survival rates, survival ranked as appropriate. Percent seedling sur- rates between June and August were greater vival data were arcsine-transformed prior to in summer-wet plots, where seedlings had analysis (Zar 1996). higher ∆ values, than in summer-dry plots, where seedlings had lower ∆ values (Fig. 2). RESULTS The ratio of intercellular to ambient CO2 ∆ concentration (ci/ca) was a poor predictor of Seedling survival rates ranged from 76–90% across the growing season, and for most dates in April to 38–58% in October (Table 2). Sur- within the growing season. In particular, ci/ca vival rates differed between treatments only in and ∆ were correlated only in August (r = June (P = 0.06), when survival was greater in 0.51, P = 0.03) and, to a lesser extent, in April the summer-dry/winter-dry treatment than in (r = 0.45, P = 0.06) and October (r = 0.40, P the summer-wet/winter-wet and long-term ∆ = 0.09). ci/ca and were uncorrelated in June mean treatments. Additional details of seed- (r = 0.02, P = 0.94) and when all dates were ling survival are in Weltzin and McPherson considered (r = 0.05, P = 0.67). (2000). Seedling predawn leaf water potentials (Ψ) DISCUSSION differed (P = 0.002) between treatments only in October, when Ψ was lower in summer-dry Q. emoryi Demography (–4.8 MPa) than summer-wet plots (–1.9 MPa), and Physiology and long-term mean plots were intermediate Surprisingly, seedling survival differed lit- (–3.5 MPa; Table 2). Ψ did not differ (P > tle between treatments, which embodied 50% 0.05) between treatments in April (–1.6 MPa) changes in quantities of summer and winter or August (–0.8 MPa). Ψ in June were less precipitation and represented a continuum of than –6 MPa (i.e., the lower limit of the pres- precipitation from 359 mm ⋅ year–1 to 846 mm sure chamber) for all sample units. ⋅ year–1 (Table 1). There were also few treat- 468 WESTERN NORTH AMERICAN NATURALIST [Volume 61

± Ψ µ –2 –1 TABLE 2. Mean ( sx–) survival (%), predawn leaf water potential ( ; MPa), net CO2 assimilation (A; mol m s ), ci/ca, –2 –1 ∆ stomatal conductance (gs; mol m s ), and carbon isotope discrimination ( ) of Q. emoryi seedlings (n = 4).

______Treatment Long-term Summer dry/ Summer dry/ Summer wet/ Summer wet/ Variable Date mean winter wet winter dry winter wet winter dry Survival (%) April 79 (2)1 76 (6) 90 (4) 85 (4) 85 (6) June 52 (11) a 70 (7) ab 87 (5) b 66 (17) a 69 (13) ab August 39 (9) 58 (7) 41 (19) 58 (18) 59 (14) October 38 (6) 58 (7) 40 (20) 58 (18) 57 (14) Ψ (MPa) April –1.5 (0.2) –1.1 (0.1) –2.0 (0.2) –1.4 (0.5) –2.0 (0.7) June <–6 <–6 <–6 <–6 <–6 August –0.9 (0.4) –1.4 (0.6) –0.6 (0.1) –0.6 (0.1) –0.6 (0.2) October –3.5 (0.2) a –4.2 (1.2) ab –5.4 (0.7) b –1.6 (0.2) c –2.1 (0.2) c A (µmol m–2 s–1) April 8.8 (2.4) 7.8 (0.9) 7.2 (1.6) 7.0 (2.0) 5.8 (2.2) June –1.9 (1.4) –0.9 (0.8) –0.5 (0.6) –0.2 (1.6) –0.1 (0.8) August 6.2 (2.0) 4.6 (1.7) 4.4 (1.2) 8.9 (3.3) 3.5 (1.4) October 0.1 (0.9) –0.7 (0.3) –0.8 (0.6) 1.0 (1.3) 0.9 (0.3) ci/ca April 0.13 (0.08) 0.40 (0.15) 0.15 (0.15) 0.44 (0.09) — June 1.18 (0.14) 1.02 (0.09) 0.97 (0.04) 1.00 (0.21) 0.96 (0.17) August 0.79 (0.04) 0.73 (0.01) 0.74 (0.04) 0.74 (0.04) 0.81 (0.04) October 0.91 (0.14) 1.01 (0.03) 1.09 (0.09) 0.79 (0.15) 0.84 (0.03) –2 –1 gs (mol m s ) April 0.041 (0.009) a 0.079 (0.023) a 0.036 (0.006) a 0.063 (0.014) a 0.016 (0.007) b June 0.059 (0.024) 0.025 (0.003) 0.061 (0.021) 0.040 (0.014) 0.014 (0.003) August 0.164 (0.008) 0.084 (0.026) 0.093 (0.011) 0.191 (0.067) 0.090 (0.026) October 0.026 (0.002) 0.026 (0.005) 0.029 (0.013) 0.034 (0.008) 0.036 (0.007) ∆ April 18.1 (0.8) 18.5 (0.5) 17.3 (1.0) 19.4 (0.4) 19.3 (0.4) June 19.0 (0.2) ac 18.2 (0.4) ab 17.7 (0.06) b 19.2 (0.3) c 19.3 (0.4) c August 17.3 (0.4) 17.5 (1.0) 17.2 (0.9) 19.7 (0.6) 18.7 (0.4) October 18.3 (0.1) a 18.1 (0.4) a 16.7 (0.8) b 19.0 (0.4) a 19.4 (0.3) a1

1Means in rows with different lowercase letters differed (ANOVA, P < 0.05).

ment effects on point measures of leaf gas ductance, negative rates of CO2 assimilation, exchange (A, gs, ci/ca) and leaf water potential ci/ca ratios near one, and predawn leaf water (Ψ). Further, there were few correlations be- potentials lower than our instrument could tween percent survival and point measures of measure. Other studies have similarly sug- Ψ seedling physiology (A, gs, ci/ca, ) at any gested that the pre-monsoon drought is a criti- given date or across the growing season. In all, cal bottleneck to Q. emoryi seedling demo- these results suggest that once established, Q. graphy (Pase 1969, Neilson and Wullstein 1983, emoryi seedlings were relatively insensitive to McPherson 1992, Germaine and McPherson environmental conditions imposed in this study 1998, 1999, Weltzin and McPherson 2000). (see also Weltzin and McPherson 2000). Although data presented herein represent However, seedling survival and physiologi- only a single growing season, survival rates for cal performance were negatively impacted by a separate Q. emoryi seedling cohort in this environmental conditions common to all treat- same experiment differed little between treatments. Seedling survival across all treatments ments after 3 growing seasons (Weltzin and declined on average from 83% to 50% during McPherson 2000). In contrast, Q. emoryi seed- the course of the growing season. The seasonal ling recruitment rates (i.e., the number of indi- drought that occurs during May and June prior viduals added to the population) for both to the onset of the summer rains in July cohorts were as much as 300% greater in sum- appeared to be most detrimental to seedling mer-wet than summer-dry treatments. Recruit- demography. First, rates of seedling survival ment was more directly attributable to treat- were lowest during this period, with apparent ment effects on emergence of seedlings from lag effects on survival lasting until August acorns planted during the summer monsoon. (Weltzin and McPherson 2000). Second, at the In contrast to point measures of leaf gas height of the pre-monsoon drought in June, exchange, seedling survival rates could roughly seedlings exhibited low rates of stomatal con- be predicted from ∆, which is an integrated 2001] CHANGING PRECIPITATION AND OAK RECRUITMENT 469

TABLE 3. Pearson correlation coefficients (top) and associated P-values (bottom) for ∆ vs. integrated seedling survival (PRE60, PRE30, POST30, POST60). ∆ ______Sample date PRE60a PRE30 POST30 POST60 All dates 0.08 –0.06 –0.03 –0.002 0.46 0.61 0.83 0.99 20 April –0.26 –0.32 –0.36 –0.34 0.28 0.19 0.13 0.16 30 June –0.34 –0.45 0.36 0.60 0.14 0.05 0.11 0.005 22 August 0.62 0.34 0.25 0.23 0.004 0.14 0.30 0.34 17 October 0.25 –0.04 — — 0.29 0.87 — — aPRE60 and PRE30 = slope of seedling survival over 60- and 30-day period (% day–1) before determination of ∆, respectively; and POST30 and POST60 = slope of seedling survival over 30- and 60-day period (% day–1) after determination of ∆, respectively. Significant coefficients after sequential Bonferroni correc- tion (Rice 1989) are in boldface. measure of leaf gas exchange. Generally, seedling survival was positively correlated with ∆, which is consistent with the expectation that metabolic activity is coupled with population demographics. Correlations between ∆ and seedling survival were greatest during the annual summer monsoon, which suggests that the demographics of these stress-tolerant seedlings can best be predicted from ∆ when environmental conditions are suitable for extended periods of carbon assimilation. However, observed differences in ∆ between summer-wet and summer-dry treatments in April and June 1996 (i.e., prior to experimental application of summer precipitation in 1996) suggest that carbon fixed during summer 1995 Fig, 2. Relationship between ∆ for Q. emoryi leaves was retained in the evergreen leaves of these sampled 22 August and seedling survival rate (% day–1) for seedlings into the next year. This could reduce 60-day period before (PRE60) determination of ∆ (r2 = the sensitivity of ∆ to environmental condi- 0.38, P = 0.004). tions in other seasons when resource gradients are less intense, and constrains the usefulness However, carbon isotope discrimination of whole evergreen leaves as measures of ∆ ∆ plant response on relatively short (i.e., a single values ( ) and ci/ca- relationships suggest season) temporal scales. that summer precipitation is more important than winter precipitation for Q. emoryi carbon Seasonal Precipitation and accumulation in leaves. This conclusion is Q. emoryi Demographics supported by recent research which indicates Seedling predawn leaf water potentials, net that recruitment of Q. emoryi seedlings was positively correlated with the quantity of sum- CO2 assimilation, and stomatal conductance indicate that growing conditions for Q. emoryi mer precipitation but was independent of seedlings at this site are generally restricted to quantity of winter precipitation (Weltzin and periods with adequate soil moisture (i.e., April McPherson 2000). Other empirical studies fur- and August). This conclusion supports obser- ther indicate the importance of summer pre- vations that 1- and 2-year-old Q. emoryi seed- cipitation to Q. emoryi seedling recruitment lings at this same research site use soil water (Pase 1969, Neilson and Wullstein 1983, derived from both winter and summer precip- McPherson 1992, Germaine and McPherson itation (Weltzin and McPherson 1997). 1999, Weltzin and McPherson 1999, 2000). 470 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Similarly, physiologic performance and demo- Brian Cypher and 2 anonymous reviewers. graphic performance of other woody plants in B. Steidl provided statistical advice. Research western North America are often coupled to was supported by USDA-CSRS National the quantity and timing of summer precipita- Research Initiative Grant 92-37101-7435, tion (Neilson and Wullstein 1983, 1985, Ehle- McIntire-Stennis project ARZT-139017-M-12- ringer et al. 1991, Lin et al. 1996, Williams 118, and the UA Graduate Student Research and Ehleringer 2000). Fund. Weltzin was supported in part by a Although changes in regional precipitation UA/NASA Space Grant Graduate Student Fel- regimes are not well predicted by general cir- lowship, a Flinn Foundation Biology.21 Grad- culation models, particularly for topographi- uate Fellowship, a William G. McGinnies cally complex regions such as the southwestern Scholarship in Arid Lands Studies, and a UA United States, predicted changes in atmos- Graduate College Fellowship. Snyder was pheric circulation and surface temperatures supported by a UA/NASA Space Grant Gradu- are likely to affect the amount and seasonality ate Student Fellowship and a William G. of precipitation and soil moisture in this region McGinnies Scholarship in Arid Lands Studies. (e.g., Kattenberg et al. 1996, Giorgi et al. 1998). Results from this study and Weltzin and LITERATURE CITED McPherson (2000) indicate that changes in BEERLING, D.J., AND F.I. WOODWARD. 1996. In situ gas summer precipitation regimes would likely exchange responses of boreal vegetation to elevated constrain Q. emoryi population dynamics CO2 and temperature: first season results. Global through changes in seedling recruitment rates. Ecology and Biogeography Letters 5:117–127. Further, although adult and sapling (ca 1 m BROWN, D.E., EDITOR. 1982. Biotic communities of the tall) Q. emoryi and coexisting grasses at this American Southwest—United States and Mexico. Desert Plants 4:1–342. savanna site access water from relatively deep BRUGNOLI, E., AND G.D. FARQUHAR. 2000. Photosynthetic and shallow in the soil profile, respectively, Q. fractionation of carbon isotopes. Pages 399–434 in emoryi seedlings are unable to access deep R.C. Leegood, T.D. Sharkey, and S. von Caemmerer, sources of soil water for at least the first 3 editors, Photosynthesis: physiology and metabolism. Advances in photosynthesis. Kluwer Academic Pub- growing seasons after germination (Weltzin and lishers, Boston, MA. McPherson 1997, 2000). These results con- BURGESS, T.L. 1995. Desert grassland, mixed shrub savanna, trast with an assumption implicit to the “two- shrub steppe, or semidesert scrub? The dilemma of layer” hypothesis (Walter 1954, 1979) that coexisting growth forms. Pages 31–67 in M.P. McCla- woody plants in all life history stages are more ran and T.R. Van Devender, editors, The desert grassland. University of Arizona Press, Tucson. dependent upon winter precipitation than CHAPIN, F.S., G.R. SHAVER, A.E. GIBLIN, K.J. NADELHOF- summer precipitation. Thus, models devel- FER, AND J.A. LAUNDRÉ. 1995. Response of arctic oped to predict the effect of changing climates tundra to experimental and observed changes in cli- on the abundance and distribution of woody mate. Ecology 76:694–711. plants (e.g., Emanuel et al. 1985, VEMAP Mem- DAWSON, T.E., AND J.S. PATE. 1996. Seasonal water uptake and movement in root systems of Australian phrae- bers 1995, Iverson and Prasad 1998) should atophytic plants of dimorphic root morphology: a consider spatial and temporal processes that stable isotope investigation. Oecologia 107:13–20. constrain the establishment of individuals EHLERINGER, J.R. 1990. Correlations between carbon iso- (Grubb 1977, Harper 1977, McPherson 1997, tope discrimination and leaf conductance to water vapor in common beans. Plant Physiology 93: Scholes and Archer 1997). This should be facil- 1422–1425. itated by the development of dynamic global EHLERINGER, J.R., AND T.E. DAWSON. 1992. Water uptake vegetation models (e.g., Foley et al. 1998, by plants: perspectives from stable isotope composi- Neilson and Drapek 1998) that incorporate tion. Plant, Cell and Environment 15:1073–1082. such transient processes as emergence and EHLERINGER, J.R., S.L. PHILLIPS, W.S.F. SCHUSTER, AND D.F. SANDQUIST. 1991. Differential utilization of sum- recruitment. mer rains by desert plants. Oecologia 88:430–434. EMANUEL, W.R., H.H. SHUGART, AND M. STEVENSON. 1985. ACKNOWLEDGMENTS Climatic change and the broad-scale distribution of terrestrial ecosystem complexes. Climatic Change G. Bolton and E. Pendall provided field 7:29–43. FARQUHAR, G.D., J.R. EHLERINGER, AND K.T. HUBICK. 1989. assistance, and G. McPherson provided help- Carbon isotope discrimination and photosynthesis. ful comments and reviewed early drafts of the Annual Review of Plant Physiology and Molecular manuscript. We appreciate the comments of Biology 40:503–537. 2001] CHANGING PRECIPITATION AND OAK RECRUITMENT 471

FISHER, R.A. 1960. The design of experiments. 7th edi- in R.C. Anderson, J.S. Fralish, and J. Baskin, editors, tion. Hafner, New York. Savannas, barrens and rock outcrop plant communi- FOLEY, J.A., S. LEVIS, I.C. PRENTICE, D. POLLARD, AND ties of North America. Cambridge University Press, S.L. THOMPSON. 1998. Coupling dynamic models of Cambridge, England. climate and vegetation. Global Change Biology 4: MCPHERSON, G.R. 1992. Ecology of oak woodlands in Ari- 561–579. zona. Pages 24–33 in P.F. Ffoliott, G.J. Gottfried, GERMAINE, H.L., AND G.R. MCPHERSON. 1998. Effects of D.A. Bennett, V.M. Hernandez, A. Ortega-Rubio, timing of precipitation and acorn harvest date on and R.H. Hamre, technical coordinators, Ecology emergence of Quercus emoryi. Journal of Vegetation and management of oak and associated woodlands: Science 9:157–160. perspectives in the southwestern United States and ______. 1999. Effects of biotic factors on emergence and northern Mexico. USDA Forest Service General survival of Quercus emoryi at lower treeline. Éco- Technical Report RM-218. science 6:92–99. ______. 1997. Ecology and management of North Ameri- GIORGI, F., L.O. MEARNS, C. SHIELDS, AND L. MCDANIEL. can savannas. University of Arizona Press, Tucson. 1998. Regional nested model simulations of present MEINZER, F.C., N.Z. SALIENDRA, AND C.H. CRISOSTO. day and 2xCO2 climate over the central plains of the 1992. Carbon isotope discrimination and gas exchange United States. Climatic Change 40:457–493. in Coffea arabica during adjustment to different soil GRUBB, P.J. 1977. The maintenance of species richness in moisture regimes. Australian Journal of Plant Physi- plant communities: the importance of the regenera- ology 19:171–84. tion niche. Biological Review 52:107–145. MELILLO, J.M., I.C. PRENTICE, G.D. FARQUHAR, E.D. HARPER, J.L. 1977. Population biology of plants. Academic SCHULZE, AND O.E. SALA. 1996. Terrestrial biotic Press, London. responses to environmental change and feedbacks to HARTE, J., AND R. SHAW. 1995. Shifting dominance within climate. Pages 445–481 in J.T. Houghton, G. Miera, a montane vegetation community: results of a climate- B. Filho, B.A. Callander, N. Harris, A. Kattenberg, warming experiment. Science 267:876–880. and K. Maskell, editors, Climate change 1995: the HAWORTH, K., AND G.R. MCPHERSON. 1994. Effects of science of climate change. Cambridge University Quercus emoryi on herbaceous vegetation in a semi- Press, Cambridge, England. arid savanna. Vegetatio 112:153–159. NEILSON, R.P. 1986. High-resolution climatic analysis and HOUGHTON, J.T., G. MIERA, B. FILHO, B.A. CALLANDER, N. Southwest biogeography. Science 232:27–34. HARRIS, A. KATTENBERG, AND K. MASKELL, EDITORS. ______. 1993. Transient ecotone response to climatic 1996. Climate change 1995: the science of climate change: some conceptual and modeling approaches. change. Cambridge University Press, Cambridge, Ecological Applications 3:385–395. England. NEILSON, R.P., AND R.J. DRAPEK. 1998. Potentially com- IVERSON, L.R., AND A.M. PRASAD. 1998. Predicting abun- plex biosphere responses to transient global warm- dance of 80 tree species following climate change in ing. Global Change Biology 4:505–521. the eastern United States. Ecological Monographs NEILSON, R.P., G.A. KING, AND G. KOERPER. 1992. Toward 68:465–485. a rule-based biome model. Landscape Ecology KATTENBERG, A., F. GIORGI, H. GRASSL, G.A. MEEHL, J.F.B. 7:27–43. MITCHELL, R.J. STOUFFER, T. TOKIOKA, ET AL. 1996. Climate models—projections of future climate. Pages NEILSON, R.P., AND L.H. WULLSTEIN. 1983. Biogeography 285–357 in J.T. Houghton, G. Miera, B. Filho, B.A. of two southwest American oaks in relation to atmos- Callander, N. Harris, A. Kattenberg, and K. Maskell, pheric dynamics. Journal of Biogeography 10:275–297. editors, Climate change 1995: the science of climate ______. 1985. Comparative drought physiology and bio- change. Cambridge University Press, Cambridge. geography of Quercus gambelii and Quercus turbin- KNOOP, W.T., AND B.H. WALKER. 1985. Interactions of ella. American Midland Naturalist 114:259–271. woody and herbaceous vegetation in a southern NICKS, A.D., AND L.J. LANE. 1989. Weather generator. African savanna. Journal of Ecology 73:235–253. Pages 2.1–2.19 in L.J. Lane and M.A. Nearing, edi- KOCH, G.W., AND H.A. MOONEY. 1996. Carbon dioxide tors, USDA–water erosion prediction project: hill- and terrestrial ecosystems. Academic Press, San slope profile model documentation. NSERL Report Diego, CA. 2, USDA-ARS National Soil Erosion Research Labo- KÖRNER, CH., AND F.A. BAZZAZ. 1996. Carbon dioxide, ratory, West Lafayette, IN. populations, and communities. Academic Press, San NOAA. 1996. Climatological data annual summary: Ari- Diego, CA. zona. Department of Commerce, Asheville, NC. LAUENROTH, W.K., D.L. URBAN, D.P. COFFIN, W.J. PARTON, NYANDIGA, C.O., AND G.R. MCPHERSON. 1992. Germina- H.H. SHUGART, T.B. KIRCHNER, AND T.M. S MITH. tion of two warm-temperate oaks, Quercus emoryi 1993. Modeling vegetation structure–ecosystem and Quercus arizonica. Canadian Journal of Forest process interactions across sites and ecosystems. Research 22:1395–1401. Ecological Modeling 67:49–80. PASE, C.P. 1969. Survival of Quercus turbinella and Quer- LIN, G., S. PHILLIPS, AND J.R. EHLERINGER. 1996. Mon- cus emoryi in an Arizona chaparral community. South- soonal precipitation responses of shrubs in a cold western Naturalist 14:149–156. desert community of the Colorado Plateau. Oecolo- RICE, W.R. 1989. Analyzing tables of statistical tests. Evo- gia 106: 8–17. lution 43:223–225. MAHLMAN, J.D. 1997. Uncertainties in projections of SAS INSTITUTE. 1989. SAS/STAT user’s guide, version 6. human-caused climate warming. Science 278: 4th edition. SAS Institute, Cary, NC. 1416–1417. SHAPIRO, S.S., AND M.B. WILK. 1965. An analysis of vari- MCCLARAN, M.P., AND G.R. MCPHERSON. 1999. Oak ance test for normality (complete samples). Bio- savanna of the American Southwest. Pages 275–287 metrika 52:591–611. 472 WESTERN NORTH AMERICAN NATURALIST [Volume 61

SCHOLES, R.J., AND S.R. ARCHER. 1997. Tree-grass interac- ster, P.F. Ffolliott, and A. Ortega-Rubio, technical tions in savannas. Annual Review of Ecology and coordinators, Biodiversity and management of the Systematics 28:517–544. Madrean Archipelago: the sky islands of southwest- STEPHENSON, N.L. 1990. Climatic control of vegetation ern United States and Northwestern Mexico. USDA distribution: the role of the water balance. American Forest Service General Technical Report RM-264. Naturalist 135:649–670. ______. 1997. Spatial and temporal soil moisture resource VEMAP MEMBERS. 1995. Vegetation/ecosystem modeling partitioning by trees and grasses in a temperate and analysis project: comparing biogeography and savanna, Arizona, USA. Oecologia 112:156–164. biogeochemistry models in a continental-scale study ______. 1999. Facilitation of conspecific seedling recruit- of terrestrial ecosystem responses to climate change ment and shifts in temperate savanna ecotones. Eco- and CO2 doubling. Global Biogeochemical Cycles logical Monographs 69:513–534. 9:407–437. ______. 2000. Implications of precipitation redistribution WALTER, H. 1954. Die Verbuschung, eine Erscheinung for shifts in temperate savanna ecotones. Ecology der subtropischen Savannengebiete, und ihre ökolo- 81:1902–1913. gischen Urscachen. Vegetatio 5/6:6–10. WILLIAMS, D.G., AND J.R. EHLERINGER. 2000. Intra- and ______. 1979. Vegetation of the earth and ecological sys- interspecific variation for summer precipitation use in tems of the geo-biosphere. Springer-Verlag, New pinyon-juniper woodlands. Ecological Monographs York. 70:517–537. WEAVER, J.E., AND F.E. CLEMENTS. 1929. Plant ecology. ZAR, J.H. 1996. Biostatistical analysis. Prentice Hall, Upper McGraw-Hill, New York. Saddle River, NJ. WELTZIN, J.F., AND G.R. MCPHERSON. 1995. Potential effects of climate change on lower treelines in the Received 26 January 2000 southwestern United States. Pages 180–194 in L.F. Accepted 24 August 2000 DeBano, G.J. Gottfried, R.H. Hamre, C.B. Edmin- Western North American Naturalist 61(4), © 2001, pp. 473–479

SASQUAPERLA HOOPA, A NEW STONEFLY GENUS AND SPECIES FROM NORTHERN CALIFORNIA (PLECOPTERA: CHLOROPERLIDAE)

B.P. Stark1 and R.W. Baumann2

ABSTRACT.—Sasquaperla hoopa, a new genus and species of Chloroperlidae, is described from adults and preemergent nymphs collected in the Coast Range of northern California. Males are characterized by the epiproct reduced to a small, tablike structure bearing stiff hairs in a narrow band along the posterior margin. Females are similar to those of Sweltsa, but the nymphs bear several erect bristles lateral to depressed hair patches on the mesosternum. Modified keys are presented for adults and nymphs.

Key words: Plecoptera, Chloroperlidae, Sasquaperla, new genus, California.

For several years we and our colleagues have ethanol; those examined with scanning electron collected occasional specimens of a Sweltsa- microscopy were dehydrated through 10-minute like stonefly in the Coast Range of northern washes in 90%, 95%, and 100% ethanol, fol- California. A small series of adults taken from lowed by two 30-minute washes in hexamethyl- the Willow Creek drainage near Berry Sum- disilizane. Specimens were attached to speci- mit during the 1998 field season allowed a men stubs with double-stick copper tape, more comprehensive examination and con- sputter-coated with gold-palladium, and exam- firmed that males could not be identified using ined with an 1810D Amray scanning electron existing literature (Surdick 1985, Stewart and microscope. Specimens examined during this Harper 1996). Preliminary results suggested study are deposited in the following museums this species might require a new generic name, or collections: thus prompting K.W. Stewart and B.P. Stark to B.P. Stark Collection, Clinton, MS (BPS) search for and rear the nymph in order to Monte L. Bean Museum, Brigham Young include the prospective genus in a forthcoming University, Provo, UT (BYU) revision of their nymphal monograph (Stewart University of North Texas, Denton, TX and Stark 1988). Although the nymph is very (UNT) similar to Sweltsa and is identified to that United States National Museum of Natural genus in Stewart and Stark (1988) and Stewart History, Washington, DC (USNM) and Harper (1996), several diagnostic characters separate the nymph from known Nearctic Sweltsa. The evolution of nymphal morphol- Sasquaperla, new genus ogy is conservative within the Chloroperlidae; TYPE SPECIES.—Sasquaperla hoopa, new therefore, we consider these nymphal charac- species, by monotypy. ters and the formation of the male genitalia ADULTS.—Body pale yellow with black and sufficient for recognition of a new genus. brown markings. Wings transparent, veins pale; anal area of hindwing with 3 anal veins; METHODS forewing 2nd anal vein forked. Pale brown mesal area covers ocelli and extends forward Specimens were collected using beating over clypeus. Pronotum with black margins sheets and aquatic kicknets and by hand-pick- and irregular brown areas near median suture ing from vegetation, rocks, and debris; a few (Fig. 1). Meso- and metascutellar U-sutures preemergent nymphs were maintained in iced black. Abdomen with dark mesal pigment band styrofoam coolers during transport until they extending to tergum 8 or 9; narrow lateral emerged. Specimens were preserved in 80% pigment bands extend through segment 4.

1Department of Biology, Box 4045, Mississippi College, Clinton, MS 39058. 2Department of Zoology and Monte L. Bean Life Science Museum, Brigham Young University, Provo, UT 84602.

473 474 WESTERN NORTH AMERICAN NATURALIST [Volume 61

1 2

3 4

Figs. 1–4. Sasquaperla hoopa adult structures: 1, head and pronotum; 2, aedeagus, ventral; 3, male terminalia, dorsal; 4, female terminalia, ventral.

Mesosternal Y-arms obsolete; dark pigment and wide; paragenital plates obsolete, anterior surrounding Y-stem projects forward beyond transband of segment 10 wide and heavily pig- stem apex (Fig. 5). Cerci of 9–10 segments. mented (Fig. 3). Aedeagus membranous and MALE GENITALIA.—Epiproct unhinged, re- sparsely armed with microtrichia (Fig. 2). duced to a small, tablike structure embedded FEMALE GENITALIA.—Subgenital plate a in upturned membrane of tergum 10, which short, entire flap extending over about one- bears a narrow band of stiff hairs around the third of sternum 9 (Fig. 4), forming a blunt posterior margin (Figs. 9–10). Basal bar short point medially. 2001] NEW GENUS AND SPECIES OF CHLOROPERLIDAE 475

near the furcal pits (Fig. 5); nymphs have the tibial fringes poorly developed (Fig. 7) and have a few erect bristles located lateral to the mesosternal clothing hair patches (Fig. 8). Nymphal morphology would suggest that Sasquaperla belongs in the tribe Alloperlini, since it bears large patches of clothing hairs on the meso- and metathorax. Since the male epiproct is unhinged, generic placement according to Surdick (1985) would seem to fall in the Chloroerlini. However, the actual sister genus is obviously Sweltsa, so Sasquaperla would need to have possessed a hinged epiproct and then secondarily lost it. This expla- 5 nation would be stronger if the present condition of the male genitalia were described in a positive way as an apomorphic state, instead of Fig. 5. Sasquaperla hoopa: adult male mesosternum. simply as a lack of a hinged epiproct, which is plesiomorphic. KEY MODIFICATIONS.—Modifications are NYMPH.—Mature nymph brown with ob- provided for keys to adults or nymphs given in scure darker pattern on dorsum of head and Surdick (1985), Stewart and Harper (1996), thorax; abdominal terga with narrow, trans- and Stewart and Stark (1988). verse apical bands. Head with a few long setae near eyes, at antennal bases, and anterior cor- Surdick (1985) ners of frons. Pronotum heavily setose around Adult Male Key anterior and posterior margins; mesonotum 7. Aedeagus terminating in pair of thin, with strong bristle row along lateral margins. feathery processes; hammer absent . . . . Apical abdominal terga with a few intercalary ...... Plumiperla bristles and an apical fringe of bristles (Fig. 6). Aedeagus lacking thin, feathery pro- Thorax, abdomen, and legs covered with red- cesses; hammer present or absent . . . . . 7A brown clothing hairs; tibial and femoral silky 7A. Terminal abdominal segments lacking fringe hairs sparse (Fig. 7). Mesosternal Y- hair brushes; small median hammer on arms obsolete (Fig. 8); strong, erect bristles abdominal sternum 7; aedeagus armed with conspicuous basal band of scale- located lateral to clothing hairs on thoracic like spines ...... Triznaka sterna. Cerci of about 18 segments; cercal seg- Terminal abdominal segments with hair ments with well-developed apical whorls of brushes (Fig. 3); hammer absent from bristles (Fig. 6). abdominal sternum 7; aedeagus without DISTRIBUTION.—Coast Range of northern basal band of scale-like spines (Fig. 2) . . . California from the greater Trinity River ...... Sasquaperla drainage. Stewart and Harper (1996) DIAGNOSIS.—Adult Sasquaperla are charac- Adult Key terized by a dark middorsal stripe, by narrow lateral stripes on anterior abdominal segments, Males and by dark pronotal margins. The male epi- 119 (118′). Epiproct large, its tip hinged and elabo- rate; epiproct set in a deep groove in proct consists of a small, upturned, unhinged tergum 10; brushes of close-set setae on sclerotized tab arising from a short, wide basal lateral margins of terminal abdominal bar (Figs. 3, 9–10). Males cannot be resolved segments ...... 120 in existing keys by Surdick (1985) and Stewart 119′. Epiproct small, about as wide as long, and Harper (1996), but females and nymphs tergum 10 entire or slightly depressed; key to Sweltsa in these publications. Sasqua- brushes of close-set setae present or absent on lateral margins of terminal perla nymphs and adults have the mesosternal abdominal segments ...... 119A Y-arms obsolete (Figs. 5, 8), although in fully 119A (119′). Brushes of close-set setae on lateral mar- pigmented adults the mesosternal Y-region is gins of terminal abdominal segments; dark except for a pair of irregular pale spots aedeagus without dense basal bands of 476 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Figs. 6–8. Sasquaperla hoopa nymphal structures: 6, partial habitus; 7, right foreleg; 8, mesosternum. 2001] NEW GENUS AND SPECIES OF CHLOROPERLIDAE 477

9 10

Figs. 9–10. Sasquaperla hoopa male terminalia: 9, terga 9 and 10, Ep = epiproct; 10, epiproct apex, anterodorsal aspect.

dark scale-like armature (Fig. 2); epiproct Thick, depressed black clothing hairs ab- tip scarcely projecting forward from sent from lateral thoracic sterna; sternal posterior margin of tergum 10 (Fig. 3) . . hairs erect, light-colored ...... 6 ...... Sasquaperla 5A. Mesosternum without erect bristles lat- 119A′. Brushes of close-set setae absent from eral to clothing hair patch; tibial fringes lateral margins of terminal abdominal of silky setae well developed; mesoster- segments; aedeagus with dense basal nal Y-arms well developed ...... Sweltsa bands of dark scale-like armature; epiproct tip projecting forward from Mesosternum with several erect bristles posterior margin of tergum 10 ...... 121 lateral to clothing hair patch (Fig. 8); tibial fringes of silky setae sparse (Fig. 7); Females mesosternal Y-arms obsolete ...... 125 (124′). Size small (4–6 mm); mesosternal Y- ...... Sasquaperla ridge with median branch extending cephalad; subgenital plate scalloped with Stewart and Harper (1996) long hairs restricted to scalloped mar- Nymphal Key gins ...... Bisancora 63 (62′). Thick, depressed black hairs present in- 125′. Size larger (8–18 mm); mesosternal Y- side coxae on thoracic sterna; both pairs ridge without anterior extension of med- of wing pads divergent from body axis; ian branch; subgenital plate variable, if long pronotal fringe hairs except on side scalloped, hairs arranged otherwise . . 125A ...... 63A 125A (125′). Mesosternal Y-arms obsolete; subgenital 63′. No black hairs inside coxae on thoracic plate with slightly projecting mesal lobe sterna; wing pads variable in divergence covered with short setae (Fig. 4) ...... Sasquaperla ...... 64 125A′ Mesosternal Y-arms normal; subgenital 63A (63). Mesosternum without erect bristles lat- plate variable but without projecting eral to depressed hair patches; tibial mesal lobe armed with short setae . . . . . fringes of silky setae well developed; ...... Sweltsa mesosternal Y-arms well developed ...... Sweltsa Stewart and Stark (1988) 63A′. Mesosternum with several erect bristles Nymphal Key lateral to depressed hair patches (Fig. 8); 5. Thick, depressed black clothing hairs tibial fringes of silky setae sparse; meso- laterally on all thoracic sterna ...... 5A sternal Y-arms obsolete ...... Sasquaperla 478 WESTERN NORTH AMERICAN NATURALIST [Volume 61

ETYMOLOGY.—Sasquaperla occurs in small States National Museum, Washington, DC. streams that contain other stoneflies with Additional paratypes: California: Del Norte restricted distributions such as Salmoperla syl- Co., unnamed small stream, Hwy 199, 3 miles vanica (Baumann and Lauck 1987) and Capnia NE Hiouchi, 1 June 1991, B.P. Stark, R.W. fiali (Nelson and Baumann 1990). The prefix, Baumann, 1 (BPS). Humboldt Co., unnamed sasqua, was chosen to honor this region that tributary of Willow Creek, 2.7 miles E Berry provides habitat for so many biological trea- Summit, Hwy 299, 8 June 1984, P. Wilkinson, sures, including “Bigfoot”! D.R. Lauck, 1 , 1 , (BYU); 22 June 1985, R.W. Baumann, C.R. Nelson, M.F. Whiting, 2 Sasquaperla hoopa, , 2 (BYU); 19 May 1998, B.P. Stark, C.R. new species Nelson, S.W. Szczytko, I. Sivec, 1 , 3 MALE.—Forewing length 9–10 mm. Gen- (BPS, BYU); 20 May 2001, B.P. Stark, K.W. eral color pale yellow marked with brown and Stewart, 10 , 10 , 8 nymphs (BPS, UNT). black. Dorsal abdominal band consists of dark Humboldt Co., Cedar Creek, Hwy 299, 6 brown quadrangular and triangular segmental miles W Willow Creek (town), 15 May 1984, P. median patches and extends to tergum 9; ante- Wilkinson, D.A. Lauck, 4 (BYU); 31 May riorly located patches (terga 3–4) are quadran- 1991, R.W. Baumann, B.P. Stark, 1 (BYU). gular, whereas those on terga 5–8 are triangu- Humboldt Co., small tributary of Willow Creek, lar; all patches through tergum 8 include a Hwy 299, 6.5 miles W Willow Creek (town), pair of pale mesal spots. Tergum 9 with small 19 May 1998, C.R. Nelson, B.P. Stark, S.W. posteromesal indentation (Fig. 3); tergum 10 Szczytko, I. Sivec, 1 , 1 nymph (BYU). Hum- with broadly rounded hemiterga. Epiproct boldt Co., Little Bidden Creek, Hwy 299, near complex greatly reduced; basal bar expanded Boise Creek Campground, 15 May 1984, P. laterally but not reaching anterior margin of Wilkinson, D.A. Lauck, 1 (BYU). Trinity Co., tergum 10; unhinged apex embedded in mem- Indian Creek, Hwy 299, 1.4 miles W Hayden brane and sparsely armed with setal-like spines Flat Campground, 21 June 1985, C.M. & O.S. and a narrow band of hairs (Figs. 3, 9–10). Flint, Jr., 1 (USNM). Aedeagus membranous and sparsely armed ETYMOLOGY.—The species name honors the with pale microtrichia (Fig. 2); apex with small Hoopa people who live in the Willow Creek ventrolateral lobes. area of the Trinity Alps, California. FEMALE.—Forewing length 11–12 mm. Dor- sal abdominal stripe extends through tergum ACKNOWLEDGMENTS 8. Subgenital plate short; mesal lobe slightly We thank our colleagues C.R. Nelson, I. developed beyond lateral margins and armed Sivec, K.W. Stewart, S.W. Szczytko, and M.F. with short setae; longer submarginal setae Whiting for assistance in fieldwork and in located lateral to expanded mesal lobe (Fig. 4). sharing specimens with us. D.R. Lauck and P. NYMPH.—Preemergent nymph 9–11 mm. Wilkinson made the first specimens of this General color brown with obscure darker pat- interesting species available from their study tern on head and pronotum, mesosternum on the Willow Creek drainage and also shared bearing several erect bristles lateral to depressed their data on the original collecting localities. clothing hair patches. Abdominal terga with transverse posterior pigment bands. Abdomi- LITERATURE CITED nal terga with conspicuous intercalary bristles through segment 8, but reduced on segments BAUMANN, R.W., AND D.R. LAUCK. 1987. Salmoperla, a new 9 and 10. Red-brown clothing hairs conspicu- stonefly genus from northern California (Plecoptera: ous over most of body (Fig. 6). Perlodidae). Proceedings of the Entomological Society of Washington 89:825–830. TYPES.—Holotype , 3 and 8 para- NELSON, C.R., AND R.W. BAUMANN. 1990. New winter types from California, Humboldt Co., Slide stoneflies (Plecoptera: Capniidae) from the Coast Creek, Hwy 13, south of Fish Lake Camp- Range of California. Pan-Pacific Entomologist 66: ground, Six Rivers National Forest, 31 May 301–306. STEWART, K.W., AND P. P. H ARPER. 1996. Plecoptera. Pages 1991, R.W. Baumann, B.P. Stark, holotype and 217–266 in R.W. Merritt and K.W. Cummins, edi- 1 , 1 paratype deposited at the United tors, An introduction to the aquatic insects of North 2001] NEW GENUS AND SPECIES OF CHLOROPERLIDAE 479

America. Kendall/Hunt Publishing Co., Dubuque, SURDICK, R.F. 1985. Nearctic genera of Chloroperlinae IA. 862 pp. (Plecoptera: Chloroperlidae). Illinois Biological Mono- STEWART, K.W., AND B.P. STARK. 1988. Nymphs of North graphs 54. University of Illinois Press, Urbana. 146 pp. American stonefly genera (Plecoptera). Thomas Say Foundation 12. Entomological Society of America. Received 15 June 2001 Lanham, MD. 460 pp. Accepted 4 September 2001 Western North American Naturalist 61(4), © 2001, pp. 480–489

A HABITAT SUITABILITY MODEL FOR PYGMY RABBITS (BRACHYLAGUS IDAHOENSIS) IN SOUTHEASTERN IDAHO

Kate I. Gabler1, Laura T. Heady1, and John W. Laundré1

ABSTRACT.—A habitat suitability model was developed for pygmy rabbit (Brachylagus idahoensis) habitat on the Idaho National Engineering and Environmental Laboratory (INEEL) in southeastern Idaho. Suitable pygmy rabbit areas were characterized by greater cover and density of total shrubs and big sagebrush (Artemisia tridentata), as well as greater forb cover. Soil texture also played an important role in distinguishing suitable pygmy rabbit areas from nonuse sites. Principal components analysis (PCA) of several vegetation variables and soil texture was used to develop a habitat suitability model for pygmy rabbit habitat. This model, which can be used to successfully distinguish between pygmy rabbit use and nonuse areas on the INEEL, has the potential for use throughout the pygmy rabbit’s range.

Key words: pygmy rabbit, Brachylagus idahoensis, habitat suitability, southeastern Idaho, INEEL, vegetation, soil texture.

Pygmy rabbits (Brachylagus idahoensis) are leporids, pygmy rabbits are relatively slow and restricted to sagebrush-steppe areas of the may better elude predators when under a shrub Great Basin and adjacent intermountain regions. canopy (Orr 1940, Wilde 1978). Within this area their distribution is further Other factors that may define pygmy rabbit limited by the availability of “suitable” habitat habitat are soil depth and texture (Weiss and for the construction of burrow systems. Several Verts 1984). Kehne (1991) found 96% of pygmy studies have attempted to describe character- rabbit burrow sites in Washington in soils at istics of this suitable habitat for pygmy rabbits. least 51 cm deep, and 72% of burrow sites had These studies generally concluded that pygmy either coarse silty, ashy, or coarse loamy soils, rabbits tend to prefer taller and denser stands all with <18% clay. of big sagebrush (Artemisia tridentata) within Beyond the general preference of pygmy sagebrush-dominated areas (Grinnell et al. rabbits for taller and denser sagebrush cover 1930, Orr 1940, Severaid 1950, Green 1978, and deep, sandy soils, little more is known of Green and Flinders 1980, White et al. 1982, their specific habitat requirements. Addition- Gahr 1993, Katzner 1994, Katzner and Parker ally, there are often areas with appropriate 1997). Weiss and Verts (1984) found shrub looking woody vegetation and physiognomy cover was the best of 10 variables for distin- that are not necessarily suitable pygmy rabbit guishing sagebrush sites occupied by pygmy habitat (Green and Flinders 1980). Green and rabbits. Mean shrub cover in areas occupied Flinders (1980) hypothesized that subtle varia- by pygmy rabbits ranged from 29% in Oregon tions other than sagebrush density in the veg- (Weiss and Verts 1984) to 43–46% in Idaho etative component make an area appropriate and Wyoming (Green 1978, Green and Flinders for pygmy rabbits. This hypothesis has not yet 1980, Katzner and Parker 1997). These values been tested and the question remains: How were in contrast to 16% (Green 1978, Green specific is habitat selection by pygmy rabbits? and Flinders 1980,Weiss and Verts 1984) to Given the pygmy rabbit’s restriction to sage- 26% cover (Katzner and Parker 1997) for brush-steppe, loss of this habitat type could nonuse areas. Possible reasons for the prefer- impact the species’ survival. This is increas- ence for greater sagebrush cover are that it ingly so if only a subset of the habitat is suit- constitutes a large portion of their diet (Green able. If there is indeed a unique and identifi- and Flinders 1980) and may offer better pro- able subset of habitat factors pygmy rabbits tection from predators. Compared to larger are selecting, this information could be used

1Department of Biological Sciences, Idaho State University, Pocatello, ID 83209.

480 2001] PYGMY RABBIT (BRACHYLAGUS IDAHOENSIS) HABITAT 481 to develop a predictive habitat suitability model. This model could then be used to assess the suitability of specific areas for pygmy rabbits and help develop sound conservation management plans for the species. We investigated habitat requirements of pygmy rabbits in a 2305-km2 area of sagebrush-steppe in southeastern Idaho. Given what is known of pygmy rabbit habitat, we generated and tested 3 predictions. First, habitat characteristics between areas of use and nonuse have the largest differences across several vegetal axes. Second, if rabbits are selecting on a finer scale, then there are significant differences in characteristics between actual burrow sites and surrounding areas. Third, if some burrow sites are better than others, there are differences in habitat characteristics between occupied and unoccupied or abandoned burrow sites. The support or refutation of these predictions will help determine the scale of habitat selection by pygmy rabbits and con- Fig. 1. Idaho National Engineering and Environmental tribute to the development of a predictive Laboratory (INEEL). habitat model for assessing the suitability of a given area for pygmy rabbits. on leeward sides of lava ridges and on alluvial STUDY AREA fans. The native vegetation at the INEEL consists of a shrub overstory with an understory of The study was conducted on the Idaho perennial grasses and forbs (Anderson et al. National Engineering and Environmental 1996). Laboratory (INEEL). The INEEL is located on the Snake River plain in southeastern METHODS Idaho (Fig. 1). Mean annual precipitation at Sampling Habitat Characteristics the INEEL is 22 cm, most of which falls from winter to early summer. Snowcover usually We measured habitat characteristics on persists for at least 2–3 months. Temperature plots within 5 different pygmy rabbit use cate- ranges from –9°C to 35°C. Prevailing winds gories (Fig. 2) defined as follows: over much of the INEEL come from the 1. Occupied burrow site: a 40 × 40-m plot southwest (Yanskey et al. 1966). centered on an occupied pygmy rabbit The surface of the INEEL is relatively flat burrow discovered during road sur- with some basalt flows and a few volcanic veys of the study area (Gabler 1997). buttes. The subsurface is made up of basalt 2. Unoccupied burrow site: a 40 × 40-m from past lava flows. Most of the soil is derived plot centered on an inactive burrow from older silicic volcanic and Paleozoic rocks discovered within areas of predicted from the surrounding mountains (McBride et habitat as defined by GIS analysis al. 1978). In the southern portion of the (Gabler 1997). INEEL, soils tend to be gravelly, while in the 3. Active area: a 360 × 360-m plot cen- northern portion the soil is made up of lake tered on occupied burrow sites. and aeolian deposits composed mainly of 4. Inactive area: a 360 × 360-m plot cen- unconsolidated clay, silt, and sand (Kramber et tered on an unoccupied burrow site. al. 1992). Soil depth on the INEEL typically 5. Nonuse area: a 360 × 360-m unoccu- varies from a few centimeters on the more pied plot in areas of predicted nonuse recent or exposed flows to several meters in habitat as defined by GIS analysis low-lying areas. Accumulation is also greater (Gabler 1997). 482 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Habitat characteristics were measured within 10 occupied burrow sites, unoccupied burrow sites, inactive areas, and nonuse areas. Inactive Because occupied burrow sites were clustered area such that 360 × 360-m plots around each site overlapped, only 3 active areas were defined in which habitat characteristics were measured. We measured habitat characteristics at each sample point within the study plots. In the occupied and unoccupied burrow site plots, habitat variables were measured at 17 sampling points. These variables were sampled at the point centered directly at the burrow sys- Active tem entrances (Fig. 2) and at the surrounding area 16 points formed by the 30 × 30-m grid (Fig. 2A). In the active, inactive, and nonuse area plots, the 360 × 360-m sampling grid was divided into 10-m intervals along the east– west axis. Four random points were chosen to serve as the origin of north–south oriented transects (Fig. 2), and 6 random points were sampled along each transect for a total of 24 sample points per grid (Fig. 2B.). Mean measurements of each habitat variable (N = 17, occupied and unoccupied burrow sites; N = 24, active, inactive, and nonuse areas) were used to compare the 5 use categories. We used 2 techniques to measure the habitat characteristics, point-quarter sampling (Brower et al. 1990) and point interception (Floyd and Anderson 1982). For the point- quarter method, distance to the nearest tall Fig. 2. Three scales used to measure habitat character- shrub (>50 cm) and distance to the nearest istics for pygmy rabbits (Brachylagus idahoensis) on the INEEL: (1) directly at pygmy rabbit burrow systems (A); short shrub (<50 cm) were measured in each (2) in a 30 × 30-m grid around each burrow system (A); quarter. A height of 50 cm was arbitrarily cho- and (3) in a 360 × 360-m grid encompassing the burrow sen to separate the tall shrub community from systems, both active and inactive, and in nonuse areas (B). the short shrub community. Distance measurements were then used to calculate total shrub density (TD)/100 m2 and relative den- including microbiotic crust, individual shrub sity (RD) for major shrub species in the tall species, total dead shrubs, individual grass and short categories according to Brower et al. species, total forbs, bare ground, litter, and (1990). The major shrub species measured were rock. These classes were then lumped into big sagebrush, green rabbitbrush (Chrysotham- larger coverage categories: total live shrubs, nus viscidiflorus), and gray rabbitbrush (C. total grasses, and total groundcover excluding nauseosus). In addition, we estimated shrub vascular vegetation and rocks. Relative cover- cover for the 2 height classes by measuring the age was calculated as: longest diameter of live canopy and the perpendicular diameter. These 2 measurements RCi = ni/s*36, were then used as the X and Y diameters of an ellipse to estimate cover area. where ni is the number of “hits” (Floyd and The point-frame method (Floyd and Ander- Anderson 1982) of cover type I, s is the num- son 1982) was used to estimate relative cover ber of sample points (17 or 24), and 36 is the (RC) for various vegetation and habitat classes, number of sample points within the frame. 2001] PYGMY RABBIT (BRACHYLAGUS IDAHOENSIS) HABITAT 483

Diversity for each site was also calculated RESULTS using the Shannon-Weiner index (Zar 1984). In addition to vegetation measurements, Of the 30 ANOVA tests performed for the we collected surface soil samples from 3 points different habitat characteristics measured, 13 directly next to burrow entrances and from 5 indicated significant differences among the 5 randomly selected points within each 360 × plot types (Fig. 3). Range tests detected differ- 360-m plot. Particle size analysis was conducted ences most often between nonuse areas and 3 for each sample. The hydrometer method of the 4 other plot types (occupied burrow described by Palmer and Troeh (1995), with sites, unoccupied burrow sites, and inactive modifications as described in Gabler (1997), areas). The separation between nonuse areas was used. and occupied burrow sites was seen most often. In 11 cases in which significant differences Data Analysis were detected, nonuse areas had maximum Means of habitat characteristics were com- mean values in 3 cases while occupied burrow pared among occupied burrow sites, unoccu- sites had maximum values in 6 cases. Of the pied burrow sites, active areas, inactive areas, same 11 cases, there were 8 and 2 minimum and nonuse areas. We used univariate compar- mean values for nonuse areas and occupied isons to first identify which individual habitat burrow sites, respectively. Among the 4 plot characteristics might differ among the various types of actual use (occupied burrow sites, sites, followed by multivariate analysis (princi- unoccupied burrow sites, active areas, and pal components analysis [PCA]; Morrison et inactive areas), range testing detected signifi- al. 1992) to determine if the collective compo- cant differences only for relative cover of big sition of the various sites differed. We then sagebrush, green rabbitbrush, and squirreltail compared the outcome of each to help identify grass (Sitanion hystrix; Fig. 3). Of the 4 actual which habitat variables likely were most use categories, nonuse areas differed statisti- important relative to selection of habitat by cally the least often with active areas. pygmy rabbits. Soil Texture Analysis A 1-way analysis of variance (ANOVA) was used for the univariate analysis to test the null At occupied burrow sites and active areas, hypothesis that no difference existed among mean percent sand was 81.0% and 87.5% while the 5 plot types for any of the habitat vari- mean percent clay was 5.1% and 5.0%, respec- ables. Alpha levels were adjusted for multiple tively. The mean sand component at inactive comparisons using the Bonferroni method. If areas and unoccupied burrow sites was 66.9% the null hypothesis was rejected in an ANOVA, and 69.6%, respectively, with a mean clay com- a Tukey multiple-comparisons test was per- ponent of 8.7% and 7.3%, respectively. The formed to determine differences among treat- portion of sand and clay at nonuse areas was ments. All variables measured as percentages 51.6% and 14.4%, respectively. No univariate were arcsine transformed. comparisons among treatments were made on For the PCA analysis, we generated stan- the soil variables (% sand, % silt, and % clay) dardized Z1 and Z2 principal component load- because they were all correlated. ings for each variable. After the 1st pass, vari- Principal Component Analysis ables with low loadings (<0.1) were eliminated. The remaining variables were reana- Seventeen vegetation variables were used lyzed and their loadings were used to generate in a PCA. The number of variables chosen was Z1 and Z2 scores for each plot type. Z1 and Z2 based on their presence at all or most of the 43 scores for the 5 plot types were compared plots. Variables used in the PCA included shrub with the ANOVA design described above. height, canopy cover per shrub, and total den- Separate PCA analyses and statistical tests sity for both tall and short shrub communities; were conducted for each vegetation character- relative density of big sagebrush >50 cm tall; istic and soil texture. The 2 resulting predic- relative densities of big sagebrush and green tive equations were used as the predictive rabbitbrush ≤50 cm tall; relative cover of bare model for pygmy rabbit habitat. All statistical ground, litter, forbs, dead shrubs, big sagebrush, analyses were conducted using Systat for Win- total live shrubs, and total grass; and diversity dows (Wilkinson et al. 1992). index of shrub and grass species, relative 484 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Fig. 3. Results of 1-way analysis of variance tests for those variables with significant F-values at the 5 sample site types. The means of groups within a variable that did not differ significantly in a Tukey multiple-comparison test are indicated by bars with the same fill pattern. 2001] PYGMY RABBIT (BRACHYLAGUS IDAHOENSIS) HABITAT 485 coverage of total forbs and total dead shrubs, (Table 1). These variables contributed inversely and relative coverage of big sagebrush. to the Z1 score; therefore, higher values for After calculating the first PCA, we deleted these variables resulted in a lower Z1 score. 7 of 17 variables from the analysis because Occupied burrow sites, inactive areas, and they generated small component loadings rel- unoccupied burrow sites all had significantly ative to the other variables. A PCA of the positive Z1 means. These corresponded to high remaining 10 variables defined the first 2 com- positive loadings of relative big sagebrush ponents, which accounted for 63% of variation density within the tall shrub community, rela- among plots (Table 1). Six of these variables tive coverage of total live shrubs, relative cov- also differed significantly among plot types in erage of big sagebrush, relative coverage of univariate analysis. forbs, total density of tall shrubs, and, to a From the first component, mean Z1 scores lesser extent, total density and height of the for each plot type were significantly different short shrub community (Table 1). These vari- (Table 2). This indicates a difference among ables contributed positively to the Z1 score; plot types for the collective description of therefore, higher values for these variables variables. Nonuse areas had a significantly resulted in a larger Z1 score. Because active negative Z1 mean compared to occupied bur- areas had a mean Z1 score that did not differ row sites, inactive sites, and unoccupied bur- significantly from the other 4 plot types, active row sites (Table 2). The negative Z1 mean for areas were assumed to be intermediate for Z1. nonuse areas corresponded to high negative Z2 scores were calculated from the 2nd loadings in mean canopy cover per shrub for component and a 1-way ANOVA was performed the tall shrub community, relative coverage of on mean Z2 scores. Significantly different Z2 litter, and height of the tall shrub community scores were detected only between nonuse

TABLE 1. Two 1st principal components derived from principal components analysis of 10 vegetation variables and the 2 components for soils in active pygmy rabbit (Brachylagus idahoensis) sites, occupied burrow sites, predicted pygmy rabbit sites, predicted (inactive) burrows, and nonuse sites on the Idaho National Engineering and Environmental Lab- oratory. Latent roots (eigenvalues) Vegetation Soils ______12 ______12 3.58 2.76 2.68 0.31

VARIABLESa 12 12 Sagebrush tall 0.833 0.141 Sand –0.994 0.100 Total live shrub 0.784 –0.117 Silt 0.948 –0.318 Sagebrush 0.737 0.482 Clay 0.893 0.450 Cover tall –0.655 0.331 Forbs 0.625 0.171 Litter –0.593 0.472 Total density tall 0.562 0.584 Height short 0.132 0.855 Total density short 0.190 –0.813 Height tall –0.438 0.633 Percent of total variance explained ______12 ______12 35.8 27.6 89.5 10.45

aSagebrush tall = relative density of A. tridentata >50 cm tall Total live shrub = relative coverage of live shrubs Sagebrush = relative coverage of A. tridentata Cover tall = mean cover per shrub for shrubs >50 cm tall Forbs = relative coverage of forbs Litter = relative coverage of litter Total density tall = total density of shrubs >50 cm tall Total density short = total density of shrubs ≤50 cm Height short = height of shrubs ≤50 cm Height tall = height of shrubs >50 cm tall 486 WESTERN NORTH AMERICAN NATURALIST [Volume 61

TABLE 2. Results of 1-way analysis of variance tests on Z1 and Z2 vegetation scores and Z1 soil scores among active pygmy rabbit (Brachylagus idahoensis) sites, occupied burrow sites, nonuse areas, predicted pygmy rabbit sites, and unoccupied burrows on the Idaho National Engineering and Environmental Laboratory. Means ± standard errors are listed and sample sizes are included in parentheses. Means of groups that did not differ significantly in a Tukey multiple- comparisons test are indicated by the same letter in superscript.

______Vegetation

Z1 mean FP Z2 mean FP Active areas (3) –0.92 ± 1.4ab 8.750 <0.001 0.64 ± 1.4ab 3.170 0.024 Occupied burrow sites (10) 2.96 ± 0.8a 1.41 ± 0.7a Nonuse areas (10) –3.99 ± 1.4b –2.29 ± 1.2b Inactive areas (10) 0.89 ± 0.5a 0.11 ± 0.5ab Unoccupied burrow sites (10) 0.42 ± 0.5a 0.81 ± 0.8ab

______Soil

Z1 mean FP Active areas (15) –2.54 ± 0.5a 20.92 <0.001 Occupied burrow sites (30) –1.84 ± 0.3ab Nonuse areas (50) 2.05 ± 0.3 Predicted sites (50) 0.04 ± 0.3c Unoccupied burrow sites (30) –0.32 ± 0.4bc areas and occupied burrow sites (Table 2). sites did not differ significantly from occupied Nonuse areas had a significantly negative Z2 burrow sites and active areas (Table 2). score, which mainly corresponded to negative A plot of the vegetation and soil Z1 scores loadings in total density of short shrubs (Table (Fig. 4B) produced distinct separations among 2). The significantly positive mean for occu- the 5 different plot types, with occupied bur- pied burrow sites corresponded with high row sites and nonuse areas the most separated positive loadings in height of both tall and from the other 3. short shrubs, total density of tall shrubs, and Habitat Suitability Model coverage of big sagebrush and litter (Table 2). Z1 and Z2 scores were plotted against each Based on results of the vegetation and soil other (Fig. 4A). PCA, we formulated the following equations For the PCA of the 3 soil texture variables, from the Z1 scores for use as a habitat suitabil- the 1st principal component accounted for ity model: 89.5% of overall variation between sites (Table 1). The 2nd component explained only 10.5% Vegetation Z1 = (0.833)(ST) + (0.784)(TLS) + of the variability and therefore was not ana- (0.737)(SB) + (–0.655)(CT) + (–0.593)(L) + lyzed. Loadings for the 1st component were (0.625)(F) + (0.562)(TDT) + (0.132)(HS) + used to calculate Z1 scores for occupied bur- (0.190)(TDS) + (–0.438HT) row sites, unoccupied burrow sites, active areas, inactive areas, and nonuse areas. Nonuse Soil Z1 = (–0.994)(%SAND) + areas had a significantly positive Z1 mean (0.948)(%SILT) + (0.893)(%CLAY) compared to the other 4 plot types (Table 2). This positive mean corresponded to high posi- where: tive loadings for silt and clay (Table 1). Active ST = relative density of A. tridentata >50 areas and occupied burrow sites, however, had cm tall negative means, corresponding to high sand TLS = relative coverage of live shrubs values. These 2 site types did not differ signifi- SB = relative coverage of A. tridentata cantly from each other and thus were assumed CT = mean cover per shrub for shrubs to have a similarly high sand content. The >50 cm tall inactive areas and unoccupied burrow sites L = relative coverage of litter were more intermediate, with a greater sand F = relative coverage of forbs component at the unoccupied burrow sites. TDT = total density of shrubs >50 cm tall Unlike inactive areas, unoccupied burrow TDS = total density of shrubs ≤50 cm 2001] PYGMY RABBIT (BRACHYLAGUS IDAHOENSIS) HABITAT 487

major contributing factors to the observed differences among use categories was the relative cover of big sagebrush; relative cover at use areas ranged from 3 to 10 times greater than at nonuse areas. This result corresponds with findings of others (Grinnel et al. 1930, Orr 1940, Severaid 1950, Green and Flinders 1980, Weiss and Verts 1984, Gahr 1993, Katzner and Parker 1997). Because 51–99% of pygmy rabbit diet consists of big sagebrush (Green and Flinders 1980), greater sagebrush cover would represent greater food resources for the pygmy rabbit. Greater shrub cover may also represent better protection from predators. Pygmy rabbits move more slowly and are more vulnera- ble in open habitats than are other leporids (Orr 1940) and therefore are thought to better elude predators while under a shrub canopy (Orr 1940, Wilde 1978). In addition to big sagebrush, our results indicate other vegetal variables, such as ground litter, relative coverage of forbs, and characteristics of the short (<50 cm) shrub community, also likely play a role in the suitability of an area for pygmy rabbits. The inability of multiple range tests to distinguish between active areas and nonuse areas likely is due to the small sample size (3) for active areas. A 2nd result of our study was that, within use areas, there were identifiable differences in vegetation characteristics between occupied burrow sites and the surrounding (360 × 360-m) active areas (prediction 2). It is not clear, however, whether those differences were caused by detailed selection by pygmy rabbits or modifications of the burrow area. For example, increased activity at burrows by pygmy ± Fig. 4. Plots of mean Z1 and Z2 vegetation scores ( sx–) ± rabbits may prevent new shrubs from estab- (4A) and mean Z1 vegetation and Z2 soil scores ( sx–) (4B) for the 5 pygmy rabbit use site types on the INEEL. lishing, allowing the existing shrubs to grow larger (Wilde 1978, Gahr 1993). Differential consumption of grasses and forbs by the pygmy HS = height of shrubs ≤50 cm rabbit may decrease grass biomass and allow HT = height of shrubs >50 cm tall forbs a competitive advantage (Green and Flinders 1980). It also may explain the higher DISCUSSION forb density found in this study at occupied burrow sites. However, neither we nor Weiss Results of our study suggest that pygmy and Verts (1984) detected any difference in rabbits select burrow sites based on a fairly grass cover among sites; thus, whether pygmy unique, and thus identifiable, combination of rabbits have an effect on forb and grass densi- vegetation variables and soil characteristics ties is not completely known. Pygmy rabbits (prediction 1). This was indicated by both the may in fact modify the environment surround- comparisons of individual habitat variables ing their burrows; however, indications are and PCA score differences between the 4 use that they also select for subtle vegetation dif- categories and the nonuse areas. One of the ferences for the placement of their burrows 488 WESTERN NORTH AMERICAN NATURALIST [Volume 61 within what could be considered acceptable the proposed habitat suitability models could habitat. be used to (1) determine suitable pygmy rab- Last, we also found differences between bit areas and nonuse areas and (2) possibly occupied burrow sites and unoccupied burrow rank sites within suitable areas. sites (prediction 3). This suggests that pygmy If the recommended variables for vegeta- rabbits are not only selecting specific habitat tion and soil are collected at a location and the characteristics within the “acceptable” range, values are standardized, resulting Z1 scores but they also may be making distinctions among can be compared to those from different use various usable burrow sites. Although both categories in this study (Fig. 4). For example, if occupied and unoccupied burrow sites are a Z1 vegetation score for an area is around –4 considered suitable pygmy rabbit habitat, when and the Z1 soil score is about +2, then it is pygmy rabbit populations are low, as they likely a nonuse area (Fig. 4B). If a Z1 vegeta- appeared to be during this study (Gabler 1997), tion score for an area is around +3 and the Z1 populations may shrink back into more optimal soil score is less than zero, then it has the burrow habitat. Occupied burrow sites may potential of a highly preferred burrow site represent this optimal habitat by providing (Fig. 4B). An area that has more intermediate more sagebrush cover. Unoccupied burrow sites Z1 vegetation scores and Z1 soil scores that are may represent secondary habitat that is uti- close to zero (Fig. 4B) may be usable but lized only when pygmy rabbit densities are would not be a preferred site. higher. Again, pygmy rabbits may be modify- ing the environment around their burrows. CONCLUSIONS Then, once the burrows are abandoned, the area reverts back to conditions similar to the Evidence from other studies (Weiss and surrounding areas. There is some support for Verts 1984, Washington Department of Fish this explanation, as little difference was observed and Wildlife 1995) suggests that pygmy rab- between unoccupied burrow sites and inactive bits have declined within their range over this areas. These 2 contradicting hypotheses could last century. Their numbers are also suscepti- be tested by a temporal study of burrow sys- ble to rapid declines (Janson 1946, Bradfield tems as they change from occupied to unoccu- 1975, Weiss and Verts 1984), and population pied. Such data could also give more insight recovery may be very slow (Wilde 1978). The into whether pygmy rabbit population density loss of suitable sagebrush habitat to agricul- affects habitat selection. Factors other than ture and the conversion of these lands to habitat differences could also explain why accommodate grazing appear to be extensive. pygmy rabbits abandon burrows, e.g., deple- These factors, coupled with an increase in fire tion of food resources, predator avoidance, frequency within this century, pose serious etc. Again, a temporal study of burrow systems threats to pygmy rabbit habitat (Chapman et would help clarify the possible role of these al. 1990, Gabler 1997). factors. This study has found that variables impor- If pygmy rabbits select for burrow locations tant to pygmy rabbit critical habitat are identi- on such a fine scale within suitable habitat, as fiable. Pygmy rabbits appear to select suitable inferred by this study, the implications could habitat based on a complex of vegetation and be considerable. For example, although 23.4% soil characteristics. By incorporating these vari- of the INEEL contains areas most likely to ables, the habitat suitability model provides an contain pygmy rabbit burrows within predicted excellent tool for identifying nonuse areas and habitat (Gabler 1997), a much smaller portion is a fairly good indicator of potential use areas. of those areas may actually be suitable for bur- As such, the habitat suitability model may row locations. Therefore, even slight habitat help land managers identify potential pygmy changes within these smaller areas could ren- rabbit habitat and thereby prevent further loss der some areas unsuitable for burrow con- and degradation of pygmy rabbit habitat when struction. making land-use decisions. This model has great potential for use throughout the pygmy Habitat Suitability Model rabbit’s range to aid in the conservation of Given the measured vegetation and soil dif- pygmy rabbit habitat and, ultimately, to aid in ferences among the different use categories, the conservation of this species. 2001] PYGMY RABBIT (BRACHYLAGUS IDAHOENSIS) HABITAT 489

ACKNOWLEDGMENTS JANSON, R.G. 1946. A survey of the native rabbits of Utah with reference to their classification, distribution, The Environmental Science and Research life histories, and ecology. Master’s thesis, Utah State University, Logan. 103 pp. Foundation (ESRF), under contract #DE- KATZNER, T.E. 1994. Winter ecology of the pygmy rabbit AC07-94ID13268 with the U.S. Department (Brachylagus idahoensis) in Wyoming. Master’s the- of Energy, Idaho Field Office, and the U.S. sis, University of Wyoming, Laramie. 125 pp. Bureau of Land Management Cost Share pro- KATZNER, T.E., AND K.L. PARKER. 1997. Vegetative characteristics and size of home ranges used by pygmy rab- gram, provided funding for this project. We bits (Brachylagus idahoensis) during winter. Journal thank O.D. Markham, R. Morris, and R. War- of Mammalogy 78:1063–1072. ren, and Tim Reynolds from the ESRF for KEHNE, J. 1991. Sagebrush Flat pygmy rabbit project— their help. Many thanks are due to the ESRF soils report. Unpublished report, Washington Depart- technicians for their assistance with fieldwork. ment of Wildlife, Olympia. 119 pp. KRAMBER, W.J., R.C. ROPE, J. ANDERSON, J. GLENNON, AND Thanks are also due to M. Gabler and D. John- A. MORSE. 1992. Producing a vegetation map of the son for their assistance with various aspects of Idaho National Engineering Lab using Landsat The- this project. matic Mapper data. ASPRS Technical Papers 1: 217–226. MCBRIDE, R., N.R. FRENCH, A.H. DAHL, AND J.E. DETMER. LITERATURE CITED 1978. Vegetation types and surface soils of the Idaho National Engineering Laboratory Site. IDO-12084, ANDERSON, J.E., K.T. RUPPEL, J.M. GLENNON, K.E. HOLTE, National Technical Information Service, Springfield, AND R.C. ROPE. 1996. Plant communities, ethno- VA. 29 pp. ecology, and flora of the Idaho National Engineering MORRISON, M.L., B.G. MARCOT, AND R.W. MANNAN. 1992. Laboratory. Environmental Science & Research Wildlife-habitat relationships. Concepts and applica- Foundation Report Series, No. 005, Lockheed Tech- tions. University of Wisconsin Press, Madison. 343 nologies Company, Idaho Falls, ID. 111 pp. pp. BRADFIELD, T.D. 1975. On the behavior and ecology of the ORR, R.T. 1940. The rabbits of California. Occasional Papers pygmy rabbit Sylvilagus idahoensis. Master’s thesis, of the California Academy of Sciences 19:1–227. Idaho State University, Pocatello. 43 pp. PALMER, R.G., AND F.R. TROEH. 1995. Introductory soil BROWER, J.E., J.H. ZAR, AND C.N. VON ENDE. 1990. Field science laboratory manual. 3rd edition. Oxford Uni- and laboratory methods for general ecology. Wm. C. versity Press, New York. Brown Company Publishers, Dubuque, IA. 237 pp. SEVERAID, J.H. 1950. The pygmy rabbit (Sylvilagus idaho- CHAPMAN, J.A., J.E.C. FLUX, A.T. SMITH, D.J. BELL, G.G. ensis) in Mono County, California. Journal of Mam- CEBALLOS, K.R. DIXON, F.C. DOBLER, ET AL. 1990. malogy 31:1–4. Conservation action needed for rabbits, hares, and WASHINGTON DEPARTMENT OF FISH AND WILDLIFE. 1995. pikas. Pages 154–167 in J.A. Chapman and J.E.C. Washington state recovery plan for the pygmy rab- Flux, editors, Rabbits, hares, and pikas: status survey bit. Wildlife Management Program, Washington and conservation action plan. IUCN/SSC Lago- Department of Fish and Wildlife, Olympia. 73 pp. morph Specialist Group, Gland, Switzerland. WEISS, N.T., AND B.J. VERTS. 1984. Habitat and distribu- FLOYD, D.A., AND J.E. ANDERSON. 1982. A new point inter- tion of pygmy rabbits (Sylvilagus idahoensis) in Ore- ception frame for estimating cover of vegetation. gon. Great Basin Naturalist 44:563–571. Vegetatio 50:185–186. WHITE, S.M., J.T. FLINDERS, AND B.L. WELCH. 1982. Pref- GABLER, K.I. 1997. Distribution and habitat requirements erence of pygmy rabbits (Brachylagus idahoensis) for of the pygmy rabbit (Brachylagus idahoensis) on the various populations of big sagebrush (Artemisia tri- Idaho National Engineering and Environmental dentata). Journal of Range Management 35:724–726. Laboratory. Master’s thesis, Idaho State University, WILDE, D.B. 1978. A population analysis of the pygmy Pocatello. 117 pp. rabbit (Sylvilagus idahoensis) on the INEL site. Doc- GAHR, M.L. 1993. Natural history, burrow habitat and use, toral dissertation, Idaho State University, Pocatello. and home range of the pygmy rabbit (Brachylagus 172 pp. idahoensis) of Sagebrush Flat, Washington. Master’s WILKINSON, L., M. HILL, J.P. WELNA, AND G.K. BIRKEN- thesis, University of Washington, Seattle. 125 pp. BEUEL. 1992. Systat for Windows: statistics. Version GREEN, J.S. 1978. Pygmy rabbit and coyote investigations 5 edition. Evanston, IL. 750 pp. in southeastern Idaho. Doctoral dissertation, Brigham YANSKEY, G.R., E.H. MARKEE, JR., AND A.P. RICHTER. 1966. Young University, Provo, UT. 88 pp. Climatology of the National Reactor Testing Station. GREEN, J.S., AND J.T. FLINDERS. 1980. Habitat and dietary IDO-12048, National Technical Information Service, relationships of the pygmy rabbit. Journal of Range Springfield, VA. 184 pp. Management 33:136–142. ZAR, J.H. 1984. Biostatistical analysis. 2nd edition. Pren- GRINNELL, J., J. DIXON, AND J.M. LINSDALE. 1930. Verte- tice Hall, Englewood Cliffs, NJ. brate natural history of a section of northern Califor- nia through the Lassen Peak region. University of Received 22 December 1999 California Publications in Zoology 35:1–594. Accepted 11 April 2000 Western North American Naturalist 61(4), © 2001, pp. 490–494

MULE DEER FORAGING PREFERENCE AMONG FIVE SAGEBRUSH (ARTEMISIA L.) TAXA

Carl L. Wambolt1

ABSTRACT.—The hypothesis that sagebrush taxa are equally utilized by mule deer (Odocoileus hemionus hemionus) on winter range was tested. Five taxa were studied for 10 years at 2 locations. The taxa were Artemisia tridentata ssp. tridentata (basin big sagebrush), A. t. ssp. wyomingensis (Wyoming big sagebrush), A. t. ssp. vaseyana (mountain big sagebrush), A. tripartita ssp. tripartita (tall threetip sagebrush), and A. arbuscula ssp. arbuscula (low sagebrush). Possible mule deer preferences were determined each year individually for the 2 sites. Utilization was high enough to conclude all taxa are important forage, but not excessive enough to mask preference. Artemisia tridentata ssp. vaseyana (34.4%) and A. arbuscula ssp. arbuscula (35.6%) were preferred over A. t. ssp. wyomingensis (10.9%) and A. t. ssp. tridentata (6.8%) at the Ashbough site. At the Scudder site there were few differences in preference for A. t. ssp. vaseyana (32.1%), A. t. ssp. wyomingensis (28.8%), and A. tripartita ssp. tripartita (32.0%). While ungulates often demonstrate a preference among taxa, all sagebrush taxa are a potentially valuable forage source.

Key words: Artemisia, forage preference, Montana, mule deer, sagebrush.

Beetle (1960) estimated that sagebrush for sagebrush taxa under natural conditions. (Artemisia L.) taxa occur on as much as 109 Several differences also exist. This study does million ha in the western United States. While not have the complication of elk browsing developments, management, and fire in the along with mule deer. Also, the sagebrush taxa region have considerably reduced this area, are somewhat different. This study examines the importance of sagebrush taxa and the com- Artemisia tridentata Nutt. ssp. tridentata (basin munities in which they occur is obvious for big sagebrush), A. t. ssp. wyomingensis Beetle natural resource management (Wambolt 1998). and Young (Wyoming big sagebrush), A. t. ssp. Wambolt (1996) stated, “Consideration of sage- vaseyana [Rydb.] Beetle (mountain big sagebrush ecology, including forage values, is a brush), A. tripartita Rydb. ssp. tripartita (tall necessity for judicious range management.” threetip sagebrush), and A. arbuscula Nutt. Several studies have reported preference ssp. arbuscula (low sagebrush). Finally, the 2 ratings of herbivores for various sagebrush study sites are separated by a large distance. taxa (Scholl et al. 1977, Sheehy and Winward The present study was conducted 200 km by 1981, Welch et al. 1981, 1983, Welch and air from the Wambolt (1996) investigation. In McArthur 1986, Wambolt 1996). With the ex- the present study the hypothesis was tested ception of Wambolt (1996), these studies based that sagebrush taxa are utilized equally as for- their conclusions on short-term projects with small numbers of tame animals under unnat- age by mule deer on winter range. The test was ural circumstances. Wambolt (1996) attempted made each year over a 10-year period under to avoid anomalies that might occur under the natural conditions for 5 taxa at 2 study sites. previously described conditions by conducting a 10-year study under natural conditions to METHODS determine preferences of wild mule deer (Odo- Study Area coileus hemionus hemionus) and elk (Cervus elaphus nelsoni) for 4 sagebrush taxa. The Ashbough and Scudder study areas are The purpose of this study was similar to located approximately 86 km apart on Bureau that of Wambolt (1996). Similarities included of Land Management land in Beaverhead conducting the study over a 10-year period to County, Montana. The Ashbough site (44°47′N, determine the preference of wild mule deer 112°38′W) is 50 km south of Dillon, and the

1Department of Animal and Range Sciences, Montana State University, Bozeman, MT 59717.

490 2001] MULE DEER SAGEBRUSH PREFERENCE 491

Scudder site (45°18′N, 113°5′W) is 40 km west TABLE 1. Percent canopy cover of 5 sagebrush (Artemisia) of Dillon. Both are at elevations of 1980 m, taxa at 2 study sites in 1982 at the beginning of the study. Data were determined by the line-intercept method on which is typical of elevations used by mule deer eight 30-m lines established along each of the belt tran- on winter range in southwestern Montana. sects used to sample sagebrush preference. Mule deer were the only ungulate of any sig- Canopy cover (%) nificance to use these sites during the winters ______of the study. Average annual precipitation at Taxon Ashbough Scudder the sites is approximately 380 mm, with half A. tridentata ssp. vaseyana 7.2 7.5 received as snow, although the peak occurs in A. tridentata spp. wyomingensis 7.9 8.0 May and June. Vegetative composition on both A. tridentata spp. tridentata 6.9 — sites is dominated by an overstory of sage- A. tripartita ssp. tripartita — 8.3 brush taxa. At the Ashbough site Artemisia tri- A. arbuscula ssp. arbuscula 6.4 — TOTAL ARTEMISIA 28.4 23.8 dentata ssp. vaseyana, A. t. ssp. wyomingensis, A. t. ssp. tridentata, and A. arbuscula ssp. arbuscula comprise the overstory. At the Scudder site A. t. ssp. vaseyana, A. t. ssp. wyomingensis, that the number of browsed leaders and total and A. tripartita ssp. tripartita are dominants. utilization obtained by measuring leader lengths The understory at both sites is dominated by were highly correlated (r = 0.94, P ≤ 0.0001) Agropyron spicatum [Rydb.] Scribn. (blue- for Purshia tridentata [Pursh] D.C. (bitter- bunch wheatgrass) and Festuca idahoensis brush). The percentage of P. tridentata leaders Elmer (Idaho fescue). The sagebrush taxa at Guenther (1989) found removed by browsing both study sites are present in nearly equal at 18 locations was adequately predicted (±10%) quantities (Table 1) and intermixed as a nat- by the proportion of leaders browsed. Because ural cafeteria for wintering mule deer due to a sagebrush leaders are shorter than those of P. mosaic of microsites. tridentata, it is logical that results between the 2 methods for sagebrush would be similar. Sampling and Analysis Preceding the winter use period, each The preference of mule deer among the autumn between 1982 and 1991, a total of sagebrush taxa for use as winter forage was approximately 1350 and 1400 available leaders evaluated on the 2 deer winter ranges. The were tagged on 103 and 131 plants at the Ash- portions of the 2 winter ranges measured were bough and Scudder study sites, respectively. confined to areas of 30 × 60 m that comprised Sagebrush plants were each divided into seg- each natural cafeteria. Sampling was conducted ments from which randomly selected leaders by establishing eight 1 × 30-m belt transects were tagged. This procedure insured all por- located parallel to each other at 8.6-m inter- tions of the sagebrush crown were sampled. vals within each browse cafeteria. All sage- The tagged leaders were reexamined each brush plants rooted within the belt transects spring to determine the percentage of total were permanently located and identified by leaders browsed the preceding winter. taxon. Thus, the same plants were measured The statistical analysis followed Wambolt for winter utilization throughout the study. (1996) and is repeated here for explanatory Sagebrush plants that have been browsed purposes: develop a very twisted growth form. This was the situation on the 2 study sites, where a rela- A 1-way ANOVA with taxon as the factor was tively high level of browsing had occurred in conducted each year individually for the 2 the past. Therefore, it was determined that an sites. This avoided year and location effects adequate sample of leaders measured for that could confuse the results. The observa- length, before and after browsing, could not tions in the ANOVA are a transformation on be obtained. Because the study purpose was the proportion of utilized leaders. Because to determine relative preference of mule deer the proportions are based on relatively small sample sizes, a variance stabilizing arcsin for the sagebrush taxa, it was decided to com- transformation was used (Snedecor and Coch- pare the proportion of leaders browsed among ran 1980, Steel and Torrie 1980). This trans- the taxa at each site. This procedure was not formation is not used to remove inequalities affected by the gnarled sagebrush crowns. A in variance, but is used when the variation is previous study (Guenther 1989) determined purely binomial, often arising from unequal 492 WESTERN NORTH AMERICAN NATURALIST [Volume 61

denominators. The Least Significant Differ- yana and A. t. ssp. wyomingensis. In this case ence (LSD) method (P ≤ 0.05) protected by a the 2 taxa were used at similar average levels prior F-test (P ≤ 0.05) was used for compar- of 32.1% and 28.8% during the entire study for ing treatment means (Snedecor and Cochran A. t. ssp. vaseyana and A. t. ssp. wyomingensis, 1980). respectively (Table 3). The range of utilization levels of 9–59% for A. t. ssp. vaseyana and RESULTS AND DISCUSSION 10–52% for A. t. ssp. wyomingensis were also Browsing levels (Tables 2, 3) at both sites similar. Statistically over the study, the 2 taxa were in the utilization range considered desir- were browsed equally at the Scudder site. During the 4 winters a statistical difference (P able to detect possible foraging preferences ≤ (Wambolt 1996). Utilization was high enough 0.05) was found; in 2 winters A. t. ssp. vase- to conclude that the taxa are important forage yana was preferred over A. t. ssp. wyomingen- sources at the 2 sites. At the same time utiliza- sis, while in the other 2 winters the opposite tion was not so heavy that possible preferences was true. might have been masked. During periods of These data for Artemisia tripartita ssp. tri- deep snow accumulation, the availability of partita provide the 1st reported utilization lev- taxa might vary from normal and mask prefer- els for the taxon in a forage preference trial. ences. In the same manner, prolonged periods Artemisia tripartita ssp. tripartita received an of severe temperatures, snow accumulation, average utilization of 32.0% over the study. and snow crusting might necessitate higher Interestingly, over that period A. t. ssp. tripar- than normal consumption of all taxa that would tita had the greatest range of utilization val- have the same result. Because the study was ues, from 8% to 74%. During the 10 years, conducted during a decade of below-average browsing of A. t. ssp. tripartita significantly (P ≤ snowfall, no severe winters occurred that would 0.05) exceeded that of A. tridentata ssp. have minimized preference differences. The 2 vaseyana 1 winter and was browsed less in 2 study sites should be considered individually winters. Compared to A. tridentata ssp. wyo- because the sagebrush taxa present varied mingensis, A. t. ssp. tripartita was twice used between the 2 locations. more and twice used less. Overall, little differ- The 4 taxa at the Ashbough site fell distinctly ence was found among utilization levels for into 1 of 2 preference classes for mule deer the 3 taxa at the Scudder site; all taxa were (Table 2). Artemisia tridentata ssp. vaseyana browsed at similar levels to the preferred taxa and A. arbuscula ssp. arbuscula were clearly at the Ashbough site. preferred over A. t. ssp. wyomingensis and A. Perhaps the greatest anomaly found in this t. ssp. tridentata. Artemisia tridentata ssp. vase- study when results are considered with previ- yana and A. a. ssp. arbuscula, with average ous investigations (Scholl et al. 1977, Sheehy utilization levels of 34.4% and 35.6%, respec- and Winward 1981, Welch et al. 1981, 1983, tively, were statistically the same 8 of 10 win- Welch and McArthur 1986, Wambolt 1996) of ters. The 2 winters that the taxa were not uti- herbivore preferences for sagebrush taxa would lized equally were 1984–85, when browsing be the acceptance by mule deer of Artemisia on A. a. ssp. arbuscula exceeded that on A. t. tridentata ssp. wyomingensis at the Scudder ssp. vaseyana, and 1991–92, when just the site. If the previous studies had a single point opposite occurred. of agreement, it was that A. t. ssp. vaseyana Artemisia tridentata ssp. wyomingensis, with was the preferred taxon. At the Scudder loca- a 10-year average utilization of 10.9%, and A. tion browsing on the 2 taxa was essentially the t. ssp. tridentata, with 6.8%, were distinctly less same throughout the study. preferred than the other 2 taxa. On only 3 occa- This finding is also made relevant by the sions (of 20 opportunities: 2 taxa × 10 years), fact that the 3rd taxon at Scudder was Arte- all during winters of light utilization, did the misia tripartita ssp. tripartita. Indeed, this utilization level for either one of these taxa taxon was preferred equally to the other 2 taxa equal the browsing received by either A. t. ssp. by mule deer at the site. That implies that vaseyana or A. arbuscula ssp. arbuscula. there was no shortage of preferred forage Mule deer browsing at the Scudder site pro- available that would lead to higher than usual vided somewhat different results with the com- consumption of A. tridentata ssp. wyomingen- parison between Artemisia tridentata ssp. vase- sis. Also, the overall utilization levels were not 2001] MULE DEER SAGEBRUSH PREFERENCE 493

TABLE 2. Percentage of sagebrush leaders utilized during winter by taxon and year at the Ashbough study site. Means among taxa differ (P ≤ 0.05) within date when followed by a different letter. Artemisia Artemisia Artemisia Artemisia tridentata tridentata ssp. tridentata arbuscula Year ssp. vaseyana wyomingensis ssp. tridentata ssp. arbuscula ------% ------1982–83 46a 8b 2b 60a 1983–84 47a 26b 19b 43a 1984–85 15b 3c 15b 38a 1985–86 44a 16b 11b 58a 1986–87 25a 4b 0b 18a 1987–88 34a 9b 2b 31a 1988–89 28a 13b 10b 38a 1989–90 19a 9a 1b 19a 1990–91 25a 5b 3b 32a 1991–92 61a 16b 5b 19b

TABLE 3. Percentage of sagebrush leaders utilized during winter by taxon and year at the Scudder study site. Means among taxa differ (P ≤ 0.05) within date when followed by a different letter. Artemisia Artemisia Artemisia tridentata tridentata ssp. tripartita Year ssp. vaseyana wyomingensis ssp. tripartita ------% ------1982–83 59a 36b 30b 1983–84 39a 38a 43a 1984–85 10b 24a 17ab 1985–86 9b 22a 8b 1986–87 24a 23a 27a 1987–88 48a 52a 52a 1988–89 23a 25a 10b 1989–90 40a 10b 42a 1990–91 40b 38b 74a 1991–92 29a 20a 17a

high enough to result in less selectivity by The variation in desirability of Artemisia mule deer as found in another Montana study tridentata ssp. wyomingensis as mule deer for- (Wambolt 1996). Utilization levels were often age found at different locations emphasizes considerably higher on the northern Yellow- the importance of careful interpretation of stone winter range (Wambolt 1996), thereby limited information. This may be relevant to tending to equalize the intakes of taxa during A. tripartita ssp. tripartita studied at the Scud- years of high ungulate populations or severe der site. My data show that this taxon was uti- periods of weather. lized by mule deer to the same degree as A. Other investigators have determined that tridentata ssp. vaseyana, a highly preferred ungulate acceptance of a taxon for browsing taxon in all studies. However, it is reasonable varies among accessions (Welch et al. 1981, to assume that this might vary at other loca- 1983, Welch and McArthur 1986, Welch et al. tions, just as preference for A. t. ssp. wyomin- 1994). Thus, it seems likely that under natural gensis has been found to vary between loca- conditions a given taxon might be more desir- tions. able to mule deer at one location than another. Both study sites are at elevations that in It appears that was the situation at the Scud- western Montana offer prime wintering oppor- der site when results there are compared to tunities for mule deer. They are high enough the Ashbough site and results of Wambolt that conifer forest and topographical security (1996). and thermal cover are usually not far away. At 494 WESTERN NORTH AMERICAN NATURALIST [Volume 61 the same time they are low enough that snow bility of seven Artemisia taxa to mule deer and sheep. depth usually does not significantly affect deer Journal of Range Management 34:397–399. SNEDECOR, G.W., AND W. G. C OCHRAN. 1980. Statistical foraging. This is also the topographic position methods. Iowa State University Press, Ames. that generally has the most diversity of sage- STEEL, R.G.D., AND J.H. TORRIE. 1980. Principles and pro- brush taxa in western Montana. Thus, mule cedures of statistics. McGraw-Hill, New York. deer often have foraging options among sage- STRIBY, K.D., C.L. WAMBOLT, R.G. KELSEY, AND K.M. brush taxa on their western Montana winter HAVSTAD. 1987. Crude terpenoid influence on in vitro digestibility of sagebrush. Journal of Range ranges. Management 40:244–248. These results should improve understanding WAMBOLT, C.L. 1996. Mule deer and elk foraging prefer- of sagebrush-ungulate relationships. Observed ence for 4 sagebrush taxa. Journal of Range Manage- differences in foraging patterns are often ques- ment 49:499–503. ______. 1998. Sagebrush-ungulate relationships on Yel- tioned in making management decisions. This lowstone’s northern range. Wildlife Society Bulletin study and the others discussed lead to the 26:429–437. general conclusion that while ungulates will WELCH, B.L., AND E.D. MCARTHUR. 1979. Feasibility of demonstrate a preference among taxa, all sage- improving big sagebrush (Artemisia tridentata) for brush taxa are a potentially valuable forage use on mule deer winter ranges. Pages 451–457 in J.R. Goodin and D.K. Northington, editors, Arid land source (Welch and McArthur 1979, Striby et plant resources. Texas Tech University, Lubbock. al. 1987, Welch and Wagstaff 1992, Wambolt ______. 1986. Wintering mule deer preference for 21 1998). accessions of big sagebrush. Great Basin Naturalist 46:281–286. WELCH, B.L., AND F. J . W AGSTAFF. 1992. ‘Hobble Creek’ LITERATURE CITED big sagebrush vs. antelope bitterbrush as a winter forage. Journal of Range Management 45:140–142. BEETLE, A.A. 1960. A study of sagebrush, the section Tri- WELCH, B.L., S.F. BRIGGS, AND S.A. YOUNG. 1994. Pine dentatae of Artemisia. Bulletin 368, University of Valley Ridge source: a superior selected germplasm Wyoming Agricultural Experiment Station, Laramie. of black sagebrush. USDA Forest Service, Research GUENTHER, G.E. 1989. Ecological relationships of bitter- Paper INT-RP-474. brush communities on the Mount Haggin Wildlife WELCH, B.L., E.D. MCARTHUR, AND J.N. DAVIS. 1981. Dif- Management Area. Unpublished master’s thesis, ferential preference of wintering mule deer for Montana State University, Bozeman. accessions of big sagebrush and for black sagebrush. SCHOLL, J.P., R.G. KELSEY, AND F. S HAFIZADEH. 1977. Journal of Range Management 34:409–411. Involvement of volatile compounds of Artemisia in ______. 1983. Mule deer preference and monoterpenoids browse preference by mule deer. Biochemical Sys- (essential oils). Journal of Range Management 36: tematics and Ecology 5:291–295. 485–487. SHEEHY, D.P., AND A.H. WINWARD. 1981. Relative palata- 2001] MULE DEER SAGEBRUSH PREFERENCE 495

Received 25 October 1999

MORTALITY OF THE ENDANGERED WRIGHT FISHHOOK CACTUS (SCLEROCACTUS WRIGHTIAE) BY AN OPUNTIA-BORER BEETLE (CERAMBYCIDAE: MONEILEMA SEMIPUNCTATUM)

Ronald J. Kass1

Key words: demographic monitoring, beetle, mortality, population decline.

Sclerocactus wrightiae Benson (Wright fish- punctatum infesting the genus Sclerocactus. hook) is a small, white-flowered barrel cactus Six species of Moneilema are recognized in endemic to the San Rafael Swell in south cen- North America and restrict feeding to the tral Utah. It grows in salt desert shrub and Cactaceae, preferring the genus Opuntia as a pinyon juniper communities at 1460 to 1865 host (Crosswhite and Crosswhite 1985). Mon- m (Welsh et al. 1993). Populations are typically eilema semipunctatum’s range is the Great small, consisting of 50–100 individuals, and Basin in Nevada and western Utah, south to restricted to fine-textured soils derived from northern Baja California, and east to the Col- the Mancos Shale, Morrison, Summerville, orado Plateau in Utah, southwestern Colorado, Curtis, Entrada, and Carmel formations. northwestern New Mexico, and northern Ari- Sclerocactus wrightiae was listed as an zona (Linsley and Chemsak 1984). endangered species by the U.S. Fish and Linsley and Chemsak (1984) discussed feed- Wildlife Service on 11 October 1979 (USFWS ing and mating habits of the genus Moneilema. 1979). It was listed based on its narrow distri- Adult beetles are large, black, nocturnal, and bution, small population size, and threats from flightless. They become active at dusk and mining and natural gas development (Mutz et climb up and down the cactus stem, feeding at al. 1985). Past inventories (Welsh 1980, Neese the base or in the stem crown. Mating occurs 1987, Kass 1989) have extended its distribu- at night on top of the plant and may continue tion from approximately Ferron in Emery throughout the night. Females choose a suit- County southwest to Hanksville in Wayne able ovipositor site near the base of the plant County, Utah (Fig. 1). and deposit eggs. After hatching, the larvae Long-term demographic monitoring plots attempt to enter the plant, and during the sec- were established throughout the range of S. ond or third instar the larvae begin to tunnel wrightiae to assess various aspects of its life into the plant. A pupal cell is constructed in history and reproductive biology. During the the fall either in soil or within hollowed-out 1993–2000 field seasons, a small, white beetle stems. larva was discovered infesting cacti at all 3 During the 1993–2000 monitoring period, plots. Upon close examination, cacti were M. semipunctatum accounted for 23% of the often found with external chew marks, pro- combined mortality at all 3 plots. Similar per- nounced constrictions between growth seg- centages have been observed throughout the ments, and a spongy and chlorotic appearance. remaining range of S. wrightiae (Kass personal These abnormalities led to decreased vigor, observation). Size class analysis of beetle- lower fecundity, and eventual death in the killed stems (n = 25) indicated 16% mortality same or subsequent years. Beetle larvae were in size class 2 (2–4 cm wide), 44% mortality in captured and reared in captivity and later size class 3 (4–9 cm wide), and 40% mortality identified as the Opuntia-borer beetle (Ceram- in size class 4 (cacti >9 cm wide; Table 1). bycidae: Moneilema semipunctatum LeConte). Beetles did not infest the smallest size class This is the first published report of M. semi- (0–2 cm wide), and so mortality in size class 1

1Intermountain Ecosystems, L.C., 270 East 1230 North, Springville, UT 84663.

495 496 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Fig. 1. Distribution of Sclerocactus wrightiae. was generally the result of dessication or cattle (BLM personal communication). It is possible trampling. that continued beetle mortality may cause a Other mortality sources recorded during shift in population structure: larger cacti with monitoring include blister beetles (Epicauta higher reproductive rates may be replaced sp.), Ord kangaroo rats (Dipodomys ordii), and with smaller cacti with lower reproductive rates. white-tailed antelope ground squirrels (Ammo- Kass (2000) reports a mortality-to-recruitment spermophilus leucurus). Reports from other ratio of approximately 2.5 to 1 since monitor- researchers indicate Moneilema infestations ing began in 1993. Menges (1991) recom- on the federally listed Uinta Basin hookless mends that recruitment be greater than mor- cactus (Sclerocactus glaucus) in Utah (USFWS tality per year to maintain population viability. personal communication) and Mesa Verde cac- These observed mortality rates, coupled with tus (Sclerocactus mesae-verde) in Colorado and increasing anthropogenic threats of commer- New Mexico (Naumann 1989). cial collecting and habitat degradation, indi- Moneilema semipunctatum appears to infest cate a slow decline for S. wrightiae. larger individuals possibly because large individuals are better host plants. Additional de- I thank Steve Wood (Brigham Young Uni- clines in larger individuals have been observed versity) for identifying the beetles and extend range-wide since inventories for S. wrightiae appreciation to Ron Bolander and Lori Arm- began in 1986. Some of these declines are the strong of the Bureau of Land Management for results of amateur and commercial collecting funding the project. 2001] NOTES 497

TABLE 1. Size-specific mortality by beetles. Values indi- Falk and K.E. Holsingers, editors, Genetics and con- cate numbers and percent of cacti killed by beetles com- servation of rare plants. Oxford University Press, pared to overall percent size class distribution. New York. MUTZ, K.M., E. NEESE, J.L. MILLER, AND G.R. JACOBS. Size class N % beetle killed % size class 1985. Wright fishhook cactus (Sclerocactus wrightiae) 1008Benson recovery plan. Prepared in cooperation with 2 4 16 23 the Wright Fishhook Cactus Recovery Committee, 3114463U.S. Fish and Wildlife Service, Denver, CO. 27 pp. 410406NAUMANN, T. 1989. Population biology of Sclerocactus mesae-verdae: performance report. Unpublished re- TOTAL 25 100 100 port submitted to Colorado Natural Areas Program, Boulder, CO. 21 pp. NEESE, E. 1987. Habitat inventory of Sclerocactus wrightiae and other associated sensitive species. Final report LITERATURE CITED to Bureau of Land Management, Richfield District Office, Richfield, UT. 116 pp. CROSSWHITE, C.D., AND F. S . C ROSSWHITE. 1985. Tricho- U.S. FISH AND WILDLIFE SERVICE. 1979. Determination cereus as a potential nursery crop in southern Ari- that Sclerocactus wrightiae is an endangered species. zona, with discussion of the Opuntia borer (Ceram- Federal Register 44: 58868. bycidae: Moneilema gigas) as a serious threat to its WELSH, S.L. 1980. Survey for Sclerocactus wrightiae and cultivation. Desert Plants 7:195–203. Townsendia aprica. Reports 1–4. Unpublished report KASS, R.J. 1989. Habitat inventory of threatened, endan- for Consolidated Coal Company, Englewood, CO. gered, and candidate plant species in the San Rafael WELSH, S.L., N.D. ATWOOD, S. GOODRICH, AND L.C. HIG- Swell, UT. Final report submitted to the Bureau of GINS. 1993. A Utah flora. 2nd edition. Jones Endow- Land Management, Salt Lake City, UT. 80 pp. ment Fund, Monte L. Bean Museum, Brigham Young LINSLEY, E.G., AND J.A. CHEMSAK. 1984. The Cerambyci- University, Provo, UT. 986 pp. dae of North America, part VII. No. 1: Taxonomy and classification of the subfamily Lamiinae, tribes Received 2 June 2000 Parmenini through Acanthoderini. University of Cal- Accepted 7 August 2000 ifornia Publications in Entomology 102:1–258. MENGES, E. 1991. The application of minimum viable population theory to plants. Pages 45–61 in D.A. Western North American Naturalist 61(4), © 2001, pp. 498–500

SEXUAL DIMORPHISM AND BODY TEMPERATURES OF SCELOPORUS SINIFERUS FROM GUERRERO, MÉXICO

Julio A. Lemos-Espinal1, Geoffrey R. Smith2,4, and Royce E. Ballinger3

Key words: lizards, sexual dimorphism, Sceloporus, México, thermal ecology.

The biology of temperate North American observed) and substrate temperature (Ts; lizards of the genus Sceloporus is relatively shaded thermometer touching substrate where well known. We know substantially less about individual first observed). We also made vari- the majority of Mexican and Latin American ous morphological measurements to analyze Sceloporus species. Indeed, for too many sexual dimorphism. We measured head width species we know only what has been pub- (HW; at the widest point), head length (HL; lished in the original descriptions. Recently, from anterior edge of ear to tip of snout), and herpetologists have begun to recognize the femur length (FL; from knee to middle of importance of studying the biology of tropical pelvic region) using calipers. reptiles (see Vitt and Zani 1996). In some Mean SVL was 51.2 ± 0.8 mm (N = 56; cases studies on tropical species have obtained range = 38–62 mm). For all morphometric results counter to those obtained in temperate variables, the relationship with SVL was highly systems that were sometimes thought to per- significant (all r2 > 0.80; P < 0.0001); thus, we tain to all reptiles (e.g., Shine and Madsen used ANCOVA to analyze for sexual dimor- 1996). Thus, it is important to study the gen- phism (on log-transformed data; after assump- eral biology and ecology of previously unstud- tions checked). ied species, especially those from tropical or Males were larger on average than females subtropical regions. Such information hope- (Table 1; df = 59, t = 4.37, P < 0.0001). Males fully can serve as the basis of future syntheses had relatively wider heads than females (Fig. on the biology of lizards. This note concerns 1; Table 1; ANCOVA with SVL as covariate: the sexual dimorphism and body temperature F1,58 = 7.6, P = 0.008). The interaction be- of Sceloporus siniferus, a relatively unstudied tween sex and the covariate was not significant species from the seasonal semiarid tropics of and was not included in the final model. México. Males and females did not differ in the length The study population was located in a trop- of their heads after the effects of SVL were ical deciduous forest located near the Bahia removed using ANCOVA (Table 1; ANCOVA Papanoa (Km 161, Highway Mex 200 Acapulco- with SVL as covariate: F1,58 = 1.61, P = 0.21). Zihuatanejo: 17°2′0.4″N, 101°3′0.0″W). Lizards The interaction between sex and the covariate were collected by rubber band during May was not significant and was not included in 1996. We measured snout-vent length (SVL; the final model. The length of a male’s femur to nearest mm) in the field. In addition, we was, on average, the same as the length of a ° took body temperatures (Tb; nearest 0.1 C) female’s femur (Table 1; ANCOVA with SVL with a quick-reading cloacal thermometer as covariate: F1,58 = 0.47, P = 0.50). The inter- immediately upon capture. We also measured action between sex and the covariate was not air temperature (Ta; shaded thermometer 1 significant and was not included in the final cm above substrate where individual first model.

1Laboratorio de Ecologia, Unidad de Biología, Tecnología y Prototipos, Escuela Nacional de Estudios Profesionales Iztacala, Av. de los Barrios s/n, Los Reyes Iztacala, Estado de México, 54090 México. 2Department of Biology, William Jewell College, Liberty, MO 64068 USA. 3School of Biological Sciences, University of Nebraska, Lincoln, NE 68588 USA. 4Corresponding author. Present address: Department of Biology, Denison University, Granville, OH 43023 USA.

498 2001] NOTES 499

TABLE 1. Measurements (mm) of SVL and measurements (mm) of head width, head length, and femur length corrected for SVL of male (N = 36) and female (N = 25) Sceloporus siniferus from Guerrero, México. Least squares ± means are given 1sx–. Male Female SVL 52.8 ± 0.9 47.1 ± 0.8 Head width 10.13 ± 0.09 9.73 ± 0.11 Head length 11.54 ± 0.09 11.36 ± 0.11 Femur length 14.56 ± 0.12 14.70 ± 0.14

± ° Mean Tb was 36.2 0.3 C (N = 64; range ° ± ° 27.6–39.4 C). Mean Ta was 30.4 0.2 C (N = Fig. 1. The relationship between head width and SVL ° for male (closed circles; upper regression line) and female 64; range 25.9–36.8 C), and mean Ts was 34.0 ± 0.7°C (N = 64; range 27.1–49.6°C). Body (open circles; lower regression line) Sceloporus siniferus from Guerrero, México. temperature was significantly influenced by 2 both Ta (N = 64, r = 0.42, P < 0.0001; Tb = 2 13.10 + 0.76Ta) and Ts (N = 64, r = 0.28, P < 0.0001; Tb = 28.2 + 0.24Ts). Body temperatures showed some diel fluctuations, as did Ta and Ts (Fig. 2). Body size did not influence 2 Tb (N = 56, r = 0.02, P = 0.36). Males and ° females had the same mean Tb (36.2 C; ANCOVA with Ta as covariate; F1,61 = 0.12, P = 0.72). Sceloporus siniferus are sexually dimorphic in both body size and head width, but not in head length or length of femur. Males were larger and had wider heads than females. Sev- eral other Sceloporus species are sexually dimorphic, with males larger than females; Fig. 2. Diel variation in body temperature (closed cir- however, not all Sceloporus species are sexu- cles), air temperature (open circles), and substrate tem- ally dimorphic (Fitch 1978). Male-biased sex- perature (closed triangles) in Sceloporus siniferus from ± ual dimorphism in head size is relatively com- Guerrero, México. Means are given 1sx–. mon in lizards (e.g., Vitt and Cooper 1985, Perez-Mellado and de la Riva 1993, Smith et 2 al. 1997). We do not have enough information by the relatively large r value for Tb on Ta to determine the cause of sexual dimorphism regression, and the diel variation in Tb. Male in S. siniferus (i.e., whether it is due to sexual and female S. siniferus did not have signifi- selection or niche diversification; see Shine cantly different mean Tb, a situation that has 1989). However, the widespread occurrence of been observed in other studies on Mexican sexual dimorphism in Sceloporus suggests it Sceloporus (e.g., S. grammicus, Lemos-Espinal may have a historical origin in the genus. and Ballinger 1995; S. gadovae, Lemos-Espinal The mean Tb of S. siniferus in this study et al. 1997c; S. ochoteranae, Lemos-Espinal et was 36.2°C, which places it among species al. 1997a). Such a lack of difference in Tb having the highest mean Tb reported in the between males and females may suggest that ° genus Sceloporus. Mean Tb ranges from 28.9 C in these species males and females behave in S. variabilis (Benabib and Congdon 1992) to similarly, such as using similar microhabitats 37.5°C in S. horridus (Lemos-Espinal et al. or being active at the same time. Further work 1997b; see Lemos-Espinal et al. 1997c for a comparing the microhabitat use and activity of review). Environmental temperatures appear males and females in species with sexual Tb to play a relatively large role in determining differences with those without sexual Tb dif- the Tb of individual S. siniferus, as evidenced ferences would be useful. 500 WESTERN NORTH AMERICAN NATURALIST [Volume 61

We thank W.E. Cooper, Jr., and an anony- PEREZ-MELLADO, V., AND I. DE LA RIVA. 1993. Sexual size mous reviewer for their comments on the dimorphism and ecology: the case of a tropical lizard, Tropidurus melanopleurus (Sauria: Tropiduridae). manuscript. Copeia 1993:969–976. SHINE, R. 1989. Ecological causes for the evolution of sex- LITERATURE CITED ual dimorphism: a review of the evidence. Quarterly Review of Biology 64:419–461. BENABIB, M., AND J.D. CONGDON. 1992. Metabolic and SHINE, R., AND T. M ADSEN. 1996. Is thermoregulation water-flux rates of free-ranging tropical lizards Scelo- unimportant for most reptiles? an example using water porus variabilis. Physiological Zoology 65:788–802. pythons (Liasis fuscus) in tropical Australia. Physio- FITCH, H.S. 1978. Sexual size differences in the genus logical Zoology 69:252–259. Sceloporus. University of Kansas Science Bulletin SMITH, G.R., J.A. LEMOS-ESPINAL, AND R.E. BALLINGER. 51:441–461. 1997. Sexual dimorphism in two species of knob- LEMOS-ESPINAL, J.A., AND R.E. BALLINGER. 1995. Com- scaled lizards (genus Xenosaurus) from México. Her- parative thermal ecology of the high-altitude lizard petologica 53:200–205. Sceloporus grammicus on the eastern slope of the VITT, L.J., AND W.E. COOPER, JR. 1985. The evolution of Iztaccihuatl Volcano, Puebla, México. Canadian Jour- sexual dimorphism in the skink Eumeces laticeps: an nal of Zoology 73:2184–2191. example of sexual selection. Canadian Journal of LEMOS-ESPINAL, J.A., G.R. SMITH, AND R.E. BALLINGER. Zoology 63:995–1002. 1997a. Body temperatures of Sceloporus ochoteranae VITT, L.J., AND P.A. ZANI. 1996. Ecology of the lizard from two populations in Guerrero, México. Herpeto- Ameiva festiva (Teiidae) in southeastern Nicaragua. logical Journal 7:74–76. Journal of Herpetology 30:110–117. ______. 1997b. Observations on the body temperatures and natural history of some Mexican reptiles. Bul- Received 24 January 2000 letin of the Maryland Herpetological Society 33: Accepted 6 June 2000 159–164. ______. 1997c. Thermal ecology of the lizard, Sceloporus gadoviae, in an arid tropical scrub forest. Journal of Arid Environments 35:311–319. Western North American Naturalist 61(4), © 2001, pp. 501–502

NOTES ON THE WINTER DIET OF SHORT-EARED OWLS IN NORTHERN CALIFORNIA

Raymond J. Bogiatto1, John A. Hindley1, and Rebecca L. Surles1

Key words: Short-eared Owl, Asio flammeus, winter diet, foraging ecology.

The Short-eared Owl (Asio flammeus), a made every 12–15 days from 15 December California species of concern, has a wide- through 1 March 1995 and 1996. spread distribution within open country habi- Pellets from sympatric Northern Harriers tats throughout the state (Zeiner et al. 1990). (Circus cyaneus) were identified using tech- Although studies of its foraging ecology are niques proposed by Clark (1972) and removed numerous (Tomkins 1936, Stegeman 1957, from all samples. We dissected pellets in the Fisler 1960, Earhart and Johnson 1970, Colvin lab, separating and removing all osteological and Spaulding 1983, Holt 1993, Stone et al. materials from both fur and feathers. 1994, Hogan et al. 1996, and others), no data Cranial and mandibular remains were iden- currently exist for the Sacramento Valley, a tified to the most specific taxonomic level pos- major wintering area for this species in Cali- sible by comparison with known specimens fornia. As a result of the destruction and frag- from the skull collection in the California State mentation of Central Valley marshland and University, Chico (CSUC), Department of Bio- grassland habitats as well as grazing in recent logical Sciences Vertebrate Museum. We iden- decades, owl numbers have declined through- tified post-cranial remains using the zooarch- out much of the range (Remsen 1978). This aeology comparative collection in the CSUC study, which provides data collected from a Department of Anthropology. remnant grassland habitat in the northeastern In all, we collected 106 pellets during this Sacramento Valley, should provide biologists study. Of 135 prey items, the Botta pocket and land managers with a better understand- gopher (Thomomys bottae), California vole ing of the dietary needs of this species during (Microtus californicus), and deer mouse/west- fall and winter months. ern harvest mouse (Peromyscus maniculatus Short-eared Owl pellets were collected from /Reithrodontomys megalotis) were the most a winter roost site at the Vina Plains Nature common prey items, accounting for 68.1%, Preserve (administered by The Nature Conser- 16.3%, and 14.9% of the total prey item sam- vancy), Tehama County, California. The Vina ple, respectively (Table 1). Deer and harvest Plains site is characterized by a mosaic of mouse remains extracted from pellets lacking open grass-forb habitats in association with distinguishing cranial or mandibular elements numerous vernal pools. The roost site is located were treated collectively. In addition, bird in a stand of purple needle-grass (Nassella pul- prey items comprised 0.7% of total prey items. chra), a native perennial bunchgrass, with Our value for percent mammal prey (99.3%) individual owls using hutlike burrows within is consistent with values of Stegeman (1957), grass clumps. Between 8 and 14 owls were Colvin and Spaulding (1983), Holt (1993), and regularly observed using this site throughout Stone et al. (1994), who generated figures of our study. 98.3%, 99.3%, 95.0% (nonbreeding season data), Pellets were first removed from the site and 99.14%, respectively. However, the pres- prior to the onset of each field season to be ence of pocket gophers (68.1%) is relatively certain of deposition period. Collections were high compared with other studies conducted

1Department of Biological Sciences, California State University–Chico, Chico, CA 95929-0515.

501 502 WESTERN NORTH AMERICAN NATURALIST [Volume 61

TABLE 1. Food items in 106 Short-eared Owl pellets collected at the Vina Plains Nature Preserve, Tehama County, California, 1995–1996. Percent of total Taxon Number prey items Botta pocket gopher (Thomomys bottae) 92 68.1 California vole (Microtus californicus) 22 16.3 Deer mouse (Peromyscus maniculatus) 4 3.0 Deer mouse/Western harvest mouse (Peromyscus/Reithrodontomys) 16 11.9 Unidentified swallow (Hirundinidae) 1 0.7 TOTAL PREY ITEMS 135 100.0

within the geographic range of the family COLVIN, B.A., AND S.R. SPAULDING. 1983. Winter foraging Geomyidae. No pocket gophers occurred in behavior of Short-eared Owls (Asio flammeus) in Ohio. American Midland Naturalist 110:124–128. pellets collected by either Fisler (1960) or EARHART, C.M., AND N.K. JOHNSON. 1970. Size dimorphism Hogan et al. (1996), who conducted research and food habits of North American owls. Condor 72: in California and Texas, respectively. Stone et 251–264. al. (1994) reported values for the northern FISLER, G.F. 1960. Changes in food habits of Short-eared Owls feeding in a salt marsh. Condor 62:486–487. pocket gopher (T. talpoides) ranging from HOGAN, K., M.L. HOGAN, J. GABLE, AND M. BRAY. 1996. 10.3% to 27.6% in Short-eared Owl pellets Notes on the diet of Short-eared Owls (Asio flam- collected in western Wyoming. Botta pocket meus) in Texas. Journal of Raptor Research 30: gophers range in size from 71 to 250 g com- 102–104. HOLT, D.W. 1993. Breeding season diet of Short-eared Owls pared with 42 to 100 g for the California vole in Massachusetts. Wilson Bulletin 105:490–496. (Burt and Grossenheider 1976). Our findings REMSEN, J.V., JR. 1978. Bird species of special concern in show that Short-eared Owls will take larger California. California Department of Fish and Game, prey items such as gophers when available. Sacramento. Wildlife Management Administrative We are indebted to The California Nature Report 78-1. 54 pp. STEGEMAN, L.C. 1957. Winter food of the Short-eared Owl Conservancy for granting us permission to in central New York. American Midland Naturalist conduct research on the Vina Plains Preserve. 57:120–124. We also thank S.A. Kirn for helpful comments STONE, E., J. SMITH, AND P. T HORTON. 1994. Season varia- on the manuscript. tion and diet selection from pellet remains of Short- eared Owls. Great Basin Naturalist 54:191–192. TOMKINS, I.R. 1936. Notes on the winter food of the LITERATURE CITED Short-eared Owl. Wilson Bulletin 48:77–79. ZEINER, D.C., W.F. LAUDENSLAYER, JR., K.E. MAYER, AND BURT, W.H., AND R.P. GROSSENHEIDER. 1976. A field guide M. WHITE, EDITORS. 1990. California’s wildlife. Vol- to the mammals. Houghton Mifflin, Co., Boston, MA. ume II, Birds. California Department of Fish and 289 pp. Game, Sacramento. 731 pp. CLARK, R.J. 1972. Pellets of the Short-eared Owl and Marsh Hawk compared. Journal of Wildlife Manage- Received 4 February 2000 ment 36:962–964. Accepted 20 June 2000 Western North American Naturalist

INDEX

Volume 61—2001

AUTHOR INDEX

Anderson, Stanley H., 236 Fiala, G.R., 403 Anhold, John A., 19 Fisher, William L., 139 Arndt, Ronney E., 434 Fresquez, Phil R., 213 Aspinall, Richard J., 428 Fugate, Michael, 11

Ballinger, Royce E., 498 Gabler, Kate I., 480 Baumann, R.W., 403, 473 Gladwin, Douglas N., 182 Beals, Lucille, 452 Goldberg, Stephen R., 248 Beaty, Barry J., 114 Grodeska, Larry S., 452 Beck, Maurie J., 109 Gunther, Kerry A., 277 Belk, Mark C., 36 Gustafson, D., 417 Benedict, Audrey D., 241 Benedict, James B., 241 Hallett, Jennifer C., 57 Benjamin, Lyn, 359 Haroldson, Mark A., 277 Bennett, Kathryn D., 213 Hayes, Marc P., 119 Bess, James A., 195 Heady, Laura T., 480 Biggs, James R., 213 Heaney, Lawrence R., 103 Bleich, Vernon C., 124 Heckmann, Richard A., 245 Bogiatto, Raymond J., 501 Hepworth, Dale K., 129 Borkowski, John J., 43 Hettinger, Ned, 257 Botkin, Daniel B., 261 Hill, James M., 381 Brand, Leonard, 395 Hindley, John A., 501 Brian, Nancy J., 159 Hodge, Jennifer S., 109 Brooks, James E., 139 Hoffman, James T., 409 Brough, Mark, 434 Holechek, Jerry L., 50 Brown, D. Kendall, 139 Humphrey, L. David, 85 Buhler, Matt L., 236 Jacobs, James S., 43 Bull, Evelyn L, 119 Bursey, Charles R., 248 Kalkhan, Mohammed A., 328 Kass, Ronald J., 495 Cade, Brian S., 182 Kedzie-Webb, Susan A., 43 Caldwell, Martyn M., 93 Kemp, William P., 195 Calisher, Charles H., 114 Knisley, C. Barry, 381 Cardenas, Manuel, 50 Cazier, L. Dianne, 375 Landau, Frederick H., 25 Chamberlain, Charles B., 129 Laundré, John W., 480 Chaudhry, Amina A., 452 Lei, Simon A., 78 Chong, Geneva W., 328 Lemos-Espinal, Julio A., 498 Collins, Kenneth P., 149 Lesica, Peter, 1 Lester, Gary T., 375 Despain, Don G., 428 Lichthardt, J., 417 Duke, Sara E., 93 Dymerski, Alan D., 19 Manning, Ann E. Ellison, 223, 229 Mattson, David J., 277 Echelle, Anthony A., 139 McClure, Craig, 347 Evans, Ann S., 452 McDonald, H. Gregory, 64

504 2001] INDEX 505

McGee, Michael N., 245 Schoettle, A.W., 252 McKinnell, Robert G., 248 Schullery, Paul, 255, 289 Miles, Scott, 1 Schupp, Eugene W., 85 Miller, Paul, 347 Seibert, Catherine, 195 Miller, Wade E., 64 Sheley, Roger L., 43 Millspaugh, Sarah H., 316 Shiozawa, Dennis K., 149 Minshall, G. Wayne, 204 Simberloff, Daniel, 308 Mladenka, Greg C., 204 Smith, Geoffrey R., 498 Moore, Barry, 395 Smith, Jonathan P., 409 Morris, Thomas H., 64 Snyder, Keirith A., 463 Munson, Allen S., 19 Stark, B.P., 473 Stewart, Kenneth W., 445 Negron, J.F., 252 Stohlgren, Thomas J., 328 Surles, Rebecca L., 501 O’Neill, Kevin M., 195 Olliff, Tom, 347 Tan, Irene S., 248 Ottenbacher, Michael J., 129 Thayer, Theodore C., 109 Thompson, Daniel B., 25 Philips, T. Keith, 195 Pierce, Becky M., 124 Valdez, Raul, 50 Price, Dave, 347 Van Kirk, Robert W., 359 Propst, David L., 139 Vander Wall, Stephen B., 109

Reich, Robin M., 328 Wagner, Eric J., 434 Reinhart, Daniel P., 277, 347 Walker, Lawrence R., 25 Renkin, Roy, 347 Wambolt, Carl L., 490 Richards, David C., 375 Weaver, T., 428, 417 Rickart, Eric A., 103 Weltzin, Jake F., 463 Roelle, James E., 182 Whipple, Jennifer J., 336, 347 Rogers, D. Christopher, 11 White, Clayton M., 223, 229 Rolston, Holmes, III, 267 Whitlock, Cathy, 316 Rolston, Marni G., 195 Whitney, Michael J., 245 Root, J. Jeffrey, 114. Whittlesey, Lee, 289 Roth, Julie K., 109 Wigand, Peter E., 57 Rowland, Diane L., 452 Williams, David G., 463 Rowlands, Peter G., 159 Wilson, Kristine W., 36

Saiwana, Lewis, 50 Yamamoto, Osamu, 395 Sandberg, John B., 445

KEY WORD INDEX

Taxa described as new to science appear in boldface type.

Abies lasiocarpa, 417 Anostraca, 11 beetle(s), 495 abundance, 129 arid lands, 50 bark, 19 Acacia greggii, 78 aridland soils, 25 tiger, 381 accident, 124 Artemisia, 490 behavior, 395, 445 Aegilops cylindrica, 93 tridentata, 93, 417 caching, 109, 241 Agropyron Asio flammeus, 501 biodiversity, 257 desertorum, 93 assessments, integrated, 328 biological spicatum, 417 autoparasite, 78 control, 308 aliens, 417 invader, 375 alpine, 417 backwater, 149 blacklist, 308 Ambrosia dumosa, 25 balance of nature, 261 body size, 57 annotated checklist, 336 bark beetles, 19 Bonasa umbellus, 236 annuals, 85 beaver, 1 Bonneville cutthroat trout, 434 506 WESTERN NORTH AMERICAN NATURALIST [Volume 61

Bouteloua gracilis, 417 Dioryctria auranticella, 252 Greater Yellowstone, 359 Brachylagus idahoensis, 480 distribution, 129 Ecosystem, 289 Braconidae, 195 spatial, 375 Green River, [Utah], 149 Branchinecta, 11 disturbance, 417 gradients, elevated, 103 hiberna, 11 drainage, 129 grizzly bears, 277 breeding habitat, 229 drawdown, 182 ground sloth, 64 Bromus drought, 78 groundwater, 452 inermis, 417 drumming, 445 tectorum, 85, 93, 417 drumming logs, 236 habitat brucellosis, 277 dung size variation, 57 riparian, 1 Bruneau Hot Springs, [Idaho], suitability, 480 204 ecological historiography, 289 type(s), 195, 417 ecology, foraging, 501 use, 236 caching behavior, 109, 241 ecophysiology, 452 Haplocleidus furcatus, 245 cage effects, 149 ecosystem helminthes, 248 California, 124, 473 dynamics, 261 Hemiptera, 252 carbon isotope discrimination, management, 308 herbaceous cover, 182 463 Elaeagnus angustifolia, 1 history, 129 Centaurea, 417, 428 elevated gradients, 103 fire, 316 Charadrius montanus, 223, 229 elk, 213 host, 78 Chloroperlidae, 473 Encyrtidae, 195 hydrobiid, 204 Chrysanthemum leucanthemum, endangered species, 139, 204, hydrologic alteration, 359 417 375 Chrysothamnus nauseosus, 93 environmental types, 417 Idaho Cicindela limbata albissima, 381 eradication, 308 Bruneau Hot Springs, 204 Cirsium, 428 Escalante River, [Utah], 129 southeastern, 480 climax, 417 Executive Order 13122, 308 INEEL, [Idaho], 480 clover, 277 exotic(s), 1, 257, 267, 428 insect(s) Coleogyne ramosissima, 25 invasive species, 328 cone and seed, 252 Coleoptera, 252 plants, 336, 417 outbreak, 19 Colorado, 114 species, 257, 261, 277 integrated Mesa County, 114 vegetation, 347 approaches to environmental Colorado River, 129 facilitation, 25 assessments, 328 drainage, 149 fairy shrimp, 11 weed management, 347 Colorado Rocky Mountains, 241 fiber consumption, 57 introduced species, 308, 359 Columbia River Gorge, 403 fire history, 316 invasion, 308 Columbia spotted frog, 119 flooding, 182 invasive plants, 43 competition, 25, 375 flora, 336 invasive species, 261, 316 conductivity, 434 forage preference, 490 invertebrate communities, 149 cone and seed insects, 252 foraging ecology, 501 conifer forest, 236 forest(s) Jeffrey pine, 109 Conophthorus contortae, 252 conifer, 236 conservation, 381 Rocky Mountain, 409 kriging, 328 Coral Pink Sand Dunes, [Utah], subalpine, 409 381 lake trout, 277 cottonwood, 1, 452 Geocoris, 195 Lake Bonneville, [Utah], 64 cover, herbaceous, 182 geographic variation, 395 Larrea tridentata, 25 critical thermal maximum, 434 Gila trout, 139 La Sal Mountains, [Utah], 103 Cronartium ribicola, 409 Gila copei, 36 Las Vegas valley, [Nevada], 78 culture, 267 GIS, 428 leaf gas exchange, 463 cutthroat trout, 129, 359 Glacier National Park, [Mon- leatherside chub, 36 tana], 417 Lepidoptera, 252 dandelion, 277 global positioning system, 213 Lepomis cyanellus, 245 DECORANA, 159 Grand Canyon, [Arizona], 159 Leptoglossus occidentalis, 252 deer mice, 109 Grand Teton National Park, life history, 204, 445 demographic monitoring, 495 [Wyoming], 236, 417 limber pine, 252, 409 Dendroctonus rufipennis, 19 granivory, 109 lizards, 498 development, 78 grazing, 50 Lygaeidae, 195 diet, winter, 501 Great Basin, 11, 64, 85, 93 fishes, 36 2001] INDEX 507 macrohabitat, 36 Oncorhynchus relict area, 159 management, 129, 139, 267 clarki, 359 Rio Grande, 452 integrated weed, 347 gilae, 139 riparian habitat, 1 maps, 428 Oregon, 403 Rocky Mountain(s) Megalonyx, 64 northeastern, 119 Colorado, 241 Melilotus, 417, 428 Oreobroma pygmaea, 241 forests, 409 Mesa County, [Colorado], 114 oxygen, 434 northern, 417 México, 498 Rodentia, 395 microhabitat, 36, 114 paleoclimate proxy data, 57 Ruffed Grouse, 236 Microtus montanus, 241 paleoecology, 316 Russian olive, 1 migrations, past plant, 316 Paramylodon, 64 modeling, 428 parasite, 78 sagebrush, 490 monitoring, demographic, 495 past plant migrations, 316 steppe, 93, 159 monogenean, 245 perennials, 85 Salix, 182 Montana, 1, 490 Perlodidae, 445 sampling Glacier National Park, 417 Peromyscus maniculatus, 109 multi-phase, 328 montane vole, 241 Phleum, 428 multi-scale, 328 mortality, 124, 495 pratense, 417 Sangamon, 64 mountain goat, 289 Phoradendron californicum, 78 Sasquaperla, 473 mountain meadows, 417 Picea engelmannii, 19 Sceloporus, 498 Mountain Plover, 223, 229 pigmy bitterroot, 241 Sciuridae, 395 Mourning Dove, 50 Pinus Scolytidae, 19 movements, 119 flexilis, 252 screening cover, 236 mule deer, 124. 490 jeffreyi, 109 seed multi-phase sampling, 328 pinyon-juniper woodland, 159 bank, 85 multi-scale sampling, 328 plant(s) dispersal, 109 mutualism, 109 exotic, 336, 417 dormancy, 85 invasive, 43 seedling(s), 182 Nabidae, 195 past p. migrations, 316 recruitment, 463 Nanonemoura, 403 vascular, 336 seral stages, 417 National Park Service policy, Plecoptera, 403, 473 sexual dimorphism, 498 261, 289 Pleistocene, 64 Short-eared Owl, 501 native, 257 Poa pratensis, 417 shrews, 103 trout, 129 population shrub-steppe, 85 nativism, 257 decline, 495 Sierra Nevada, 109 natural, 257, 267 demographics, 463 snail, 204 invasions, 261 dynamics, 381 Snake River cutthroat trout, nature, 267 Populus, 182 434 balance of, 261 deltoides, 452 snow, 241 Nemouridae, 403 position acquisition rate, 213 soil Neotoma cinerea, 57 precipitation seasonality, 463 nitrogen distribution, 93 nesting distribution, 223 predation, 149 salinity, 452 Nevada, southern, 78 predictive spatial modeling, texture, 480 new genus, 473 328 Sorex, 103 New Mexico, southern, 139 pristine, 267 South Dakota, 248 new records, 103 Pseudoroegneria spicata, 93 southeastern Idaho, 480 new species, 11 Pseudotsuga menziesii, 417 southern Nevada, 78 nonindigenous species, 308 pygmy rabbit, 480 southern New Mexico, 139 nonnative trout, 129 Pyrgulopsis bruneauensis, 204 spatial nonnative species, 289 distribution, 375 North Dakota, 248 Quercus emoryi, 463 heterogeneity, 93 northeastern Oregon, 119 species northern Rocky Mountains, radio collar, 213 diversity, 43 417 Rana endangered, 139, 204, 375 noxious rangeland weeds, 43 luteiventris, 119 exotic, 257, 261, 277 pipiens, 248 exotic invasive, 328 occurrence, 129 range extension, 245 introduced, 308, 359 Odocoileus hemionus, 124 rangeland, 50, 195 invasive, 261, 316 off-highway vehicle impacts, ranid, 119 new, 11 381 Ranidae, 248 nonindigenous, 308 508 WESTERN NORTH AMERICAN NATURALIST [Volume 61

nonnative, 289 tree diseases, 409 vegetation, 480 richness, 43 Trifolium, 417 vocalizations, 395 sympatric, 114 trout syntopic, 103 Bonneville cutthroat, 434 watersheds, 359 threatened, 375 cutthroat, 129, 359 weeds, 417, 428 spit, 64 Gila, 139 whitebark pine, 409 spotted knapweed, 43 lake, 277 white list, 308 stocking, 129 native, 129 white pine, 409 stonefly, 445 nonnative, 129 white pine blister rust, 277, 409 subalpine forests, 409 Snake River cutthroat, 434 wild, 257, 267 survival, 78, 182 Yellowstone cutthroat, 434 wilderness, 267 sympatric species, 114 TWINSPAN, 159 wildlife, 50 syntopic species, 103 winter diet, 501 Uinta Basin, 223, 229 World Trade Organization, 308 Tamarix, 182 Ursus arctos, 277 Wyoming, 236 Tamias Utah minimus, 114 Coral Pink Sand Dunes, 381 Yellowstone, 257, 277, 316 rufus, 114 Green River, 149 Yellowstone cutthroat trout, Tamiasciurus, 395 Lake Bonneville, 64 434 temperature, 434 La Sal Mountains, 103 Yellowstone National Park, 289, thermal 336, 347, 417, 428 ecology, 498 variation springs, 204 dung size, 57 Zenaida macroura, 50 threatened species, 375 geographic, 395 zoogeography, 103 tiger beetle, 381 vascular plants, 336 2001] INDEX 509

TABLE OF CONTENTS Volume 61

No. 1—January 2001

Articles Natural history and invasion of Russian olive along eastern Montana rivers ...... Peter Lesica and Scott Miles 1 Branchinecta hiberna, a new species of fairy shrimp (Crustacea: Anostraca) from western North America ...... D. Christopher Rogers and Michael Fugate 11 Spruce beetle (Dendroctonus rufipennis) outbreak in Engelmann spruce (Picea engelmannii) in central Utah, 1986–1998 ...... Alan D. Dymerski, John A. Anhold, and Allen S. Munson 19 Experimental manipulations of fertile islands and nurse plant effects in the Mojave Desert, USA ...... Lawrence R. Walker, Daniel B. Thompson, and Frederick H. Landau 25 Habitat characteristics of leatherside chub (Gila copei) at two spatial scales ...... Kristine W. Wilson and Mark C. Belk 36 Relationships between Centaurea maculosa and indigenous plant assemblages ...... Susan A. Kedzie-Webb, Roger L. Sheley, John J. Borkowski, and James S. Jacobs 43 Mourning Dove numbers on different seral communities in the Chihuahuan Desert ...... Lewis Saiwana, Jerry L. Holechek, Raul Valdez, and Manuel Cardenas 50 The role of dietary fiber in dung size of bushy-tailed woodrat, Neotoma cinerea: its potential application to paleoclimatic interpretation ...... Jennifer C. Hallett and Peter E. Wigand 57 Taphonomy and significance of Jefferson’s ground sloth (Xenarthra: Megalonychidae) from Utah ...... H. Gregory McDonald, Wade E. Miller, and Thomas H. Morris 64 Survival and development of Phoradendron californicum and Acacia greggii during a drought ...... Simon A. Lei 78 Seed banks of Bromus tectorum–dominated communities in the Great Basin ...... L. David Humphrey and Eugene W. Schupp 85 Nitrogen acquisition from different spatial distributions by six Great Basin plant species ...... Sara E. Duke and Martyn M. Caldwell 93 Shrews of the La Sal Mountains, southeastern Utah . . . . Eric A. Rickart and Lawrence R. Heaney 103 Scatter-hoarding behavior of deer mice (Peromyscus maniculatus) ...... Stephen B. Vander Wall, Theodore C. Thayer, Jennifer S. Hodge, Maurie J. Beck, and Julie K. Roth 109 Microhabitat partitioning by two chipmunk species (Tamias) in western Colorado ...... J. Jeffrey Root, Charles H. Calisher, and Barry J. Beaty 114 Post-breeding season movements of Columbia spotted frogs (Rana luteiventris) in northeastern Oregon ...... Evelyn L. Bull and Marc P. Hayes 119

Note Accidental mass mortality of migrating mule deer ...... Vernon C. Bleich and Becky M. Pierce 124

Book Review Biology and Management of Noxious Rangeland Weeds Roger L. Sheley and Janet K. Petroff, editors ...... Larry D. Howery 126 510 WESTERN NORTH AMERICAN NATURALIST [Volume 61

No. 2—April 2001

Articles Occurrence of native Colorado River cutthroat trout (Oncorhynchus clarki pleuriticus) in the Escalante River drainage, Utah ...... Dale K. Hepworth, Michael J. Ottenbacher, and Charles B. Chamberlain 129 Catastrophic wildfire and number of populations as factors influencing risk of extinction for Gila trout ...... D. Kendall Brown, Anthony A. Echelle, David L. Propst, James E. Brooks, and William L. Fisher 139 Exclusion experiments with backwater invertebrate communities of the Green River, Utah ...... Kenneth P. Collins and Dennis K. Shiozawa 149 Fishtail Mesa: a vegetation resurvey of a relict area in Grand Canyon National Park, Arizona ...... Peter G. Rowlands and Nancy J. Brian 159 Establishment, growth, and early survival of woody riparian species at a Colorado gravel pit ...... James E. Roelle, Douglas N. Gladwin, and Brian S. Cade 182 Natural enemy assemblages on native and reseeded grasslands in southwestern Montana: a family- level analysis ...... Kevin M. O’Neill, William P. Kemp, Catherine Seibert, Marni G. Rolston, James A. Bess, and T. Keith Philips 195 Variation in the life history and abundance of three populations of Bruneau hot springsnails (Pyrgulopsis bruneauensis) ...... Greg C. Mladenka and G. Wayne Minshall 204 Relationship between home range characteristics and the probability of obtaining successful global positioning system (GPS) collar positions for elk in New Mexico ...... James R. Biggs, Kathryn D. Bennett, and Phil R. Fresquez 213 Breeding biology of Mountain Plovers (Charadrius montanus) in the Uinta Basin ...... Ann E. Ellison Manning and Clayton M. White 223 Nest site selection by Mountain Plovers (Charadrius montanus) in a shrub-steppe habitat ...... Ann E. Ellison Manning and Clayton M. White 229 Ruffed Grouse (Bonasa umbellus) drumming log and habitat use in Grand Teton National Park, Wyoming ...... Matt L. Buhler and Stanley H. Anderson 236

Notes Subnivean root caching by a montane vole (Microtus montanus nanus), Colorado Front Range, USA ...... James B. Benedict and Audrey D. Benedict 241 The monogenean Haplocleidus furcatus Mueller, 1937 (phylum Platyhelminthes) on Lepomis cyanellus Rafinesque, 1819 from Utah: a range extension ...... Michael N. McGee, Michael J. Whitney, and Richard A. Heckmann 245 Helminths of northern leopard frogs, Rana pipiens (Ranidae), from North Dakota and South Dakota ...... Stephen R. Goldberg, Charles R. Bursey, Robert G. McKinnell, and Irene S. Tan 248 First report of two cone and seed insects on Pinus flexilis ...... A. W. Schoettle and J.F. Negron 252

No. 3—July 2001

Articles What is natural? Philosophical analysis and Yellowstone practice ...... Paul Schullery 255 Defining and evaluating exotic species: issues for Yellowstone Park policy ...... Ned Hettinger 257 The naturalness of biological invasions ...... Daniel B. Botkin 261 2001] INDEX 511

Natural and unnatural; wild and cultural ...... Holmes Rolston III 267 Effects of exotic species on Yellowstone’s grizzly bears ...... Daniel P. Reinhart, Mark A. Haroldson, David J. Mattson, and Kerry A. Gunther 277 Mountain goats in the Greater Yellowstone Ecosystem: a prehistoric and historical context ...... Paul Schullery and Lee Whittlesey 289 Biological invasions—How are they affecting us, and what can we do about them? ...... Daniel Simberloff 308 A paleoecologic perspective on past plant invasions in Yellowstone ...... Cathy Whitlock and Sarah H. Millspaugh 316 New approaches for sampling and modeling native and exotic plant species richness ...... Geneva W. Chong, Robin M. Reich, Mohammed A. Kalkhan, and Thomas J. Stohlgren 328 Annotated checklist of exotic vascular plants in Yellowstone National Park ...... Jennifer J. Whipple 336 Managing a complex exotic vegetation program in Yellowstone National Park ...... Tom Olliff, Roy Renkin, Craig McClure, Paul Miller, Dave Price, Dan Reinhart, and Jennifer Whipple 347 Status and conservation of salmonids in relation to hydrologic integrity in the Greater Yellowstone Ecosystem ...... Robert W. Van Kirk and Lyn Benjamin 359 Spatial distribution of three snail species, including the invader Potamopyrgus antipodarum, in a freshwater spring ...... David C. Richards, L. Dianne Cazier, and Gary T. Lester 375

No. 4—October 2001

Articles Biology and conservation of the Coral Pink Sand Dunes tiger beetle, Cicindela limbata albissima Rumpp ...... C. Barry Knisley and James M. Hill 381 Variation in the bark call of the red squirrel (Tamiasciurus hudsonicus) ...... Osamu Yamamoto, Barry Moore, and Leonard Brand 395 Nanonemoura, a new stonefly genus from the Columbia River Gorge, Oregon (Plecoptera: Nemouridae) ...... R.W. Baumann and G.R. Fiala 403 Site and stand characteristics related to white pine blister rust in high-elevation forests of southern Idaho and western Wyoming ...... Jonathan P. Smith and James T. Hoffmann 409 Exotic plants in early and late seral vegetation of fifteen northern Rocky Mountain environments (HTs) ...... T. Weaver, D. Gustafson, and J. Lichthardt 417 A rule-based model for mapping potential exotic plant distribution ...... Don G. Despain, T. Weaver, and Richard J. Aspinall 428 Comparative tolerance of four stocks of cutthroat trout to extremes in temperature, salinity, and hypoxia ...... Eric J. Wagner, Ronney E. Arndt, and Mark Brough 434 Drumming behavior and life history notes of a high-altitude Colorado population of the stonefly Isoperla petersoni Needham & Christenson (Plecoptera: Perlodidae) ...... John B. Sandberg and Kenneth W. Stewart 445 Physiological, morphological, and environmental variation among geographically isolated cottonwood (Populus deltoides) populations in New Mexico ...... Diane L. Rowland, Lucille Beals, Amina A. Chaudhry, Ann S. Evans, and Larry S. Grodeska 452 Experimental manipulations of precipitation seasonality: effects on oak (Quercus) seedling demography and physiology ...... Jake F. Weltzin, Keirith A. Snyder, and David G. Williams 463