Full Issue, Vol. 65 No. 1

Western North American Naturalist

Volume 65 Number 1 Article 23

1-27-2005

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

Recommended Citation (2005) "Full Issue, Vol. 65 No. 1," Western North American Naturalist: Vol. 65 : No. 1 , Article 23. Available at: https://scholarsarchive.byu.edu/wnan/vol65/iss1/23

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 65(1), © 2005, pp. 1–10

FUNCTIONAL CHARACTERISTICS OF WILDERNESS STREAMS TWENTY YEARS FOLLOWING WILDFIRE

Christopher T. Robinson1, Urs Uehlinger1, and G. Wayne Minshall2

ABSTRACT.—We compared functional attributes of streams draining catchments burned by wildfire 20 years previously to those of streams in unburned catchments. Long-term analyses of channel profiles indicated most channel change occurred within the first 10 years after fire with little subsequent change the following 10 years. Much of the standing dead timber had fallen, and its effect on stream morphology was directly related to stream size, with important ramifications for future years as decay progresses. The volume of wood in the active channel was 5X higher in a 3rd- order burn stream than in other burn or reference streams, but >80% of this wood was still bridging the stream. Reten- tion of leaves was strongly associated with channel morphology and location of debris dams. Sediment respiration was significantly greater (1.7X) in streams of burned catchments, resulting from greater amounts of loosely attached organic matter in the sediments of these streams. In concordance with respiration results, coefficients of exchange (kex) were almost 5X higher in burn streams than in reference streams, although estimates of transient storage were similar between stream types. We expect the input of large woody debris to increase in the next 10 years in fire-impacted streams as bridging trees collapse into the stream, thereby enhancing channel complexity and habitat heterogeneity, instream metabolism and retention, and consequently stream function. The results emphasize the importance of landscape history, such as large-scale wildfires, on present patterns and processes in stream ecosystems.

Key words: retention, uptake, woody debris, metabolism, storage, channel morphology.

Landscape history profoundly influences legacies that influence ecosystem structure and ecological patterns and processes of lotic eco- function for decades to centuries. These lega- systems today. For example, past glacial events cies (biotic remnants) have direct bearing, for have directly affected current distributions example, on successional patterns and overall of many freshwater organisms such as fish recovery of ecosystems following disturbance (Hershey et al. 1999) and macroinvertebrates (Turner et al. 1998). Because large disturbances (Sweeney et al. 1992). In a more recent con- are infrequent and ecosystem changes follow- text, historical land use patterns by humans ing large disturbances are long-term (Minshall have direct consequences on the present diver- and Brock 1991), few studies have documented sity of stream macroinvertebrates (Harding et recovery processes over the decades following al. 1998). These historical “habitat filters” (sensu large disturbances, particularly in stream eco- Tonn 1990) shape biotic assemblages and pro- systems. Indeed, Gresswell (1999) argues that vide a mechanistic understanding of the re- the temporal response of lotic ecosystems to sponse of these assemblages to disturbance (e.g., wildfire has been essentially ignored. Poff 1997). Consequently, a historical context Wildfire is a landscape-level disturbance is required to more fully appreciate and even that has profound long-term effects on ecosys- elucidate habitat constraints (spatial and tem- tem response and recovery. Wildfires vary in poral) on the structure and function of lotic extent and intensity, and this variability has ecosystems. History is an important but little important abiotic and biotic consequences for appreciated component of habitat templet the- streams and rivers in burned catchments (Min- ory (sensu Southwood 1977). shall et al. 1997). Because of the tight coupling Large-scale disturbances are infrequent and between streams and the terrestrial landscape can vary substantially in their spatial extent through which they flow (Hynes 1975), the and temporal intensity (Foster et al. 1998). For recovery of streams following wildfire can be instance, wildfire can produce abiotic and biotic partitioned into temporal components that

1Department of Limnology, Swiss Federal Institute of Environmental Science and Technology (EAWAG/ETHZ), Ueberlandstrasse 133, 8600 Duebendorf, Switzerland. 2Department of Biological Sciences, Box 8007, Idaho State University, Pocatello, ID 83209.

1 2 WESTERN NORTH AMERICAN NATURALIST [Volume 65 reflect successional changes in the regrowth of predominantly at higher elevations, Douglas- terrestrial vegetation and decomposition char- fir (Pseudotsuga menziesii) at mid-elevations, acteristics of burned terrestrial organic matter and ponderosa pine (Pinus ponderosa) mostly (Minshall and Brock 1991). Minshall et al. at lower elevations. Sagebrush (Artemisia) and (2004) categorized these stages in stream re- grasses are common at mid- and low eleva- sponse and recovery as (1) immediate changes tions, particularly on south-facing slopes. Ripar- (the time of active burning to a few days after); ian vegetation includes alder (Alnus), aspen (2) short-term changes (a few days to the be- (Populus tremuloides), water birch (Betula occi- ginning of spring runoff); (3) mid-term changes dentalis), cottonwood (Populus), and willow (from spring runoff of the 1st postfire year to (Salix). Minshall et al. (2001a, 2001b) provide sometime beyond the 10th year); and (4) long- additional details of the study area. term changes (occurring decades or centuries All study streams are tributaries of the later). Middle Fork Salmon River, with 3 of them An important terrestrial component that having a significant part of their catchment directly affects stream systems in burned catch- (54%–81%) burned by wildfire (Mortar Creek ments is the relatively rapid recovery of ripar- fire) in 1979 (Table 1). The other 3 streams ian vegetation and the falling of most standing were unburned by the Mortar Creek fire but dead timber within the first 5–20 years (Min- experienced less extensive wildfires in their shall et al. 1990). The falling of dead trees has respective catchments during the 20 years fol- important intermediate to long-term conse- lowing 1979, especially in 1988 (Minshall et al. quences in the retention, stability, and function 2001b). The streams are located within 15 km of streams that are expected to be in direct of each other at elevations between 1329 m contrast to the immediate and short-term re- and 1414 m a.s.l. and range in size from 2nd to sponses (e.g., lower channel stability and re- 5th order. Drainage aspect is primarily south- tention) of these systems to wildfire. Readers ern for burned catchments and northern for are referred to Minshall et al. (2001a, 2001b) unburned catchments. Stream slopes are 6%– for data regarding various short-term (1st year) 11% for the smaller streams and 2%–4% for and mid-term (years 1–10) responses of streams the larger study streams. Maximum water tem- following wildfire. The goal of the present peratures in summer were 12° to 17°C. Stream study was to compare functional attributes water pH (daytime) ranged from 8.3 to 8.6, ⋅ –1 (e.g., ecosystem metabolism) of streams in catch- alkalinity from 38 to 102 mg L CaCO3, ments burned 20 years previously relative to and specific conductance between 50 and 120 reference streams in unburned catchments. µs ⋅ cm–1.

STUDY SITE DESCRIPTION METHODS

The study was conducted in late July 1999 Physical attributes of the streams were on 6 streams in the remote Frank Church River characterized using measures of substrate size of No Return Wilderness in central Idaho, USA. and embeddedness, water depth, and bankfull The area is roadless, being accessible only by width (as defined in Davis et al. 2001). We trail or via occasional airfields by light aircraft. determined substrate size (mid-axis) and per- Elevations range from about 1200 m to 3150 m cent embeddedness (quartile percent) by mea- a.s.l., with valley side-slopes averaging 45%– suring 100 randomly selected rocks along a 70%. The geology of the area is underlain by 100-m length of stream. Study reaches were Challis Volcanics (rhyolitic to andesitic rock) selected in 1979, the initial year of study, to intruded by Idaho Batholith (Ross 1934); some represent relative conditions along each stream. river valleys also contain Quaternary glacial We visually estimated embeddedness as the deposits. Soils are loamy sand to sandy loam degree of interstitial filling of the substratum (Cater et al. 1973) and poorly developed. Pre- by fine inorganic particles (after Minshall et al. cipitation, mostly as snow, ranges from 38–50 1997). Water depth also was recorded at these cm in the valleys to 76–100 cm at higher ele- 100 locations. Bankfull widths were recorded vations. The area is lightly to moderately at 5 transects 25 m equidistant from each other, forested, with subalpine fir (Abies lasiocarpa) with the 1st transect being the initial perma- and whitebark pine (Pinus albicaulis) found nent transect determined in 1979 (Minshall et 2005] STREAMS AND WILDFIRE 3

TABLE 1. Catchment characteristics of study sites. Area burned refers to percent of the catchment that was burned by wildfire in 1979: B = burned catchment and U = unburned catchment. Specific Area Stream Elevation Area Channel Drainage conductance Stream burned (%) order Link (m) (km2) slope (5) aspect at 25°C Little (B) 81 2 7 1408 5.1 6 S 138 Pungo (U) trace 2 10 1396 13.8 7 NE 112 WF L Loon (B) 54 3 24 1341 20.7 4 SW 100 EF Indian (U) trace 3 12 1413 17.3 3 NE 111 EF L Loon (B) 61 5 105 1329 84.0 2 S 117 Indian (U) trace 5 89 1414 212.7 1.5 NW 78 al. 2001a, 2001b). Subsequent transects were release point and proceeding upstream, sum- placed upstream of the 1st one. Channel cross ming for each 5-m increment. Each 50-m sections were recorded each year at the per- reach was examined for leaves twice. The pre- manent transect established in 1979. For the dominant habitat type of each 5-m segment purpose of this paper, channel profiles for 3 within the 50-m reach was scored as pool, run, years (1980, 1988, and 1999) are presented in or riffle, and also for presence of debris dams. the results for the study streams. Streams were Sediment metabolism was quantified in each characterized chemically in the field using stream using an in situ oxygen-depletion portable pH and specific conductance meters method (after Jones et al. 1995). Fine sedi- (Orion Corp., USA). We recorded diel water ment (<8 mm diameter) was collected by temperatures for each stream using tempera- shovel from the stream bottom after removal ture loggers (Vemco minilogs) during the time of surface sediment (typically 10–15 cm in of sampling, usually 1–2 days. Sample days depth) and placed into clear plexiglass tubes (5 were sunny and represented typical summer cm wide × 30 cm long, volume = 0.6 l, n = 3 temperatures for each study stream. per stream). Half of each tube was filled with The volume of large (>5 cm diameter) sediment and the remaining space with stream woody debris was recorded within a 50- to water. Water oxygen concentrations were mea- 100-m reach in each stream beginning imme- sured (WTW Oxi 340, Weilheim, Germany) diately upstream of the 1st transect used for and the tube capped off with a rubber stopper. determining bankfull width. Each piece of The rubber stopper had a vent that allowed all large woody debris within the bankfull width air trapped in the tube to be released; thus, was measured (diameter and length) and cate- the tube was filled only with stream sediment gorized as either bridging the stream or hav- and ambient stream water. Once the tubes ing potential for enhancing stream retention. were sealed, they were buried in the stream to Conglomerates of woody debris, i.e., debris eliminate effects of solar radiation on metabo- masses retained by larger pieces of woody lism. The tubes were allowed to incubate for a debris, also were measured for total volume. period of 3–4 hours, retrieved, and the oxygen Volume was determined based on respective concentration remeasured. Tubes were incu- geometric shapes of the measured debris. bated in the stream beginning in late morning Because of time constraints in the field, e.g., around 1030–1100. For comparison among sites, 3–6 hours per stream, we could determine leaf sediment respiration rates were normalized to retention at only 4 of the study streams (Little, 20°C using an Arrhenius temperature coeffi- EF Indian, Pungo, and WF Little Loon Creeks). cient of 1.072. The contents from each tube At each site we collected 300–400 fresh alder were stored in plastic zip-type bags and kept leaves from trees in the riparian zone. Essen- in the dark for transport to the laboratory. tially no fresh leaves were observed in any of In the laboratory the organic content of the study reaches prior to release of the treat- the sediments was determined. First, loosely ment leaves. Collected leaves were transported attached particulate organic matter (LPOM) to the most upstream transect and individually was elutriated, split into size fractions <63 released over a period of about 3 minutes. and >63 µm, and determined as ash-free dry After 1 hour we collected leaves from the stream mass (AFDM). These sediments were dried at channel beginning 50 m downstream of the 60°C for at least 4 days, weighed, ashed at 4 WESTERN NORTH AMERICAN NATURALIST [Volume 65

500°C for 4 hours, and reweighed. The difference in weights was an estimate of the AFDM. In addition, the remaining organic matter that was attached to the sediments also was measured as AFDM. ANOVA was used to test for significant differences in the measured variables between streams of burned and unburned catchments (Zar 1984). We estimated transient water storage using a pulse release of a conservative tracer (NaCl) and recording changes in specific conductance of the stream water at 2 locations downstream of the release point as the pulse of tracer moved through the system. The 1st location was immediately downstream of the effective Fig. 1. Box plots of diel water temperatures recorded in mixing zone, the 2nd location about 100 m midsummer for the study streams: B = burned catchment downstream of the first. The transient storage and U = unburned catchment. 2 zone (At as m ) and coefficient of exchange 2 ⋅ –1 (kex as m s ) were calculated for each stream as described by Meier and Reichert (2005) in channel stability. An exception was Little using average current velocity, channel geom- Creek, where additional channel scouring etry, channel roughness (Bathurst 1978), slope, occurred in the latter 10 years. Pungo Creek, and discharge. Differences in At and kex be- the reference for Little Creek, also showed tween burn and reference streams were tested some channel movement during the second 10 using the Mann-Whitney U-test (Zar 1984). years of the study perhaps in response to a wildfire that partially burned its catchment in RESULTS 1988. Channel profiles for EF Indian and Indian Creeks were essentially unchanged during the Most habitat variables measured in 1999 20-year study period. Differences in channel were similar between comparably sized streams profiles in EF Indian between 1980 and 1988 of burned and unburned catchments (Table 2). were likely caused by a change in transect For example, cobble-size substrata dominated location after that 1st year. most streams, although coefficients of variation Total volume of large woody debris in (CVs) for substrata were higher for some streams streams of burned catchments relative to of burned catchments than for respective ref- streams of unburned catchments varied with erence streams. Average water depths, sub- stream size. Wood volume was essentially the stratum embeddedness, and CVs for these same in burned and reference 2nd-order and variables were comparable between similar- 5th-order streams (Fig. 2). However, the vol- sized streams. Indian Creek was substantially ume of large woody debris was ca. 5X greater wider (bankfull width) than respective EF L in the 3rd-order stream in the burned catch- Loon, but this is primarily because of geomor- ment than in the reference stream. In addi- phology and drainage size. Diel temperature tion, the function of woody debris differed recordings indicated little difference between considerably between burn and reference sites comparably sized burn and reference streams with respect to stream size. For example, 80% (Fig. 1), although values generally increased of the large woody debris was categorized as with stream size. providing retention in the 2nd-order burn Substantial channel scouring was evident stream, compared to 35% in the reference for the streams in burned catchments relative stream where the majority of woody debris to those in unburned catchments in the 20- bridged the stream. In contrast, most woody year study period (Fig. 2). Stream channel debris was bridging 3rd-order streams (>60%), change was most evident in burned catch- especially in the burn stream. Last, most large ments during the first 10 years following the woody debris acted as bridges in 5th-order wildfire, with little additional change obvious streams, although the percent bridging was during the 2nd decade, suggesting an increase 15% greater in the reference stream. 2005] STREAMS AND WILDFIRE 5 idth: Depth W tudy streams: B = burned and U -value . –1 . Loosely attached particulate –1 sFP (kg sediment) –1 d 2 C) O ° BurnReference 0.37 1.20 0.16 1.03 5.77 0.029 BurnReference 0.41 2.45 0.21 1.60 14.42 0.002 ReferenceReferenceReference 0.013Reference 11.1 10.4 0.004 0.78 1.6 1.6 0.34 C and amount of organic matter in bed sediments of study streams. ANOVA table of the C and amount of organic matter in bed sediments study streams. ANOVA = 5 for bankfull width; std standard deviation, cv coefficient of variation as %. ° n C) Burn 0.022 0.006 13.41 0.002 ° m m µ µ 3. Respiration at 20 ABLE T arameter Stream type Mean otal sediment POMotal LPOM Burn 10.6 Burn 3.300 3.95 0.15 0.69 0.704 11.38 0.004 LMPM < 63 organic matter (LPOM) and attached to the sediments (AOM) in g AFDM (kg sediment) comparison between burned and reference sites. Respiration (20 P Respiration (20 T AOMT LPOM > 63 Burn 7.0 0.8 25.47 0.0002 Substratum size (cm) Depth (cm) Embeddedness (%) Bankfull width (m) ______2. Summary statistics for substratum size, water depth, embeddedness, bankfull width, and width:depth ratio the s ABLE T Stream Mean std cv Mean std cv Mean std cv Mean std cv ratio unburned catchment. Sample size was 100 for each, except unburned catchment. Sample size was 100 for each, except Little (B)Pungo (U)WF L Loon (B)EF Indian (U)EF L Loon (B)Indian (U) 18.9 10.5 24.4 18.1 18.2 20.0 14.9 20.7 20.0 23.5 22.2 105 141 85 111 129 21.6 15.9 9.0 14.6 12.1 97 26.7 9.9 5.8 11.2 10.0 13.2 31.1 62 65 77 83 49 13.8 28.0 15.3 30.8 44 28.5 31.5 27.8 26.8 35.3 30.8 33.3 23.8 99 176 115 108 106 25.0 5.4 2.6 105 3.6 3.8 6.5 2.3 0.6 0.5 1.4 17.5 1.4 42 22 13 37 1.8 21 10 34 28 24 32 24 56 6 WESTERN NORTH AMERICAN NATURALIST [Volume 65

Fig. 2. Cross-sectional profiles for the study streams recorded from a permanent transect installed in 1979. Profiles from 1980, 1988, and 1999 are depicted to represent changes occurring in the first 10 years relative to changes in the second 10 years following the wildfire. Reference streams are indicated by U and burned streams by B.

Retention of leaves was strongly related to of released leaves after 50 m, and over 65% of the amount of large woody debris acting as the large woody debris was noted as bridging retention devices in each stream. For example, the stream; no major debris dams were noted ca. 80% of the released leaves were retained in the release reach of Pungo Creek. Woody within the first 10 m in Little Creek (Fig. 4), debris in Pungo Creek likely reflects the impact and 80% of the large woody debris recorded in of miners that lived there around the 1930s or this stream was categorized as having reten- earlier and used the surrounding forest for tion capability (Fig. 3). Channel maps indicated mining activities. WF L Loon retained <50% that a debris dam was present within this first of the released leaves in 50 m, and over 80% 10 m that effectively retained the released of its large woody debris was bridging the leaves. In contrast, Pungo Creek retained 80% stream. An increase in retention of leaves in 2005] STREAMS AND WILDFIRE 7

Fig. 4. Cumulative percentage of leaves retained within a 50-m reach in 4 study streams. Logistic constraints prevented this experiment from being conducted in all streams (see Methods).

ence streams than burned streams, whereas the amount of loosely associated organic matter (all measured size fractions) was significantly greater in burned streams (ca. 4–6X as high) than in reference streams (Table 3). Loose organic material is apparently more metaboli- cally active than attached organic matter. The transient storage zone (At) for examined streams ranged between 0.21 and 0.71, averaging 0.43 for burn streams and 0.46 for reference streams. Results of the Mann-Whit- ney U-test indicated no significant differences between burn and reference streams (P = 0.83) for relative sizes of storage zones. However, coefficients of exchange (kex) were significantly ± greater in burn streams (0.0078 0.0013 sx–) than in reference streams (0.0017 ± 0004; P < 0.05).

DISCUSSION Fig. 3. Bar graphs illustrating total volume of wood within a 50-m reach in the study streams, and percent of Our results indicate that the functioning of this wood categorized as either bridging the stream or contributing to channel stability and retention. Streams stream ecosystems can be altered over time in are paired by size (stream order). response to changes caused by wildfire, a large, infrequent disturbance. Many mid-term changes (1 to >10 years) in stream function are probably WF L Loon and EF Indian after about 25 m a response to the increase in physical distur- also was because of a debris dam. bance from enhanced peak discharges and sed- Sediment metabolism, in terms of respiration iment input from valley side-slopes the first few ⋅ –1 (mg O2 d ), was significantly greater (1.7 X) years following wildfire (Minshall et al. 1997). in streams of burned catchments than in refer- Burned streams are expected to become more ence streams (Table 3). The total amount of autotrophic as solar and nutrient inputs are organic matter associated with sediments was greater and stream temperatures higher than similar between burned and unburned streams. before the fire or later in the recovery sequence However, the amount of organic matter attached (Gresswell 1999, Minshall et al. 2001b), as evi- to sediments was significantly greater in refer- denced by changes in periphyton (Robinson et 8 WESTERN NORTH AMERICAN NATURALIST [Volume 65 al. 1994, Minshall et al. 2001b) and macroinver- tional Park, USA, burned by the 1988 wildfires tebrate assemblages (Mihuc and Minshall 1995, (see Minshall et al. 1997). Minshall et al. 1997). Spencer and Hauer (1991) The higher stream complexity from the in- documented increased phosphorus and nitro- crease in large woody debris should enhance gen loads in streams during wildfire from the instream retention as found for forested streams smoke and ash from burning vegetation. After (Bilby and Likens 1980). Our results indicate >10 years of recovery, burned streams should that retention of leaves is associated strongly become increasingly heterotrophic due to de- with the presence of large woody debris with- creased solar inputs and increases in alloch- in channels. However, increased channel sta- thonous organic matter as riparian vegetation bility inherent in burned streams during the matures and recovering terrestrial vegetation second 10 years also infers a greater potential on valley side-slopes retains surface sediments for these streams to retain coarse organic mat- and water (Minshall and Brock 1991). The inter and thereby enhance their functional effi- crease in side-slope vegetation also should ciency. Indeed, observations of a stream in a reduce peaks in surface runoff. We found that similar climate draining a catchment that burned channels of burned streams, while still struc- 50 years previously suggests that organic mat- turally simple, became more stable during the ter retention will become even greater in the second 10 years following fire, and the observed next 20–30 years in our study streams (GWM increase in woody debris should enhance struc- personal observation), especially as riparian tural heterogeneity. vegetation continues to mature and is incorpo- Our results suggest that stream morphology rated into stream channels. becomes more complex once the large woody Sediment respiration was significantly higher debris bridging fire-impacted streams is incor- in streams of burned catchments probably be- porated into the active stream channel. Wallace cause of the greater amount of loosely attached et al. (1995) document major channel changes organic matter in the sediments. Reference (increases in complexity) in response to adding and burned streams exhibited similar levels of large woody debris to a stream channel. How- organic material adhering to sediment sur- ever, our findings regarding woody debris were faces, suggesting that this loosely attached strongly related, although nonlinearly, to stream material was of recent origin and provided an size. There is probably a particular size stream important functional component, e.g., a resource in which the input and utilization of fallen for microbes, to these streams. Minshall et al. burned timber is maximized; our results sug- (1997) showed greater levels of benthic organic gest that this may occur in mid-order streams matter in streams of burned catchments that as the volume of large woody debris bridging were attributed to input from valley side-slopes. these streams was substantially greater than This suggests that as stream channels become that observed for the reference stream and more stable in burned catchments over time other burn streams of different sizes. However, (>10 years), more organic material transferred an alternative explanation is the difference in from terrestrial sources is retained and pro- percentage catchment burned among streams cessed within the system; i.e., the stream of different size. This gradient in percentage becomes more functionally efficient. Minshall catchment burned may influence the degree of et al. (1989) posited that litter inputs will be- woody debris input in later years following come even greater between 20 and 70 years wildfire. For instance, Minshall et al. (1997) following wildfire as succession of the terres- also found that larger streams had less of their trial landscape occurs. catchment burned than smaller headwater In concordance with sediment respiration streams. Further, our measures of embedded- results, burn streams had higher coefficients ness and substratum size were less variable (as of exchange even though transient storage CV) in burned streams with lower percentage zones were similar between stream types. The catchment burned. The exact timing of this higher kex likely reflects higher sediment per- enhanced input is likely related to climate; our meability of burn streams, as supported by the study was completed in a semiarid environ- greater amount of loosely attached organic ment, suggesting input may be slower than for matter in the sediment samples of burn streams. streams in burned catchments of more temper- This enhanced permeability may have been a ate climates, e.g., streams in Yellowstone Na- result of substantial channel alteration during 2005] STREAMS AND WILDFIRE 9 the first 10 years following the wildfire. In DAVIS, J.C., G.W. MINSHALL, C.T. ROBINSON, AND P. L AN- effect, wildfires may indirectly increase hypo- DRES. 2001. Monitoring wilderness stream ecosystems. General Technical Report RMRS-GTR-70, rheic exchange properties of affected streams USDA Forest Service, Rocky Mountain Research by increasing turnover of instream habitats Station, Ogden, UT. 137 pp. and delivery of smaller substrata by flow dis- FOSTER, D.R., D.H. KNIGHT, AND J.F. FRANKLIN. 1998. Land- turbances. By maintaining greater porosity of scape patterns and legacies resulting from large, bed sediments, more organic matter can be infrequent forest disturbances. Ecosystems 1:497–510. GRESSWELL, R.E. 1999. Fire and aquatic ecosystems in retained in the system for processing by forested biomes of North America. Transactions of microbes and other organisms, thus enhancing the American Fisheries Society 128:193–221. overall system metabolism. HARDING, J.S., E.F. BENFIELD, P.V. BOLSTAD, G.S. HELFMAN, In summary, wildfire is a major large-scale AND E.B.D. JONES. 1998. Stream biodiversity: the ghost of land use past. Proceedings of the National disturbance in many regions of the world. Academy of Science 95:14843–14847. Because wildfire is an infrequent event, it im- HERSHEY, A.E., G.M. GETTEL, M.E. MCDONALD, M.C. poses abiotic and biotic legacies on the land- MILLER, H. MOOERS, W.J. O’BRIEN, J. PASTOR, ET AL. scape. These legacies play important roles in 1999. A geomorphic-trophic model for landscape control of arctic lake food webs. BioScience 49: the observed structure and function of stream 887–897. ecosystems that differ spatially over long peri- HYNES, H.N.B. 1975. The stream and its valley. Interna- ods of time. Our results support previous con- tionale Vereinigung für theoretische und angewandte tentions (e.g., Minshall et al. 1997) that tempo- Limnologie 19:1–15. ral changes in stream structure and function JONES, J.B., S.G. FISHER, AND N.B. GRIMM. 1995. Vertical hydrologic exchange and ecosystem metabolism in a following wildfire are closely associated with Sonoran Desert stream. Ecology 76:942–952. successional changes occurring in adjacent ter- MEIER, W.K., AND P. R EICHERT. 2005. Mountain streams: restrial environments. These temporal changes modelling hydraulics and substance transport. Jour- in the terrestrial environment are transferred nal of Environmental Engineering 131. MIHUC, T.B., AND G.W. MINSHALL. 1995. Trophic general- to streams via changes in solar radiation and ists vs. trophic specialists: implications for food web inputs of large woody debris, leaves, and other dynamics in post-fire streams. Ecology 76:2361–2372. organic litter that alter channel stability and MINSHALL, G.W., D.A. ANDREWS, J.T. BROCK, C.T. ROBIN- complexity, organic matter retention and pro- SON, AND D.E. LAWRENCE. 1990. Changes in wild cessing, and ultimately ecosystem efficiency. trout habitat following forest fire. Pages 111–119 in F. Richardson and R.H. Hamre, editors, Wild Trout IV: proceedings of the symposium. U.S. Government ACKNOWLEDGMENTS Printing Office 774-173/25037, Washington, DC. MINSHALL, G.W., AND J.T. BROCK. 1991. Observed and We greatly appreciate assistance in the field anticipated effects of forest fire on Yellowstone stream by K.E. Bowman, C.D. Myler, J.N. Minshall, ecosystems. Pages 123–135 in R.B. Keiter and M.S. Boyce, editors, The Greater Yellowstone Ecosystem: G. Mladenka, A.M. Prussian, C. Richards, and redefining America’s wilderness heritage. Yale Uni- K. Richards. We thank C. Dambone-Boesch versity Press, New Haven, CT. and C. Relyea for assistance in the laboratory. MINSHALL, G.W., J.T. BROCK,D.A. ANDREWS, AND C.T. Partial funding for the study was provided to ROBINSON. 2001a. Water quality, substratum, and GWM by the U.S. Forest Service–Payette biotic responses of five central Idaho (USA) streams during the first year following the Mortar Creek National Forest with special thanks to Dr. Fire. International Journal of Wildland Fire 10: David Burns, and the Department of Biologi- 185–199. cal Sciences, Idaho State University. Comments MINSHALL, G.W., J.T. BROCK, AND J.D. VARLEY. 1989. Wild- by 2 anonymous reviewers improved the paper fire and Yellowstone’s stream ecosystems. BioScience presentation. 39:707–715. MINSHALL, G.W., C.T. ROBINSON, AND D.E. LAWRENCE. 1997. Postfire responses of lotic ecosystems in Yellow- LITERATURE CITED stone National Park, USA. Canadian Journal of Fish- eries and Aquatic Sciences 54:2509–2525. BATHURST, J.C. 1978. Flow resistance of large-scale rough- MINSHALL, G.W., C.T. ROBINSON, D.E. LAWRENCE, D.A. ness. American Society of Civil Engineers, Journal ANDREWS, AND J.T. BROCK. 2001b. Benthic macroin- of Hydrology Division 104:1587–1603. vertebrate assemblages in five central Idaho (USA) BILBY, R.E., AND G.E. LIKENS. 1980. Importance of organic streams over a 10-year period following disturbance debris dams in the structure and function of stream by wildfire. International Journal of Wildland Fire ecosystems. Ecology 61:1107–1113. 10:201–213. CATER, F.W., D.M. PINCKNEY, W.B. HAMILTON, R.L. PARKER, MINSHALL, G.W., T.V. ROYER, AND C.T. ROBINSON. 2004. R.D. WELDIN, T.J. CLOSE, AND N.T. ZILKA. 1973. Stream ecosystem responses following the 1988 Yel- Mineral resources of the Idaho Primitive Area and lowstone wildfires: the first 10 years. Pages 87–97 in vicinity, Idaho. U.S. Geologic Survey Bulletin 1304. L. Wallace, editor, After the fires: the ecology of 10 WESTERN NORTH AMERICAN NATURALIST [Volume 65

change in Yellowstone National Park. Yale University and biogeography of aquatic insects in eastern North Press, New Haven, CT. America. Pages 143–176 in P. Firth and S.G. Fisher, POFF, N.L. 1997. Landscape filters and species traits: editors, Global climate change and freshwater eco- towards mechanistic understanding and prediction systems. Springer-Verlag, New York. in stream ecology. Journal of the North American TONN, W.M. 1990. Climate change and fish communities: Benthological Society 16:391–409. a conceptual framework. Transactions of the Ameri- ROBINSON, C.T., S.R. RUSHFORTH, AND G.W. MINSHALL. can Fisheries Society 119:337–352. 1994. Diatom assemblages of streams influenced by TURNER, M.G., W.L. BAKER, C.J. PETERSON, AND R.K. PEET. wildfire. Journal of Phycology 30:209–216. 1998. Factors influencing succession: lessons from ROSS, C.P. 1934. Geology and ore deposits of the Casto large, infrequent natural disturbances. Ecosystems quadrangle, Idaho. U.S. Geologic Survey Bulletin 854. 1:511–523. SOUTHWOOD, T.R.E. 1977. Habitat, the templet for ecolog- WALLACE, J.B., J.R. WEBSTER, AND J.L. MEYER. 1995. Influ- ical strategies? Journal of Animal Ecology 46: ence of log additions on physical and biotic charac- 337–365. teristics of a mountain stream. Canadian Journal of SPENCER, C.N., AND F.R. HAUER. 1991. Phosphorus and Fisheries and Aquatic Sciences 52:2120–2137. nitrogen dynamics in streams during wildfire. Jour- ZAR, J.H. 1984. Biostatistical analysis. 2nd edition. Prentice- nal of the North American Benthological Society Hall, Inc., Englewood Cliffs, NJ. 10:24–30. SWEENEY, B.W., J.K. JACKSON, J.D. NEWBOLD, AND D.H. Received 16 December 2003 FUNK. 1992. Climate change and the life histories Accepted 22 June 2004 Western North American Naturalist 65(1), © 2005, pp. 11–23

A LIFE HISTORY STUDY OF THE SNAKE RIVER PLAINS ENDEMIC LEPIDIUM PAPILLIFERUM (BRASSICACEAE)

Susan E. Meyer1,3, Dana Quinney2, and Jay Weaver2

ABSTRACT.—Lepidium papilliferum is an ephemeral species that occupies “slick spot” microhabitats in the matrix of sagebrush steppe vegetation of the southwestern Snake River plains, Idaho, USA. We related population demographic data collected from 1993 to 1996 to on-site precipitation data on the Orchard Training Area west of Boise. We also carried out field seed-retrieval and in situ seed bank studies. We found that L. papilliferum has a dual life history strategy. A fraction of each cohort sets seed as summer annuals, while the remaining plants remain vegetative and potentially biennial. Surviving biennials flower and set seed along with the annual cohort of the following year. The switch to flowering as an annual appears to be based on threshold rosette size. Probability of survival to flowering was much lower for biennials than for annuals of the same cohort, but surviving biennials sometimes had enhanced seed production. The summer- dry environment of the Snake River plains combined with the slick spot habitat has apparently selected for a primarily summer annual life cycle for this species. Seeds were highly dormant at dispersal and were not responsive to dormancy- breaking cues. Those from a given cohort of L. papilliferum remained viable in the soil for at least 11 years. This persistent seed bank provides a buffer against extinction in sequences of years when seed production is low or absent. Estimated seed bank size varied from near zero for a heavily disturbed site that formerly supported the species to 18 viable seeds ⋅ dm–2 for an extant population in high-quality habitat. Management for population preservation for L. papilliferum should focus on protecting the seed bank from destruction caused by livestock trampling and other anthropogenic disturbances.

Key words: Lepidium papilliferum, endangered species, demography, life history strategy, rare plant, seed bank dynamics, slickspot peppergrass, desert, Idaho.

The current renaissance in the study of the formed decisions regarding management for population biology of plants began with the preservation of this unique plant species. publication of the pivotal book by Harper (1977). Lepidium papilliferum has the highest doc- The field also received impetus, especially for umented extirpation rate of any of Idaho’s rare rare plants, with the passage of the Endangered plant taxa, due largely to the combination of Species Act in 1973. It has become increas- outright loss and dramatic ecological decline of ingly clear since then that information beyond its sagebrush steppe habitat (Michael Mancuso, detailed geographic distribution, or even beyond Conservation Data Center, Idaho Department census data generated by most monitoring of Fish and Game, unpublished reports). It has programs, is necessary to critically assess the been the object of intensive monitoring efforts conservation status of rare plants or to project for over a decade, especially on the Idaho Army likely outcomes of various management actions National Guard Orchard Training Area (OTA) (Schemske et al. 1994). In contrast to monitor- southwest of Boise (Quinney 1998). The species ing, demographic studies establish the basic was formally proposed for listing as endangered features of species life history and can help by the U.S. Fish and Wildlife Service (2002). identify life history stages that are particularly Lepidium papilliferum is limited in its dis- vulnerable to negative impacts. In the study tribution to the southwestern Snake River reported here, we used a demographic approach plains, Idaho, USA (Fig. 1). According to the to characterize the life history of the Snake River Conservation Data Center, Idaho Department plains endemic Lepidium papilliferum. We used of Fish and Game, there are 42 populations or the resulting information to make inferences metapopulations (element occurrences) known about the impacts of management activities. A to be extant (Mancuso and Moseley, unpub- more complete understanding of L. papilliferum lished report, 1998). Many of these populations life history strategy will greatly facilitate in- are very small, but a few metapopulations are

1USDA Forest Service, Rocky Mountain Research Station, Shrub Sciences Laboratory, 735 North 500 East, Provo, UT 84606. 2Idaho Army National Guard, Gowen Field, Boise, ID 83707. 3Corresponding author.

11 12 WESTERN NORTH AMERICAN NATURALIST [Volume 65

Fig. 1. Map of current known geographic distribution of Lepidium papilliferum and the location of the metapopulation containing the States study site and associated areas where supplemental data were taken (note white arrow indicating location). Data source: Conservation Data Center, Idaho Department of Fish and Game. large and can include thousands of plants in comparable nearby long-term weather station). years of favorable spring precipitation. Lepid- Lepidium papilliferum is specialized to occupy ium papilliferum is related to the more south- a particular microhabitat within the sagebrush erly and more widely distributed taxon L. steppe vegetation of the Snake River plains, montanum Nutt., a biennial to short-lived peren- namely the “slick spots” that are small-scale nial, and was formerly considered a variety of sites of water accumulation in the gently undu- that taxon. Lepidium papilliferum is found in lating landscape. Seedlings of the dominant the sagebrush steppe vegetation type on shallow perennial species are usually excluded from to deep loess soils that are underlain primarily slick spots, presumably because of their inabil- by basaltic bedrock (Fisher et al. 1996). The ity to tolerate winter flooding. Slick-spot soils climatic regime of this region is characterized are characterized by higher clay content and by low and variable winter and spring precipi- salinity than adjacent zonal soils as well as lower tation and dry summers, with a mean annual organic matter content (Fisher et al. 1996). The precipitation of 250 mm (Kuna, ID, the most slick spots represent an environment where 2005] LEPIDIUM PAPILLIFERUM LIFE HISTORY 13 spring water availability extends into the sum- Census Data and Survival mer, an advantage to species that can tolerate Lepidium papilliferum plants were first or escape the associated disadvantages. censused at the States study site in early April Field observations made on the OTA, south- 1993, when all overwintering plants (1992 west of Boise, in 1990–1992 revealed that L. cohort) were tagged. We tagged individuals of papilliferum has a dual life history strategy. the 1993 cohort as they appeared in 4 micro- Seedlings emerged in early spring. Plants that sites; detailed count data were obtained from survived until June either flowered and set seed the other 8. The microsites were censused every immediately, functioning as summer annuals, 3 weeks through fruit set in late June. At each or remained vegetative. Vegetative individuals census the status of each tagged plant was that survived functioned as biennials, flowering recorded as vegetative, budding, flowering, and setting seed with the annual cohort of the fruiting, dead, or missing; the number of un- subsequent year. Successful biennials appeared tagged plants in the first 4 categories was ob- to have higher reproductive output than the tained for the microsites with untagged plants. annuals that set seed with them, but their All plants alive and vegetative at fruiting time chances of surviving to reproduction seemed in mid-June 1993 were then tagged and fol- greatly reduced. We used these preliminary ob- lowed at monthly intervals. Similar procedures servations as the basis for the study reported were followed for the 1994 and 1995 cohorts, here. We wished to document the existence of except that it was possible to tag plants in all the 2 life history forms, to quantify the survival 12 microsites from the time of recruitment for and reproductive output of annuals versus bi- these cohorts because of their lower numbers. From these data we determined numbers for ennials, and to relate variation in demographic the recruited cohort, the fraction surviving until parameters to between-year differences in flowering time, the fraction to flower as annuals, precipitation. In addition, we wanted to exam- the fraction to remain vegetative and poten- ine the role of seed dormancy and of the seed tially biennial, and the fraction of biennial hope- bank in population persistence and to estimate fuls to oversummer and overwinter successfully. the size of the in situ seed bank. Size Class Distributions and Reproductive Output MATERIALS AND METHODS At flowering time in June 1993 and 1994, we Study Site Selection measured the rosette diameter of each tagged The distribution of Lepidium papilliferum plant. Size class distributions based on inter- on the OTA was mapped in some detail in vals of 10 mm were prepared for the flowering 1990–1992. The States study site, in the White biennial cohort, the flowering annual cohort, Dog area (43°16′49.5″N, 116°04′3.9″W, 970 m), and the vegetative (potentially biennial) cohort. was chosen for detailed demographic study. It To determine the relationship between was known to support thousands of L. papil- rosette diameter and seed output, we tagged liferum plants in years of high spring rainfall and measured a set of plants (n = 54) chosen and was surrounded by sagebrush steppe veg- to represent the full range of sizes in 1993. We avoided destructive sampling at the States study etation in relatively good ecological condition. site, choosing instead to sample at the nearby Supplemental data were also collected in the and similar Red Tie study site. Plants were nearby Red Tie, Orchard Corner, South Stan- then harvested prior to seed dispersal and difer, and T Rex Hill study sites. The States their dry mass was determined. We counted site consists of a series of 12 subpopulations in fruiting pedicels and total seed number and 2 microsites varying in size from 9 m to 165 obtained total seed mass for a subset of these 2 m , with a combined total area of approxi- plants (n = 33). Because ovule number in the mately 600 m2. The microsites are scattered genus Lepidium is fixed at 2, we could calcu- through the matrix vegetation over an area of late fruit fill percentages as well as seed yield. approximately 2 ha. Only the microsites them- We harvested and processed a similar set of selves are considered potential habitat for the plants from Red Tie in 1994 (n = 34). species, which is found only rarely in the matrix We used linear regression to relate plant vegetation (Quinney 1998). dry mass to rosette diameter. To relate seed 14 WESTERN NORTH AMERICAN NATURALIST [Volume 65 number to plant dry mass, we used analysis of period for 2 weeks, and the number of germi- covariance with year as the class variable and nated seeds was scored. We then determined plant dry mass as the continuous variable. viability of the remaining seeds using tetra- Variables were transformed as necessary to ob- zolium staining. For the tetrazolium viability tain linear relationships. To examine the effect evaluation, we immersed seeds in 1% tetrazol- of year on mean fruit fill and mean individual ium solution for 24 hours at room tempera- seed mass, we used 1-way analysis of variance. ture. We then bisected the seeds and scored as To estimate total seed production and aver- viable those with firm, uniformly red embryos. age seed output per plant for annuals and Retrieved seeds were therefore classified as biennials in 1993 and 1994, we combined size recently germinated, germinable, dormant, or class distribution data from the States study dead. The difference between the sum of these site with regressions relating plant size to seed fractions and the number of seeds initially viable production. (100) was assumed to represent seeds that had germinated in earlier episodes. Examination Seed Germination and of the sum across all years of seeds actually ob- Retrieval Studies served to be recently germinated corroborated Laboratory seed germination studies were this approach. carried out with 1991, 1992, and 1993 L. papilliferum collections from the Orchard Corner In Situ Seed Bank Study and T Rex Hill sites. These included various We did not attempt to directly quantify the combinations of high-temperature (30°, 40°, in situ seed bank at the States study site be- and 50°C) dry after-ripening, incubation at cause we did not want to disturb the seed bank superoptimal temperature, moist chilling, and during the course of the demographic study. gibberellic acid treatments prior to incubation We later realized that it would be helpful to at a range of temperatures considered likely to develop a method for evaluating the seed bank be optimal. These experiments are not described directly, particularly from a management per- in detail because the resulting germination spective. We chose microsites in 3 adjacent percentages were consistently very low. OTA locations with different histories of L. We used an artificial seed bank retrieval papilliferum abundance: Orchard Corner, South approach with seeds of known age to study Standifer, and T Rex Hill. At Orchard Corner patterns of seed bank persistence and seed the population of L. papilliferum had appar- germination under field conditions. The study, ently become extirpated some years earlier. which included T Rex Hill seed collections from Although it once had a steady and plentiful 1991 and 1992, was installed in August 1992 at population, heavy livestock use damaged the the Orchard Corner study site, a location where site and it has not recovered. The South Stan- the plots could be fenced. We enclosed groups difer site is characterized by moderate, yet con- of 100 seeds in flat nylon mesh bags, 5 × 5 cm. sistent L. papilliferum populations every year, These were placed in sets (1 bag for each seed while the T Rex Hill site houses an excep- collection) in a typical microsite habitat, buried tional example of a population that contains under 5–10 mm of soil, and covered with 1- hundreds of plants annually. cm-mesh hardware cloth cones to protect them We collected seed bank samples in October from disturbance. We set the cones out in 3 2002. From 1 microsite at each of the 3 sites, blocks with 25 cones in each block. Seed bags 21 seed bank sample locations (3 × 3 cm) were were retrieved at intervals, at first very fre- randomly selected, 7 each from the center, the quently and then with decreasing frequency, edge, and the intermediate band between the over a period of 11 winters (August 1992–June center and the edge (middle). For each of these 2003). One set of bags from each block was sample points, the potentially seed-containing retrieved on each of 19 dates. soil was carefully removed by layers into zip- If the retrieval took place when the ground lock bags. These layers were the surface silt was wet, we first counted any recently germi- layer, clay hardpan layer, and subsurface clay nated seeds. The remaining intact seeds were layer. These layers were collected separately then placed on blotters in petri dishes in an regardless of their thickness. The depth of each incubator at 10°/20°C with a 12-hour photo- sample excavation varied depending on the 2005] LEPIDIUM PAPILLIFERUM LIFE HISTORY 15 thickness of the layers. The lowermost layer and the 1993 recruited cohort was by far the was collected to a depth of no more than 6 cm. largest observed. The fraction that functioned Soil was dry at the time of sampling. as summer annuals was high (78%). Average Each of the 189 seed bank samples (3 sites seed production for these annual plants was × 3 positions × 3 layers × 7 replications) was 125 seeds per plant. Overwintered biennials processed individually. First, the sample was of the 1992 cohort had an average seed output emptied onto the surface of a small, fine-mesh of 787, six times higher than the average annual. soil sieve, and the fine soil (most of each sample) The biennial hopeful cohort exhibited relatively was washed through the sieve with running high over-summer survival (22%), with sum- water. The clay hardpan material became quite mer precipitation almost twice the long-term soft when wet and could easily be broken up average (Table 2). Twelve percent of the bien- and washed away without damage to the seeds. nial hopeful cohort survived to flower with the The coarse fraction remaining on the sieve, annuals of the 1994 cohort. which contained the seeds, was spread to dry Growing season precipitation was below on a labeled paper towel. Once the sample average in 1994, and the recruited cohort was was dry, it was processed by hand to extract small (Tables 1, 2). Most surviving plants failed the seeds. Seeds of all species were extracted, to flower, with only 29% of the cohort func- separated by species, and counted. Seeds of 3 tioning as annuals. Seed output per plant was species were found: Lepidium papilliferum, L. also low, and the difference between seed pro- perfoliatum (clasping peppergrass), and Ranun- duction by annuals (46 per plant) and bienni- culus testiculatus (bur buttercup). The latter 2 als (105 per plant) was not as evident. Because species are common annual weeds on the Snake of this strong contrast in environmental qual- River plains. Seeds of the 2 Lepidium species ity between years, annuals of the 1993 cohort could easily be distinguished by size; the weedy actually averaged higher seed production than species has much larger seeds. surviving biennials of that cohort. Biennial hope- Weed seeds were not processed further, but fuls also fared poorly in 1994, with complete L. papilliferum seeds were subjected to addi- over-summer mortality due to summer precip- tional procedures. Seeds were placed on mois- itation well below the average (Table 2). tened germination blotters in petri dishes and Growing season precipitation for the 1995 incubated at room temperature for 5 days. At cohort was near the average, and the recruited the end of this time, germinated seeds were cohort was intermediate in size (Tables 1, 2). counted and removed, and the remaining seeds About two-thirds of the surviving plants func- were pierced (an artificial method for obtaining tioned as summer annuals. There were no seedlings from dormant seeds) and incubated overwintered biennials because of complete for another 5 days. Germinated seeds were summer mortality the previous summer. Sum- again counted and removed, and the remaining mer precipitation was well above average in seeds were subjected to tetrazolium staining 1995, resulting in almost 60% over-summer as described above to determine whether they survival of biennial hopefuls. Twenty-three per- were viable. Total seed number, readily germin- cent of the 1995 biennial hopeful cohort sur- able seed number, number of seeds germinable vived to set fruit in the summer of 1996. after piercing, and number of nongerminable The 3 years of our study had contrasting seeds staining viable were recorded for each weather scenarios, permitting us to infer some sample. Data were analyzed using analysis of relationships between weather patterns and variance. demographic response. Late winter precipitation appeared most important in mediating RESULTS the size of the recruited cohort, with a large increase in the number of seedlings recruited Census Data and Survival with increased February–March precipitation. Examination of the demographic data for Survival of potential biennials was tied pri- the 1993, 1994, and 1995 cohorts showed widely marily to summer precipitation, with increased varying patterns of recruitment and survival survival in years with substantial summer rain- among years (Table 1). Growing season precip- fall. But it was also influenced negatively by itation was above average in 1993 (Table 2), early winter precipitation. Heavy early winter 16 WESTERN NORTH AMERICAN NATURALIST [Volume 65

TABLE 1. Measured Lepidium papilliferum demographic data for the years 1993–1996 at the States study site on the Orchard Training Area, southwestern Idaho.

______Year Demographic measure 1993 1994 1995 1996 Established cohort 4065 203 833 na Fruiting annualsa 2503 46 384 2 Biennial hopefulsa 708 110 216 0 Over-summered biennial hopefulsb 156 0 123 0 Fruiting biennials from previous yearc 60 85 0 49 Mean seed output of an annual plantd 125 46 na na Mean seed output of a biennial plantd 787 105 na na Estimated seed raine 360,095 11,041 na na aFruiting annuals plus biennial hopefuls represent the subset of each spring-established cohort alive at fruiting time in June. bThe subset of biennial hopefuls alive at the end of the summer. cThe subset of over-summered biennial hopefuls from the previous year that survived to set fruit with the annual cohort of the subsequent year. dEstimated for plants present in a calendar year, i.e., for current-year annuals and biennials established the previous year. eCalculated by multiplying the number of plants by the mean seed output per plant for annuals and for biennials, then summing the 2 numbers.

TABLE 2. Precipitation data (mm) for key seasons from the Orchard Training Area Range 2 weather station for the period 1991–1997, along with long-term means for these key seasons from Kuna, ID, the most climatically comparable NOAA weather station in the area.

Kuna, ID, OTA Range 2 weather station year of record long-term ______Season mean 1992 1993 1994 1995 1996 Early winter (N-D-J) 55 46 60 36 88 116 Late winter (F-M) 49 35 93 27 49 44 Spring (A-M) 54 20 44 46 52 58 Growing season (F-M-A-M) 104 55 137 73 101 102 Summer (J-J-A) 33 46 62 9 80 7 Water year (10/1–9/30) 250 169 279 138 288 243

precipitation tends to maintain the microsites spread. Nonreproductive plants, the majority in a flooded condition, with subsequent mortal- of the surviving cohort that year, were almost ity the following spring, particularly of rosettes exclusively members of the smallest size class located in microsite centers where water stands (5 mm). These results suggest that (1) mean plant longer (Quinney personal observation). size is positively related to growing season precipitation, (2) overwintering biennials grow Size Class Distributions and larger in response to high growing season pre- Reproductive Output cipitation than annuals, and (3) flowering as an Size class distributions for plants present in annual is based primarily on achieving a mini- June 1993 and 1994 showed major differences mum resource status and consequent thresh- between years (Fig. 2). In 1993 the plants of old size, as nonflowering individuals are con- the overwintered 1992 biennial cohort were sistently concentrated in the smallest classes. mostly in the middle to large size classes, with Rosette diameter at flowering was a good a mode of 55 mm. The modal size class for the predictor of fruiting plant dry mass (Fig. 3A). 1993 annuals was 25 mm, while nonreproduc- This relationship was linear when both vari- tive (biennial hopeful) plants were concentrated ables were expressed on log scales. Analysis of in the smallest size classes (5–15 mm). In 1994 covariance showed that plant dry mass was in the modal class for overwintered 1993 bienni- turn a good predictor of seed number per plant als was 25 mm. The modal class for 1994 annu- (Fig. 3B; F = 403, df = 1,62, P < 0.0001). Slopes als was 5 mm, but there was considerable were not significantly different between years 2005] LEPIDIUM PAPILLIFERUM LIFE HISTORY 17

Fig. 2. Size class distributions (based on rosette diameter at flowering) for L. papilliferum fruiting biennials (overwintering cohort), fruiting annuals (annual cohort), and biennial hopefuls (vegetative cohort) at the States study site in June 1993 and 1994.

(F = 0.3, df = 1,62, n.s.), but there was a sig- gree of cue nonresponsive primary dormancy. nificant between-year difference in the eleva- Seed viability as determined by tetrazolium tion of this linear relationship (F = 5.02, df = staining was consistently very high (>95%). 1,62 P < 0.05). More seeds were set on aver- In the field seed-retrieval study at Orchard age per unit of plant mass in 1993, the wet Corner, no seeds germinated during the 1st year, than in 1994, the dry year. One reason for winter (1993), despite tens of thousands of L. this is that the fill percentage was significantly papilliferum seedlings that established from higher in 1993 than in 1994 (69% vs. 45%; F = the in situ seed bank on the OTA during this 17.2, df = 1,35, P < 0.0002). Individual seed period of above-average precipitation (Table 2, mass was also significantly higher in the wet Fig. 4). We conclude that few, if any, of these year (0.44 mg vs. 0.36 mg; F = 22.0, df = naturally occurring seedlings were generated 1,60, P < 0.0001). from seeds produced the previous summer. Combining the size class and reproductive Field seed germination was observed only output data permitted us to estimate total L. in late winter and early spring retrievals (Fig. papilliferum seed rain at the States study site 4). Germination pulses were observed in 1994, for both 1993 and 1994. The combination of 1995, and 1996 at levels that were apparently fewer plants and smaller plants resulted in a unrelated to precipitation regimes. We unfor- 33-fold decrease in estimated seed rain in 1994 tunately do not have observational data on relative to 1993 (Table 1). field germination for the period 1997–2003, but remaining viable seed percentages in the Seed Germination and 1999, 2000, and 2003 early summer retrievals Retrieval Studies suggest that this pattern of germination of The highest germination percentage obtained approximately 6% of the initially viable seeds in any of the laboratory seed germination exper year continued. periments was 10%, in response to 6 weeks of Some seeds lost viability in the field re- dry after-ripening at 50°C followed by 8 weeks trieval bags before germinating (Fig. 4). We of moist chilling. Most treatments yielded no first began to detect these nonviable seeds in germination. The seeds thus exhibit a high de- the 2nd year of the retrieval, and by the end of 18 WESTERN NORTH AMERICAN NATURALIST [Volume 65

Fig. 3. The relationship between (A) L. papilliferum rosette diameter and plant dry mass (statistics from linear regression) and (B) plant dry mass and seed production for plants measured at flowering, then harvested just before seed dispersal in 1993 and 1994 (statistics from analysis of covariance).

8 winters approximately 25% of the seeds were by data from seed retrieval in June 2003, which nonviable. This average annual 3% loss of via- showed a small but measurable fraction of the bility also appears to be largely independent seeds still alive after 11 winters in the ground of yearly weather variation. The combined (Fig. 4). effects of germination and viability loss are Most seeds in the seed-retrieval experiment thus 9% of the original seed cohort per year at any given time were dormant, which is after the 1st year, yielding a maximum lon- apparently how they avoided germination and gevity of 12 years for seeds in the seed bank. were able to persist as seeds through multiple This estimate of seed longevity was confirmed winters (Fig. 4). The 2 seed populations in the 2005] LEPIDIUM PAPILLIFERUM LIFE HISTORY 19

Fig. 4. Results of an 11-year seed-retrieval experiment with 1991 and 1992 L. papilliferum seed collections installed at Orchard Corner in August 1992. The difference between inverted triangles and triangles represents the seed fraction that lost viability without germinating. retrieval experiment were somewhat different tion had the smallest seed bank. In 63 Orchard in their patterns of dormancy loss and subse- Corner samples, only a single viable seed was quent germination. The 1991 population showed isolated, from the clay subsoil layer in the cen- higher levels of dormancy, especially over the ter of the microsite, for an estimated seed den- first few years, and also showed a slightly sity of <1 ⋅ dm–2. The microsite at the most higher rate of viability loss (Fig. 4). Both seed pristine location, T Rex Hill, had an estimated collections apparently contained a fraction that seed density of 18 ⋅ dm–2, while the South could reenter dormancy after becoming tem- Standifer microsite showed an intermediate porarily germinable in the summer. This frac- seed density (7.4 ⋅ dm–2). Densities were calcu- tion was evident for the 1992 collection in the lated by summing numbers across soil layers. fall of 1992, soon after experiment initiation. It If we consider that perhaps seeds in the clay was also evident and quite large for the 1991 subsoil layer are too deeply buried to be part collection in the fall of 1994. The 1992 collection of the active seed bank, these numbers would also contained a fraction that lost dormancy in change to 0 ⋅ dm–2 for Orchard Corner, 7.4 ⋅ the summer of 1994, but most nondormant dm–2 for South Standifer, and 10.6 ⋅ dm–2 for seeds of the 1992 collection germinated the T Rex Hill. following spring rather than reentering dor- In analysis of variance with location, micro- mancy, whereas most of the nondormant 1991 site position, and soil layer as fixed main effects seeds reentered dormancy. and log-transformed viable seed number ⋅ dm–2 as the response variable, a significant differ- In Situ Seed Bank Study ence among locations occurred as suggested There were large differences among sites in above (F = 5.27, df = 2,162, P < 0.005; Fig. 5). the estimated size of the in situ L. papilliferum The difference among positions within micro- seed bank (expressed in terms of number of sites was also marginally significant (F = 2.51, viable seeds ⋅ dm–2; Fig. 5). As expected, the df = 2,162, P < 0.084). Samples taken from microsite at the degraded Orchard Corner loca- the middle zone of the microsite had more 20 WESTERN NORTH AMERICAN NATURALIST [Volume 65

sample too great to detect small differences, even if present. If we assume that the T Rex Hill microsite in the seed bank study is comparable to the States study site during the demographic study in terms of seed bank size, we would arrive at an estimate of 1,080,000 viable seeds in the seed bank in the 12 combined microsites of the States site (600 m2). If we exclude seeds in the subsoil from this calculation, the total estimated seed bank at the States study site during the demographic study would be 696,000. Of the viable seeds extracted from soil samples in the in situ seed bank study, 38% germinated without treatment, 50% germinated after piercing, and the remaining 12% did not germinate but were scored as viable in the tetrazolium test. Germinable seed percentages were thus somewhat comparable to those observed in the 1994 autumn retrieval (Fig. 3). It is not known whether these seeds would subsequently have reentered dormancy or germinated.

DISCUSSION

In adapting to the environment of the Snake River plains, L. papilliferum has undergone modifications in its life history strategy relative to its close congener and putative ances- tor L. montanum, a widely distributed species with wide ecological amplitude that is found in a variety of open habitats in arid to semiarid regions of the southern Intermountain area. Most obvious of these life history changes is the shift from biennial to summer annual as the primary life history expression. The dry summers in southwestern Idaho have apparently ⋅ Fig. 5. Number of viable Lepidium papilliferum seeds applied strong selection pressure in the direc- dm–2 recovered in October 2003 from microsites at 3 sampling locations, from 3 positions within each microsite, tion of the annual habit. Even in years when and from 3 soil layers at each position. Values are means biennials are successful, their contribution to of 7 replicates. See text for statistical interpretation. the seed rain may be small, and in some years, such as 1995, they make no contribution at all (Table 1). Life history theory predicts that the viable seeds than those taken from the edges optimum strategy in terms of fitness or repro- and the centers (Fig. 5). There was no signifi- ductive output for monocarpic plants is achieved cant difference in seed numbers among soil through reproduction at the age where the layers (F = 1.71, df = 2,62, P < 0.171), al- product of survival and fecundity is maximized though we observed a trend for increased (Silvertown and Charlesworth 2001). Thus, numbers of seeds in the surface silt layer (Fig. longer lifespan is expected under circumstances 5). The low value for proportion of variance where the survival curve is concave, with higher accounted for by the model in the overall survival for older plants than for seedlings, ANOVA (0.226, P = 0.0134) suggests that rep- and where fecundity increases steeply with licate number in this preliminary study was age. A shift to an annual life history strategy, probably too low and variation from sample to on the other hand, is expected when mortality 2005] LEPIDIUM PAPILLIFERUM LIFE HISTORY 21 is higher for older plants than for seedlings, The nature of the environmental control of and when the increase in fecundity with age is floral initiation is clearly a key aspect of these not sufficient to counterbalance this greater contrasting life history expressions (Klinkhamer risk of mortality. et al. 1987). One possible explanation of the Many species exhibit life history variation shift to the annual habit is elimination of the that includes both annuals and biennials. These vernalization requirement that would prevent contrasting life histories may result from genetic 1st-year flowering in biennials, as has happened differentiation (Terborg et al. 1980, Lacey 1988), in southern European biotypes of wild beet or they may represent a plastic response to (van Dijk et al. 1997). If this were then cou- growing conditions (Lee and Hamrick 1983). pled with a size or resource requirement for There is ample theoretical treatment of the rel- the photoinduction of flowering, the result ative advantages of the annual versus biennial might be the one we observed, namely that life history strategy (Hart 1977, Silvertown 1983, larger plants function as annuals, while the Kelly 1985). The crux of these arguments seems smallest plants attempt to delay reproduction to depend on whether population growth rate until the return of long days the following or reproductive output per plant is the best spring, when they are able to reach the thresh- measure of fitness. When there is a long-lived old size for floral induction. seed bank and only episodic stand establish- Another major change in life history strat- ment, the delay in reproduction experienced egy for L. papilliferum is the evolution of cue by biennials is largely unimportant, even though nonresponsive seed dormancy that permits it lowers the population growth rate (Kelly seeds to persist in the seed bank. The evolu- 1985). More important is the question of tion of persistent propagule banks in stochasti- whether the increased reproductive output of a cally varying environments is also predicted biennial can compensate for its increased risk by a large body of theoretical literature (Cohen of pre-reproductive mortality, as described 1966, Brown and Venable 1986, Tuljapurkar above. In L. papilliferum the answer is often and Istock 1993). Without a persistent seed negative. Depending on the sequence of years, bank, L. papilliferum probably could not suc- there may not even be increased fecundity for ceed as an annual in its stochastically varying biennial members of a cohort, and their mor- habitat. Lepidium montanum, in contrast, has tality risk under the current summer-dry climate largely nondormant seeds, as does the closely scenario appears to be high. This leads to the related shrubby species L. fremontii Wats. of question of why biennials persist in the popu- the Mojave Desert (Meyer unpublished data). lation at all. The difference between annuals This suggests that the persistent seed bank and biennials is unlikely to be genotypic in this and the annual strategy go hand in hand, as L. species, as selection would be likely to elimi- fremontii inhabits an even more arid environ- nate the biennial biotype. But plants that are ment than L. papilliferum, but is long-lived too small to set seed in the current year have and thus less in need of a persistent seed nothing to lose, so to speak, by attempting to bank. This conclusion is supported by the fact function as biennials. The annual-biennial de- that L. lasiocarpum Nutt. ex Torr. and Gray, an cision is probably made as a plastic response, annual species of the warm deserts of the so that larger plants, which stand a good chance Southwest, has cue nonresponsive dormancy of seed production as annuals, flower as annu- and forms persistent seed banks (Philippi 1993). als, while the smallest plants, that would prob- In contrast, the seeds of L. davisii Rollins, a ably not successfully set seed as annuals, are long-lived herbaceous species endemic to larger programmed to wait until the following year. dry lakes of the Snake River plains, show only The role of climatic variation in the persis- cue responsive dormancy and germinate in the tence of the biennial life history strategy in L. field within a year (Meyer and Quinney un- papilliferum could best be addressed through published data). simulation modeling using long-term stochas- We terminated our formal demographic tic variation in precipitation as a driver vari- study at the States study site in 1996, but the able. It may be that the sequence of years we area has continued to be censused as part of sampled was insufficient to estimate the con- the monitoring program at the OTA. Starting tribution of the biennial subset of the popula- in 1996, the population at the study site went tion to the seed bank in the long term. into decline. Based on late winter and growing 22 WESTERN NORTH AMERICAN NATURALIST [Volume 65 season precipitation, 1996 should have been a training also poses a potential threat, but, at reasonably good year for L. papilliferum (Table least on the OTA, this threat is minimal because 2), and count data from the OTA in general L. papilliferum population centers have been confirmed that it was (Quinney unpublished voluntarily placed off limits to all military train- data). But at the States site, there were only 2 ing for over 10 years. These areas are still grazed individuals of the current year cohort present at by both cattle and sheep, however. While live- flowering time, even though we have evidence stock rarely consume the plants, they have that thousands of plants present in 1993 pro- been observed to destroy many individuals duced over 300,000 long-lived seeds (Table 1). through uprooting, defecation, and trampling The overwintered biennials of the 1995 cohort (Quinney and Weaver unpublished data). And were still present in summer 1996, but there because there is no natural surface water in was virtually no recruitment of new plants. This the area, other than the slick spots, livestock recruitment failure was associated with exten- are attracted to them at just the time when sive livestock trampling when the microsites trampling does the most damage to slick spot were filled with water earlier in the spring, an hydrology and soil structure and presumably event that “reduced them to pockmarked mush” to the persistent seed bank as well. (Quinney and Weaver, field notes, 1996). In It is also important to recognize that many 1997 there were again only a handful of plants of the practices commonly used to revegetate at the study site, but in this case the low num- sagebrush steppe communities after wildfire, bers could have been because of recruitment including the introduction of slick spot–adapted failure associated with the exceptionally dry exotic reclamation species like Kochia prostrata late winter. Monitoring data over the subse- (L.) Schrader, the use of preemergent herbi- quent years, however, have shown that the cides, and the use of seeding equipment, such population at the States study site has never as the rangeland drill that disrupts the soil recovered. For example, in 2003, a year with surface, may have serious negative effects on L. papilliferum habitat and on the status of the high L. papilliferum abundance overall in rein situ seed bank (Mancuso unpublished report, sponse to good spring precipitation, only 8 Scholten 2000). We have clear evidence to sug- plants were observed at the States site, and 7 gest that any form of soil disturbance is likely of these were clustered at one end of a single to have a deleterious effect on the in situ seed microsite. We hypothesize that the severe bank and therefore on the probability of suc- trampling disturbance in the spring of 1996 in cessful recruitment episodes in subsequent some way disrupted or buried the in situ seed years. Until we have more comprehensive data bank. It is not likely that the observed popula- on the status of L. papilliferum seed banks in tion decline there can be explained in terms of relation to variation in slick spot soils, hydrol- weather patterns alone. ogy, and disturbance regime, the most prudent The importance of the seed bank in man- approach to preservation of this species is to agement for population persistence of L. papil- manage all potential habitat as if it were cur- liferum cannot be overemphasized. There will rently occupied by L. papilliferum. This means definitely be years when no actively growing keeping all forms of surface disturbance to a plants are observed, but this is not necessarily minimum. evidence that L. papilliferum is not present. Recognition of the ephemeral nature of actively ACKNOWLEDGMENTS growing plants has led to the development of a Habitat Integrity Index for assessing the habi- This research was supported in part by the tat quality and status of L. papilliferum popu- Department of Defense Legacy Program. Mem- lations using a method that does not rely ex- bers of the OTA research technical staff pro- clusively on censusing (Moseley and Mancuso vided valuable field assistance each season. unpublished report, Mancuso unpublished re- Megan Ferguson carried out the in situ seed ports, Conservation Data Center, Idaho Depart- bank sample analyses. ment of Fish and Game). Threats to L. papilliferum habitat include wildfire, weed invasion, LITERATURE CITED off-road vehicular traffic, land conversion, and BROWN, J.S., AND D.L. VENABLE. 1986. Evolutionary ecol- livestock disturbance (Mosely unpublished re- ogy of seed-bank annuals in temporally varying envi- port, Mancuso unpublished report). Military ronments. American Naturalist 127:31–47. 2005] LEPIDIUM PAPILLIFERUM LIFE HISTORY 23

COHEN, D. 1966. Optimizing reproduction in a randomly Evaluating approaches to the conservation of rare varying environment. Journal of Theoretical Biology and endangered plants. Ecology 75:584–606. 12:119–129. SCHOLTEN, D. 2000. Effects of rangeland rehabilitation FISHER, H., L. ESLICK, AND M. SEYFRIED. 1996. Edaphic practices on Lepidium papilliferum (Henderson) Nel- factors that characterize the distribution of Lepidium son and McBride. Master’s thesis, University of Idaho, papilliferum. Bureau of Land Management, Idaho Moscow. State Office, Technical Bulletin 96-6. 23 pp. SILVERTOWN, J.W. 1983. Why are biennials sometimes not HARPER, J.L. 1977. The population biology of plants. Aca- so few? American Naturalist 121:448–453. demic Press, London. SILVERTOWN, J.W., AND D. CHARLESWORTH. 2001. Introduc- HART, R. 1977. Why are biennials so few? American Natu- tion to plant population biology. 4th edition. Blackwell ralist 111:792–799. Science, Oxford, UK. 347 pp. KELLY, D. 1985. Why are biennials so maligned? Ameri- TERBORG, S.J., A. JANSE, AND M.M. KWAK. 1980. Life cycle can Naturalist 125:473–479. variation in Pedicularis palustris. Acta Botanica Neer- KLINKHAMER, P.G.L., T.J. DEJONG, AND E. MEELIS. 1987. landica 29:397–405. Life history variation and the control of flowering in TULJAPURKAR, S., AND C. ISTOCK. 1993. Environmental un- short-lived monocarps. Oikos 49:309–314. certainty and variable diapause. Theoretical Popula- LACEY, E.P. 1988. Latitudinal variation in reproductive tim- tion Biology 43:251–280. ing in a short-lived monocarp, Daucus carota (Api- U.S. FISH AND WILDLIFE SERVICE. 2002. Endangered and aceae). Ecology 69:220–232. threatened wildlife and plants: listing the plant Lep- LEE, J.M., AND J.L. HAMRICK.1983. Demography of two idium papilliferum (slickspot peppergrass) as endan- natural populations of musk thistle (Carduus nutans). gered. Federal Register 67(135):46441–46450. Journal of Ecology 71:923–936. VAN DIJK, H., P. BAUDRY, H. MCCOMBIE, AND P. V ERNET. PHILIPPI, T. 1993. Bet-hedging germination of desert 1997. Flowering time in wild beet (Beta vulgaris ssp. annuals: variation among populations and maternal maritima) along a latitudinal cline. Acta Oecologica effects in Lepidium lasiocarpum. American Natural- 48:47–60. ist 142:488–507. QUINNEY, D. 1998. LEPA (Lepidium papilliferum). Natural Received 5 December 2003 Resources Group, Environmental Management Accepted 18 May 2004 Office, Idaho Army National Guard, Boise. 25 pp. SCHEMSKE, D.W., B.C. HUSBAND, M.H. RUCKELSHAUS, C. GOODWILLIE, I.M. PARKER, AND J.G. BISHOP. 1994. Western North American Naturalist 65(1), © 2005, pp. 24–35

CHARACTERIZING A FIRST OCCURRENCE OF BISON DEPOSITS IN SOUTHEASTERN NEVADA

William Gray Johnson1, Saxon E. Sharpe1, Thomas F. Bullard1, and Karen Lupo2

ABSTRACT.—Late Holocene bison bones eroding from the badlands topography of Cathedral Gorge State Park in southeastern Nevada have been identified as the remains of Bison and date between approximately 400 and 850 years old. The bones were originally thought to be turn-of-the-century cattle and then early bison, so park personnel had need to solve this interpretive problem. A multidisciplinary team sought information on archaeology, geomorphology, paleontology, and paleoenvironment to document the 1st established occurrence of bison in this part of Nevada and to develop a hypothesis explaining their demise.

Key words: bison, southeastern Nevada, late Holocene, Cathedral Gorge State Park.

In 1975 a skull, several long bones, vertebrae, ern corner of Meadow Valley at an average and bone fragments were removed from the elevation of 1460 m (4800 feet) above mean picnic area of Cathedral Gorge State Park in sea level. Nevada. Initial analyses suggested they were Geology and Geomorphology bones from extinct Bison bison antiquus and were estimated to date between 12,000 and Cathedral Gorge State Park, a north–south 15,000 years old. However, excavations at trending gorge opening to the south, is the other localities within the state park (Fig. 1), product of a long period of geological history. coupled with further analysis and radiometric The region surrounding Cathedral Gorge be- dating, indicate the remains are modern bison longs to one of the numerous fault-bounded dating between 400 and 850 years old. Recov- structural basins of the Great Basin that accu- ery and dating of these remains extend the mulated volcanic tuffs and terrigenous sedi- spatial and temporal range of B. bison into ments throughout much of the Tertiary. During southeastern Nevada. Furthermore, an absence the late Tertiary, the region was characterized of archaeological association with these deposits by lake systems depositing thick packages of suggests natural death for these animals that fine-grained lacustrine, fluvial, and eolian sed- may have a geological explanation. iments (Tschanz and Pampeyan 1970, Stewart We identified 6 locations of bison remains 1980, Reynolds and Lindsay 1999). In Cathedral during 3 seasons of investigation; all remains Gorge these deposits comprise the Pliocene- are located within 1 km (0.62 miles) of the age Panaca Formation (Stewart 1980). park picnic area. Remains consist of bones, Cathedral Gorge formed as a result of stream many of which were articulated, vegetative erosion during the Pleistocene and Holocene matter, hide, hair, and Diptera (flies) pupal following the integration of the Meadow Valley casings. We excavated 2 of the locations (Fig. 2 wash fluvial system into the Muddy River–Vir- shows the excavation at location 5); bone sam- gin River–Colorado River system, which re- ples were collected from the other 4 locations. sulted in the exposure of the thick lakebed sequence (Tschanz and Pampeyan 1970). The resulting landscape within the gorge consists STUDY SITE of spectacular badlands topography along the Cathedral Gorge State Park is located in margins of the broad, low-relief valley floor Lincoln County, Nevada, about 260 km (160 and isolated Pleistocene and Holocene fluvial miles) northeast of Las Vegas, near the town of terrace remnants within the gorge. Erosional Panaca. The park is situated in the northwest- remnants of colluvial mantled bedrock slopes,

1Desert Research Institute, 755 E. Flamingo Road, Las Vegas, NV 89119-7363. 2Washington State University, Anthropology Department, Pullman, WA 99164-4910.

24 2005] BISON DEPOSITS IN SOUTHEASTERN NEVADA 25

Fig. 1. Bison locations in Cathedral Gorge State Park. now detached from the rim of the gorge, are Vegetation common along the margins of the gorge. Fluvial Vegetation in the area today consists of terraces associated with Meadow Valley wash northern desert and salt desert shrubs and near Panaca and numerous smaller drainages grasses. Common plants include black sage- that dissect the piedmont surfaces are present brush (Artemisia nova), big sagebrush (Arte- outside park boundaries. Resistant, nonpedo- misia tridentata), barberry (Berberis fremontii), genic carbonate lenses within the Panaca For- greasewood (Sarcobatus vermiculatus), white mation act as protective caps for less resistant, sage (Eurotia lanata), saltbush (Atriplex can- underlying sediments. Differential erosion has escens and A. confertifolia), rabbitbrush (Chrys- resulted in the development of large areas of othamnus nauseosus), ephedra (Ephedra tor- relatively flat, carbonate-capped surfaces that reyana and E. viridis), spiny hopsage (Grayia are at different levels above the gorge floor. spinosa), cliffrose (Purshia tridentata), and snake- Differential erosion promotes preservation of weed (Gutierrezia sarothrae). Utah juniper ( Jun- tall spires of sediment that impart the cathe- iperus osteosperma) is widespread on well- dral-like appearance to the gorge (Fig. 3). drained alluvial fans. Grasses such as Stipa, Adjacent to the spires are deep, vertical Muhlenbergia, Sporobolus, Hilaria, and the piping vents, which are prominent features in introduced Bromus tectorum are also common. the lower canyon walls. Most pipes are part of an interconnected network of subterranean Depositional Context of tunnels and void spaces formed by mechanical Bison bison Bones expansion of clay minerals and high infiltra- All 6 locations of bison skeletons were with- tion of water through materials with intrinsi- in pipes developed in the Tertiary lake sedi- cally low permeability (Bryan and Yair 1982). ments. Some remains were found buried in Numerous pipes in the gorge are of sufficient pipes 15–30 m above the present valley floor, diameter to serve as effective traps of large whereas others were found in large pipe cham- debris, including large animals. bers only a few meters above the valley floor. 26 WESTERN NORTH AMERICAN NATURALIST [Volume 65

Fig. 2. Exposed bison remains at location 5. The relatively fine-grained deposits, having pebbles and gravel suspended in a fine-grained matrix, are consistent with a fluid having a high-sediment concentration similar to fine-grained debris flows. They contrast with the sedimentology of the adjacent bedrock.

We observed some bone fragments and exotic sible, bones that clearly conjoined were iden- clasts in the small alluvial fans at pipe outlets, tified. Dental eruption sequences and epiphy- indicating transport from elsewhere in the seal fusion of selected elements were used to piping network as well as from surfaces above determine age composition of the animals in the rim of the gorge. the assemblage (Fuller 1959, Duffield 1973, Both excavated locations were located at Reher 1973, Frison 1982). Bone surfaces were the same stratigraphic level and immediately visually examined for presence of taphonomic below an indurated calcium carbonate layer. damage, such as weathering, carnivore, and The carbonate layer approaches 10 cm in thick- cultural modification. ness and ranges from nearly pure carbonate to We made taxonomic identification by refer- a heterogeneous mix of fine sand, silt, and clay. ring to comparative collections housed in the This layer does not display characteristic cal- Department of Anthropology at Washington cium carbonate morphology consistent with a State University and by using published man- pedogenic origin (e.g., Birkeland 1999). Physical uals (Lawrence 1951, Olsen 1959, 1960, Brown appearance is suggestive of a chemical precip- and Gustafson 1990, Balkwill and Cumbaa itate formed in a lacustrine setting (e.g., Fouch 1992). All identifiable bones clearly came from and Dean 1982). large-bodied bovids of Quaternary age corresponding to size-class 6 (>325 kg following METHODS Thomas 1969). Two potential species, Ameri- can bison (Bison bison) and cattle (Bos taurus), Faunal Assemblage from could be represented by bone specimens in Cathedral Gorge this family and size range. We examined each bone specimen to ascer- American bison and cattle are osteologically tain taxonomic affiliation, skeletal part and very similar. Both are within the larger tribe segment, and bone symmetry. Whenever pos- Bovini, share the same basic osteological 2005] BISON DEPOSITS IN SOUTHEASTERN NEVADA 27

Fig. 3. Badlands topopography of Cathedral Gorge State Park. Note thin, discontinuous, resistant carbonate layers within the fine-grained sediments of the Panaca Formation (Pliocene). The resistant layers form protective caps for many of the spires. Resistant layers also form weak bridges between pipes of differing sizes. structure, and can overlap in body size. The Stomach Contents positive distinction between cattle and bison Organic remains from 3 discrete areas were must be based on the presence of several excavated from location 5. We processed sam- qualitative morphological features (Lawrence ples to obtain plant remains by soaking them 1951, Olsen 1959, 1960, Balkwill and Cumbaa 1992). Balkwill and Cumbaa (1992) used the in distilled water until contents disaggregated. maximum likelihood theory to quantitatively We screened plant remains through 1.7-mm, evaluate 192 morphologic and diagnostic fea- 710-mm, and 355-mm nested sieves and air- tures identified by Lawrence (1951), Olsen dried them. Contents were sorted and identi- (1960), and others. By scoring the success rate fied using a 2X magnification lamp and a of different qualitative features, they were able microscope at 6–40X magnification. Plant iden- to identify the most useful diagnostic charac- tifications were established by using the paleo- teristics for distinguishing cattle from bison. botanical collection in the Laboratory of Paleo- In our analysis bone specimens were assigned ecology at the Desert Research Institute. taxonomic designations if they displayed 2 or more diagnostic characteristics, and ≥1 of these RESULTS was associated with a high success rate as Radiocarbon-Dating identified by Balkwill and Cumbaa (1992). Specimens that displayed only 1 characteristic We radiocarbon-dated 7 samples of bone and of either bison or cattle or no diagnostic fea- plant remains from 3 locations (Fig. 4). These tures were assigned to the category size-class indicate at least 3 periods of bison existence in 6/bovid. Cathedral Gorge. Table 1 provides summary 28 WESTERN NORTH AMERICAN NATURALIST [Volume 65

Fig. 4. Graphic depiction of radiocarbon data demonstrating the episodic occurrence of bison in the study area. See Figure 1 for location numbers.

data with calibrations based on Stuiver and class 6/bovid elements are a femur and a meta- Reimer (1993) and Stuiver et al. (1998). carpal that display features associated with cattle. Because each of these specimens displays Bison Bones only 1 diagnostic feature associated with cat- The faunal assemblage from Cathedral Gorge tle, they were classified as size-class 6/bovid. consists of approximately 1445 bones and Of the identifiable bones, 20% (n = 208) bone fragments. Bone preservation at the site display qualitative features ascribable to bison. is excellent, and 73% (n = 1059) of the bone Most bones that could be identified positively specimens are identifiable to taxonomic level as bison are from adult specimens. While only or body class size and body part. Of the identi- a small percentage of the entire assemblage fiable bones, 80% (n = 851) are assigned to can be identified definitely as bison, it is prob- the category size-class 6/bovid. More than half able that some, if not all, of the size-class (n = 506) of the bones in this category are 6/bovid subadult and juvenile specimens are from subadult and juvenile individuals. These also bison. Table 2 shows the number of iden- bones lack complete epiphyseal fusion and tifiable specimens (or NISP) and minimum diagnostic landmarks and cannot be identified number of elements (MNE) by skeletal part definitely as bison or cattle. Among the size- for size-class 6/bovid and bison specimens. 2005] BISON DEPOSITS IN SOUTHEASTERN NEVADA 29

TABLE 1. Summary table of radiocarbon dates from Cathedral Gorge. Laboratory Corrected 14C age ± cal B.P. cal A.D. Material Location number 637 62 650 580 570 1300 1370 1380 metacarpal 1 DRI 3215 675 37 650 1300 vegetative 1 DRI 3187 806 62 730 720 710 1230 1240 metacarpal 3 DRI 3214 450 60 510 1440 stomach contents 5 BETA 141818 510 40 530 1420 vegetative 5 BETA 122063 640 40 650 580 570 1300 1370 1380 insect casings 5 BETA 141819 650 50 650 580 570 1300 1370 1380 vegetative 5 BETA 122062

TABLE 2. Skeletal element representation for size-class 6/bovid and bison. Element NISPa MNEa NISPb MNEb Cranium 17 6 38 2 Mandiblec 54176 Hyoid 0 0 13 13 Cervical vertebrae 11 11 102 31 Thoracic vertebrae 4 4 58 55 Lumbar vertebrae 7 4 34 14 Caudal vertebrae 0 0 41 41 Ribd 44 39 182 36 Sternum 0 0 30 29 Scapula 5 5 5 3 Humerus 6 6 11 4 Radioulna 8 6 14 6 Carpals 41 41 16 16 Metacarpal 6 5 5 5 Innominate 0 0 23 4 Femur 2 2 18 7 Patella 3 3 9 9 Tibia 4 3 14 5 Tarsals 24 24 17 17 Metatarsal 4 4 7 1 Phalanx 1 7 7 33 30 Phalanx 2 2 2 37 35 Phalanx 3 8 8 25 24 Sesmoid 0 0 32 32 Vertebrate fragment 0 0 66 0 Metapodial fragment 0 0 4 0 Total 208 851 aBison NISP. bSize-class 6/bovid NISP. cMNEs represent half mandibles. dIncludes costal cartilage.

Nearly all skeletal elements are represented in tooth eruption (Fuller 1959, Reher 1973, Frison the assemblage, which is consistent with a death 1982). Similarly, the age when different skele- assemblage containing complete carcasses. tal elements fuse is also known for bison (Duffield 1973). Since the age when teeth erupt Age Categories and bones fuse can be impeded by nutritional A large proportion of the assemblage (53%) and other environmental factors, age ranges consists of subadult and juvenile specimens. given here are merely approximations of the Age composition of the Cathedral Gorge spec- composition of the assemblage. imens is reflected by tooth eruption sequences There are at least 4 complete skulls (cra- and degree of bone epiphyseal fusion. As with nium with mandible) and portions of maxilla most animals, bison have a known sequence of and mandibles with full dentition. Table 3 shows 30 WESTERN NORTH AMERICAN NATURALIST [Volume 65

TABLE 3. Dental eruption sequences displayed by the specimens at Cathedral Gorge. Abbreviations: PCI: permanent central incisor; LPP2/3: lower permanent premolar; LDP4: lower deciduous premolar 4; LM1/2/3: lower molars; UDP2/3/4: upper deciduous premolars; UPP2/3: upper permanent premolars; UM1/2/3: upper molars. Specimen description Approximate agea Skull/Mandible PC1 erupting, LPP2, LPP3, LDP4, LM1, LM2 erupted, LM3 erupting with first 2 cusps visible 30–33 months Skull/Mandible UDP2, UDP3, UDP4 erupted, UM1 just erupting above alveolar (no wear) 5–6 months Skull/Mandible Complete adult dentition >54 months Skull/Mandible UPP2, UPP3 erupting, UDP4, UM1, UM2 (little wear) erupted, UM3 erupting (first 2 cusps visible) 30–33 months Maxilla UDP2, UDP3, UDP4, UM1 erupted, UM2 erupting (UPP2, UPP3 visible under UDP2, UDP3) 18–21 months Maxilla Adult dentition ≥54 months Maxilla Adult dentition ≥54 months Mandible Adult dentition ≥54 months aData derived from Fuller 1959, Reher 1973, Frison 1982. tooth-eruption data and age categories for these 2–4 months old. Based on overall bone size specimens. The dentition indicates that a range and the lack of cortical bone, these elements of individuals from 5–6 months old to adult is appear to be from the same animal; most were represented in the assemblage. Table 4 lists recovered from the same excavated area. These the degree of epiphyseal fusion displayed by specimens include unfused vertebral centra and limb bone elements in the assemblage. Adult neural arches, unfused innominate, scapulae, limb bones are represented by full fusion of limb bones, crania fragments, and loose decid- proximal and distal ends to the limb bone uous teeth. The vertebrae generally fuse in shaft. These elements are limited in number, bison by 4 months of age (Duffield 1973). and some are clearly pairs from the same indi- Additionally, the vertebrae, innominate, and vidual. Most of the subadult limb bones (except scapula compare very favorably in size and the metapodials) lack fusion on both ends of overall morphology (i.e., degree of cortical the bones and are represented by loose, unfused bone covering the shaft) with a 2- to 4-month- proximal and distal epiphyses and limb bone old calf in the Washington State University shafts with exposed metaphyses. Lack of limb reference collection. bone fusion indicates that most of these indi- Age composition displayed by the speci- viduals are <72 months old at death. If it is mens is summarized in Table 5. Age ranges assumed that all limb bones came from the represented by the dental-eruption sequences same group of individual animals, then, based are supported by some of the epiphyseal fusion on the presence of unfused humeri, most speci- information, albeit less precisely. Two dental mens are from animals that were <36 months sequences indicate a range of 30–33 months. old at death. However, 2 radioulna specimens, This is matched by some of the epiphyseal a right and left pair from the same animal, fusion data indicating that at least 2 individu- consist of shafts with fused proximal ends but als are under 36 months. Dental-eruption evi- unfused distal ends and indicate the presence dence also indicates that at least 1 individual of at least 1 individual between 48 and 72 is 18–21 months old. Based on the number of months old at death. intact skulls and maxilla and mandibular por- In addition to specimens listed in Tables 3 tions, the assemblage contains the remains of a and 4, at least 116 size-class 6/bovid skeletal minimum number of individuals (MNI) of 3 elements are from 1 individual, approximately adults, 4 subadults, and 2 juvenile animals. 2005] BISON DEPOSITS IN SOUTHEASTERN NEVADA 31

TABLE 4. Epiphyseal fusion data for size-class 6/bovid and bison. Element Age at fusion (months)a NISP unfused NISP fused Humerus 5b 4 Proximal humerus 36 5 Distal humerus 36 2 Radioulna 1b 4 Proximal radioulna 48 1 Distal radioulna 72 4 Metacarpal 0 5 Distal metacarpal 48 5 Femur 6b 2 Proximal femur 54 7 Distal femur 60 5 Tibia 5b 3 Proximal tibia 60 3 Distal tibia 5 Metatarsalb 03 Distal metatarsal 48 6 aAfter Duffield 1973 bShafts with exposed metaphyses

Taphonomic Damage Stomach Contents The assemblage displays remarkably few Organic material in all samples was well taphonomic indicators; no evidence of cultural preserved, generally uniform in size (approxi- or carnivore modification exists. Absence of mately 2–8 mm), and consisted of plant mater- the latter is particularly intriguing because it ial in 3 discrete boluses. It is probable that the might be expected that local scavenging birds material was contained in the omasum or the and mammals would ravage exposed carcasses. abomasum portion of the stomachs as it was Lack of carnivore modification suggests a rapid found in conjunction with the rib cavity. Its burial of the carcasses after death. This is fur- uniform size; degraded, crushed, or shredded ther suggested by the excellent preservation of nature (appears as if it had been chewed and the bones (and the presence of articulated car- partially digested); and blunt ends are consis- casses). A handful of specimens (NISP = 18) tent with this part of the digestive system. collected from the surface of location 6 display Vegetal material consisted mostly of twigs extensive weathering. Weathering displayed rather than forbs and grasses. This is unusual by these specimens is consistent with Behrens- because bison usually graze. However, it is pos- meyer’s (1978) stages 4 and 5. The bone sur- sible that this is a result of differential preser- faces are heavily exfoliated, and portions of vation and not representative of the entire diet. the shaft are splintered and falling apart. Macrobotanical fragments identified in the Behrensmeyer (1978) found that this stage of bison stomach contents are all from plants that weathering occurs on bones exposed 6–15 grow in Cathedral Gorge State Park today years after death of the animal. Since Behrens- (Table 6). Three families are represented: Poa- meyer’s weathering experiments were con- ceae (grasses), Asteraceae (asters), and Cheno- ducted in East Africa, it is unknown if this span podiaceae (chenopods). accurately reflects the same time interval of Insect Casings exposure for these specimens in the Great Basin. Nevertheless, weathering indicates a Pupal cases of the order Diptera (flies) longer interval of exposure for these speci- were associated with matter that we believe to mens than the rest of the assemblage. Most be stomach contents. Diptera usually infest bones in the assemblage are very well pre- carcasses within minutes of death and lay eggs served, and it is likely that these few speci- in crevices/creases (out of direct sunlight) of mens, collected from the surface, weathered the flesh of the carcass (nostrils, anus, etc). after the bones were buried and subsequently Eggs hatch in a few hours, and the maggots go exposed by erosion. Weathering on these through several life stages within days of specimens is likely a recent event. hatching. The last stage before adulthood is 32 WESTERN NORTH AMERICAN NATURALIST [Volume 65

TABLE 5. Summary of the age groups represented at TABLE 6. Identified species from stomach contents. Cathedral Gorge. Grayia Diptera Age range (months) MNI Location Poaceae A. spinescens spinosa pupal cases Dental evidence 1 twigs twigs present 5–6 months 1 2 lemma twigs present 18–21 months 1 3 twigs present 30–33 months 2 >54 months (adult) 3 Epiphyseal fusion <36 months 2 Van Vuren and Dietz (1993) report 3 bison 48–72 months 1 specimens in or near the basin drained by the Adult (>72 months) 3 Humboldt River. One is associated with a radio- ± Other evidence carbon-date of 950 60 B.P. (BETA 555844). 2–4 months 1 Examination of multiple archaeological sites by Aikens (1965) in nearby southwestern Utah finds no bison represented in the faunal assem- the puparium stage (e.g., pupae found at blages. Further north in Ute territory, numer- Cathedral Gorge). Total life span varies depend- ous researchers have reported bison use (e.g., ing on species and environmental conditions Steward 1938, Jennings 1978, Janetski 1991). (hotter is better). For North America the life Studies that examine cultural use of bison span (egg to adult) for various nonsubtropical have found fluctuations in the frequency of species is between 13 and 35 days (Haskell et populations during periods of the Holocene. al. 1997). Creel (1991) indicates one of these fluctuations takes place around A.D. 1300 where “. . . the DISCUSSION few previous centuries [were] a period of especially low frequency in at least many portions American bison or buffalo are the largest of the southern Plains.” He correlates this members of the family Bovidae indigenous to shift with technological changes, noting the the Great Basin during the late Quaternary. reappearance of end scrapers and the 1st While prehistoric evidence exists that bison appearance of 2- and 4-edge beveled knives. once roamed parts of Nevada, based on remains Ricklis (1992) corroborates this position with from sites such as Gatecliff Shelter (Thomas the additional observation that peoples in his 1983), Hidden Cave (Grayson 1985), O’Malley study area of south central coastal Texas adopted Rock Shelter (Fowler et al. 1973), and James a tool kit to take advantage of the expanding Creek Shelter (Grayson 1990), all predate A.D. bison populations. 1000 (Grayson 1990, Van Vuren and Dietz Historical references to bison in portions of 1993, Lupo and Schmitt 1997). Archaeological northern Nevada are often vague and impre- investigations near Cathedral Gorge have not cise (Roe 1951, Grayson 1990). Few actual his- revealed any evidence of bison utilization torical sightings of bison are recorded in (Elston et al. 1987). Moreover, a lack of infor- Nevada (Seton 1929, Hall 1946, Roe 1951) and mation on use of bison among the southern none exist for southern Nevada. Given that the Nevada aboriginal peoples suggests bison were youngest date range derived from Cathedral not part of the economy of the Cathedral Gorge Gorge is A.D. 1400–1550, it is also possible region. that part of the assemblage contains cattle. McDonald (1981) mentions 6 locations in Cattle were first introduced to North America Nevada where bison remains have been radio- in A.D. 1521 and transported to the southwest carbon-dated, all locations exceeding 10,000 in 1540 when Francisco Vasquez de Coronado years B.P. Steward (1938), following Seton brought 500 cattle to southern Arizona (Olsen (1929), indicates the former range of the bison 1960, Rouse 1973). Some of these animals pos- extended into northern Nevada around A.D. sibly escaped and quickly formed feral popula- 1500. He provides corroborative, anthropolog- tions that expanded north as cattle did in north- ical evidence to substantiate this position ern Mexico, Texas, and New Mexico (Rouse (including informant-based data on their pres- 1973). However, the earliest historic and ethno- ence along the Humboldt River and in the graphic references by explorers and indige- region of Steptoe Valley in eastern Nevada). nous peoples make no mention of feral cattle 2005] BISON DEPOSITS IN SOUTHEASTERN NEVADA 33 populations in southern Nevada. The most likely earliest Holocene expansion of bison into this interval for the introduction of cattle into territory. The middle death event, represented southern Nevada occurred with the opening of by 4 dates, occurred about 710–570 calendar the Old Spanish Trail in the 1700s. The Span- years B.P. and synchronizes well with observa- ish and commercial traders regularly used the tions by Creel (1991), Ricklis (1992), and others. Old Spanish Trail from 1831 to 1848 and likely The youngest death event, represented by brought cattle with them. Early Mormon settlers 2 dates, occurred around 400–550 calendar in southern Utah also brought cattle in the years B.P. 1850s (Euler 1966). Cattle remains could have potentially been introduced to the Cathedral CONCLUSIONS Gorge region more recently. In the recent past local residents used this area to discard farm Cathedral Gorge bison deposits represent animal carcasses. The analysis, however, indi- new evidence for extending the bison range cates that with the exception of 2 specimens, into southeastern Nevada. This occurrence no conclusive evidence exists for the presence happened, for the most part, at a time when of cattle in the assemblage at Cathedral Gorge. bison were expanding into new territory else- Age ranges displayed by the specimens give where in North America. Unlike other locations a rough estimate for the timing of bone depo- where human predators followed, no evidence sition for at least part of the assemblage. The exists that cultural groups of the region caused presence of 2 juvenile specimens, aged 2–4 the deaths of Cathedral Gorge bison or scav- and 5–6 months, is suggestive of the presence enged their carcasses. One could speculate of cow and calf social groups at Cathedral that the animals may have been attracted to Gorge. Juvenile bison (and cattle) in this age the carbonate layers that served as a salt lick range maintain a close relationship with their of sorts (see McHugh 1958 for bison use of salt mothers (McHugh 1958). If these juveniles licks). The periodicity, lack of taphonomic were born between mid-April and May, a peak marks, and presence of fly pupal casings indi- period for births among some wild bison pop- cate the bison were periodically trapped and ulations (McHugh 1958), then these remains died within an unspecified time frame, their were deposited sometime between July and carcasses exposed for relatively short periods. November. Assuming the same birthing period In this environment the slightest rainfall can for the individual 18–21 months old, then the become a flash flood associated with a flow death might have occurred some time in the having elevated sediment concentrations simi- winter period (between November and Febru- lar to debris flows, which can rapidly bury ani- ary). For the 2 individuals 30–33 months old, mals already lodged in the confined space of a death could have been at the same time. pipe. Evidence supporting this idea consists of The Diptera pupal casings indicate exposed pipes completely filled with matrix-supported carcasses, but the presence of well-preserved pebbles and gravel similar to the deposits asso- stomach contents along with the geologic and ciated with debris flows. For a certain number geomorphic context suggests exposure of rela- of individuals, it is plausible that some decom- tively short duration. Lack of taphonomic indi- posing, articulated skeletal remains washed cators such as fractured bones suggests that into the pipes from the surface above the the bison did not run or fall off the cliff and pipes. Similarly, previously buried remains were not trapped or struggling before death. may have been reworked during episodic Sediment filling the pipes throughout the area transport in the pipes. is consistent with mudflow or debris-flow McHugh’s (1972) documentation of 2000 material that filled the pipes rapidly, perhaps bison becoming mired in mud flats of the Platt during localized cloudbursts that could have River, Nebraska, in 1867 may serve as an caught the animals unaware and unable to appropriate analogy. While many eventually escape. got out, some did not. Similarly, he discusses Radiocarbon ages indicate that 3 “death bison dying in mud bogs in Yellowstone where events” occurred between about 850 and 400 calves are caught when they follow cows into calendar years B.P. The oldest death event, the marshes. Age distribution of the Cathedral based on 1 date, occurred between 850 and Gorge population appears to mimic Child’s 710 calendar years B.P. and may represent the (1997) description of American moose that die 34 WESTERN NORTH AMERICAN NATURALIST [Volume 65 of exhaustion and starvation after being caught FOWLER, D.D., D.B. MADSEN, AND E.M. HATTORI. 1973. in bogs, snow, mineral licks, ponds, etc. Prehistory of southeastern Nevada. Publications in the Social Sciences 6, Desert Research Institute, McHugh specifies that calves are especially Reno, NV. susceptible to getting caught. FRISON, G.C. 1982. Bison dental studies. In: G.C. Frison and D. Stanford, editors, The Agate Basin site. Acad- ACKNOWLEDGMENTS emic Press, New York. FULLER, W.A. 1959. Horns and teeth as indicators of age Funding for this project was provided by in bison. Journal of Wildlife Management 23:342–344. GRAYSON, D.K. 1985. The paleontology of Hidden Cave: the Nevada State Parks Cooperative Associa- birds and mammals. Pages 125–161 in D.H. Thomas, tion, Nevada Applied Research Initiative, and editor, The archeology of Hidden Cave. Anthropo- Lander Endowment Funds. The authors thank logical Papers of the American Museum of Natural Cathedral Gorge State Park personnel, espe- History 61, Part 1, New York. ______. 1990. The James Creek Shelter mammals. Pages cially Alan Newberry and Barbara Rohde. 87–98 in R.G. Elston and E.E. Budy, editors, The Thanks are extended to David Gilette, Law- archaeology of James Creek Shelter. University of rence Stevens, and an anonymous reviewer, Utah Anthropological Papers 115, Salt Lake City. whose comments and suggestions greatly HALL, E.R. 1946. Mammals of Nevada. University of Nevada improved this paper. Additionally, thanks go to Press, Reno. HASKELL, N., R.D. HALL, V.J. CERVENKA, AND M.A. CLARK. Mike Auerbach for his support. 1997. On the body: insects’ life stage presence and their post mortem artifacts. Pages 415–448 in W. LITERATURE CITED Haglund and M. Sorg, editors, Forensic taphonomy: the postmortem fate of human remains. CRC Press, AIKENS, C.M. 1965. Excavations in southwest Utah. Uni- Boca Raton, FL. versity of Utah, Anthropological Papers 76, Salt Lake JANETSKI, J.C. 1991. The Ute of Utah Lake. University of City. Utah Anthropological Papers 116, Salt Lake City. BALKWILL, D.M., AND S.L. CUMBAA. 1992. A guide to the JENNINGS, J.D. 1978. Prehistory of Utah and the eastern identification of postcranial bones of Bos taurus and Great Basin. University of Utah Anthropological Bison bison. Canadian Museum of Nature, Syllogeus Papers 98, Salt Lake City. 71, Ottowa. LAWRENCE, B. 1951. Post-cranial skeletal character of BEHRENSMEYER, A.K. 1978. Taphonomic and paleoecologic deer, pronghorn, and sheep-goat with notes on Bos information from bone weathering. Paleobiology 4: and Bison. Peabody Museum, Harvard University 150–162. Papers 35:9–43. BIRKELAND, P.W. 1999. Soils and geomorphology. 3rd edi- LUPO, K., AND D. SCHMITT. 1997. On late Holocene vari- tion. Oxford University Press, New York. ability in bison populations in the northeastern Great BROWN, C.L., AND C.E. GUSTAFSON. 1990. A key to the Basin. Journal of California and Great Basin Anthro- postcranial skeletal remains of cattle/bison, elk and pology 19:50–69. horse. Washington State University Laboratory of MCDONALD, J.N. 1981. North American bison: their classi- Anthropology 57. fication and evolution. University of California Press, BRYAN, R., AND A. YAIR. 1982. Perspectives on studies on Berkeley. badland geomorphology. Pages 1–12 in R. Bryan and MCHUGH, T. 1958. Social behavior of the American buf- A. Yair, editors, Badland geomorphology and piping. falo (Bison bison bison). Zoologica 43:1–40. GEO Books, Norwich, England. _____. 1972. The time of the buffalo. Alfred A. Knopf, CHILD, K. 1997. Incidental mortality. Pages 275–302 in A. New York. Franzmareny and C. Schwartz, editors, Ecology and OLSEN, S.J. 1959. Similarity in skulls of the Bison and the management of North American moose. Smithson- Brahman. American Antiquity 24:321–322. ian Institution Press, Washington, DC. ______. 1960. Post-cranial skeletal characters of Bison and CREEL, D. 1991. Bison hides in late prehistoric exchange Bos. Peabody Museum, Harvard University Papers in the southern plains. American Antiquity 56:40–49. 35(4):1–61. DUFFIELD, L. 1973. Aging and sexing the post-cranial skele- REHER, C.A. 1973. Appendix II: The Wardell bison bone ton of bison. Plains Anthropologist 18:132–139. sample: population dynamics and archaeological im- ELSTON, R.G., K. JUELL, D. SCHMITT, AND M. DREWS. 1987. plication. In: The Wardell buffalo trap 48SU301: com- Intensive investigation of seven sites on Panaca Sum- munal procurement in the Upper Green River basin, mit. Pages 145–206 in R.G. Elston and K. Juell, edi- Wyoming. University of Michigan, Anthropological tors, Archaeological investigations at Panaca Summit. Papers of the Musuem of Anthropology 48:89–105. Bureau of Land Management Cultural Resource REYNOLDS, R.E., AND E.H. LINDSAY. 1999. Late Tertiary Series 10, Reno, NV. basins and vertebrate faunas along the Nevada-Utah EULER, R. 1966. Southern Paiute ethnohistory. University border. Pages 469–478 in D.D. Gillette, editor, Verte- of Utah Anthropological Papers 78, Salt Lake City. brate paleontology in Utah. Utah Geological Survey, FOUCH, T.D., AND W.E. DEAN. 1982. Lacustrine and asso- Miscellaneous Publication 99-1. ciated clastic depositional environments. Pages 87– RICKLIS, R.A. 1992. The spread of a late prehistoric bison 114 in P.A. Scholle and D. Spearing, editors, Sand- hunting complex: evidence from the south-central stone depositional environments. American Associa- coastal prairie of Texas. Plains Anthropologist 37: tion of Petroleum Geologists, Tulsa, OK. 261–273. 2005] BISON DEPOSITS IN SOUTHEASTERN NEVADA 35

ROE, F.G. 1951. The North American buffalo: a critical THOMAS, D.H. 1969. Great Basin hunting patterns: a study of the species in wild state. 2nd edition. Uni- quantitative method for treating faunal remains. versity of Toronto Press, Toronto. American Antiquity 34(4):392–401. ROUSE, J.E. 1973. World cattle III: cattle in North Amer- ______. 1983. The archaelogy of Monitor Valley 2: Gate- ica. University of Oklahoma, Norman. cliff Shelter. Anthropological Papers of the American SETON, E.T. 1929. Lives of game animals. New York. Museum of Natural History 59:1. STEWARD, J.H. 1938. Basin-plateau aboriginal sociopoliti- TSCHANZ, C.M., AND E.H. PAMPEYAN. 1970. Geology and cal groups. Smithsonian Institution Bureau of Amer- mineral deposits of Lincoln County, Nevada. Nevada ican Ethnology Bulletin 120, Washington, DC. Bureau of Mines Bulletin 73:188. STEWART, J.H. 1980. Geology of Nevada. Nevada Bureau VAN VUREN, D., AND F.C. DIETZ. 1993. Evidence of Bison of Mines and Geology Special Publication 4:136. bison in the Great Basin. Great Basin Naturalist 53: STUIVER, M., AND P. J . R EIMER. 1993. Extended 14C data base 318–319. and revised CALIB 3.0 14C age calibration program. Radiocarbon 35:215–230. Received 2 October 2002 STUIVER, M., P.J. REIMER, E. BARD, J.W. BECK, G.S. Accepted 13 April 2004 BURR, K.A. HUGHEN, B. KROMER, ET AL. 1998. INT- CAL98 radiocarbon age calibration, 24,000-0 cal B.P. Radiocarbon 40:1041–1083. Western North American Naturalist 65(1), © 2005, pp. 36–44

HOME RANGE AND SEASONAL MOVEMENTS OF COLUMBIAN SHARP-TAILED GROUSE ASSOCIATED WITH CONSERVATION RESERVE PROGRAM AND MINE RECLAMATION

Jennifer H. Boisvert1,3, Richard W. Hoffman2,4, and Kerry P. Reese1

ABSTRACT.—During 1999 and 2000 we trapped and radio-marked 156 Columbian Sharp-tailed Grouse (Tympanu- chus phasianellus columbianus) on leks in Conservation Reserve Program (CRP, n = 73) and mine reclamation (MR, n = 83) lands in northwestern Colorado. Median spring–fall home range sizes using the 95% fixed kernal and minimum convex polygon estimators for 54 grouse were 86 ha and 61 ha, respectively. Median fixed kernal home range size did not differ between males (79 ha) and females (87 ha). Home ranges of grouse associated with CRP (112 ha) were larger than those of grouse in MR (75 ha). Directional orientation of movements from leks of capture to wintering areas was nonrandom, and there was a positive elevation gain (median = 102 m) associated with these movements. Movements did not differ between grouse captured in CRP and MR for any season but did differ between genders for the spring–fall period. Males exhibited stronger fidelity and less variation in their movements than females; 96% of males compared with only 77% of females remained within 2.0 km of their lek of capture from spring through fall. Ninety percent of females nested within 2.5 km of their lek of capture. During winter all grouse were found farther (median = 21.5 km) from lek sites than in any other season. Males remained on the breeding range longer in the fall and returned earlier in the spring than females even though they wintered similar distances away (median males = 21.5 km, median females = 21.4 km). Our findings support the 2.0-km radius used in the Habitat Suitability Index model for Columbian Sharp- tailed Grouse to assess nest and brood-rearing cover around leks, but not the 6.5-km radius used to evaluate winter cover.

Key words: Columbian Sharp-tailed Grouse, home range, seasonal movements, Conservation Reserve Program, mine reclamation, Tympanuchus phasianellus columbianus, Colorado.

Columbian Sharp-tailed Grouse inhabit sea- native habitats, such as Conservaton Reserve sonally distinct vegetation types, using grass- Program (CRP) and mine reclamation (MR) land and shrub-steppe communities during lands, for breeding, nesting, and brood-rearing spring, summer, and fall, and tall deciduous (Meints 1991, Sirotnak et al. 1991, Apa 1998, riparian and mountain shrub cover types dur- McDonald 1998, Boisvert 2002). In northwest- ing winter (Giesen and Connelly 1993). Move- ern Colorado, 44% of 133 active leks surveyed ments within and between seasonally occupied between 1997 and 2000 were located on CRP habitats vary depending on the quality and jux- (26%) or MR (18%) lands (Hoffman 2001). While taposition of these habitats (Meints et al. 1992). others have attempted to document movements Available information suggests that most grouse of Sharp-tailed Grouse in relation to leks located remain within 2.0 km of the lek where they in CRP (Meints 1991, Apa 1998, McDonald were captured from spring through fall and 1998), their results were based on small sam- within 6.5 km of the lek during winter (Marks ples of radio-marked grouse that were often and Marks 1987, Meints 1991, Ulliman 1995, biased toward one sex. There have been no Giesen 1997, Apa 1998, McDonald 1998). published studies of seasonal movements of Meints et al. (1992) used these figures as the Sharp-tailed Grouse attending leks in MR lands. basis for developing the Habitat Suitability This information is necessary for describing Index (HSI) model for Columbian Sharp-tailed habitat use patterns, implementing meaningful Grouse. management practices, and evaluating the More recently, Columbian Sharp-tailed impacts of land use changes around these non- Grouse have been documented using non- traditional lek sites. It is also important to

1Department of Fish and Wildlife Resources, University of Idaho, Moscow, ID 83844-1136. 2Colorado Division of Wildlife, 317 West Prospect Road, Fort Collins, CO 80526. 3Present address: ABR, Inc., Box 240268, Anchorage, AK 99524. 4Corresponding author.

36 2005] SHARP-TAILED GROUSE HOME RANGE AND MOVEMENTS 37 know whether the distances used to develop variations in topography, soils, moisture condi- the HSI model apply to leks in CRP and MR tions, elevation, and aspect. The natural tran- lands. sition is from big sagebrush (Artemisia triden- We monitored seasonal movements of tata) at the lower elevations to shrub-steppe, Columbian Sharp-tailed Grouse associated upland shrub, quaking aspen (Populus tremu- with CRP and MR lands in northwestern Col- loides), mixed conifer/aspen, and finally to con- orado during 1999 and 2000 in conjunction ifer at the highest elevations. The extensive with investigations of habitat use, productivity, deciduous shrub component dominated by and survival (Boisvert 2002). Here we report Saskatoon serviceberry (Amelanchier alnifolia) spring–fall home range sizes, distances trav- interspersed with sagebrush, native grasslands, eled in relation to leks of capture, and eleva- CRP, MR, aspen, and agricultural lands pro- tion changes and directional orientation of vides optimal habitat for Columbian Sharp- movements from breeding to wintering areas. tailed Grouse. Land ownership is mostly (>70%) We tested the hypotheses that home range private. Livestock grazing and coal mining are and distances traveled did not differ between the primary land uses, with some irrigated hay, genders or between grouse captured in CRP alfalfa (Medicago sativa), and dryland wheat and those captured in MR. We also address (Triticum spp.) farming. timing of movements, fidelity to lek sites, and gender segregation. In addition, we define METHODS appropriate buffer sizes around lek sites that can be used for assessing habitat use and suit- Grouse were captured using walk-in funnel ability, and for directing management of Colum- traps placed on the leks (Schroeder and Braun bian Sharp-tailed Grouse populations using 1991). During 1999 we trapped and radio- nonnative habitats such as CRP and MR. marked 50 grouse (23 females, 27 males) on 8 leks in MR and 35 grouse (22 females, 13 males) STUDY AREA on 4 leks in CRP. During 2000 our numbers were 34 grouse (22 females, 12 males) on 7 leks Our study was conducted in Routt and in MR and 37 grouse (25 females, 12 males) on Moffat Counties in northwestern Colorado 5 leks in CRP. The radio-transmitters weighed (40°22′N, 107°05′W) within the Upper Yampa <15 g and were attached with an elastic neck- River watershed. Boundaries of the 276,602- lace. Captured birds were banded with a seri- ha study area were delineated based on maxi- ally numbered aluminum leg band (size 12), mum movements of grouse from their leks of classified to gender using crown and tail capture. However, trapping and most fieldwork feather characteristics (Henderson et al. 1967), during the spring–fall period were confined to and aged based on the shape and wear of the 2 a 20,215-ha region known as Twentymile Park, distal primaries (Ammann 1944). Two age classes located 28 km southwest of Steamboat Springs were recognized for analyses: subadults (≤12 in Routt County. The Twentymile Park area is months) and adults (>12 months). Eighteen a mosaic of shrub-steppe, upland shrub, and subadult females and only 2 subadult males well-established CRP (627 ha) and MR (2513 were captured over the 2-year period. Thus, ha) lands in close proximity to each other comparisons between age classes were based (Boisvert 2002). only on females. Trapping and handling proto- Average annual precipitation ranges from cols used in this study were conducted under <26 cm near Craig in Moffat County to >127 the approval of the Colorado Division of Wild- cm at Steamboat Springs in Routt County. life and University of Idaho Animal Care and During this study mean annual precipitation Use Committees. was 51 cm and mean monthly temperature Monitoring began 7–10 days post-capture was 5°C (range = –8° to +20°C); snow depth at which time we flushed each bird to ensure was ≥3 cm for over 100 days during winter they were still alive and had adjusted to the and averaged 53 cm. transmitter. All radio-marked grouse were sub- The area is topographically diverse with sequently located twice per week during spring elevations and slopes ranging from 2000 m to and summer, once every 2 weeks during fall, 2600 m and 0° to 60°, respectively. Vegetation and at least once per month during winter. All types in the area are equally diverse due to grouse were located from the ground using 38 WESTERN NORTH AMERICAN NATURALIST [Volume 65 the loudest signal method (Springer 1979). brood-rearing season extended from 16 June Aerial searches were used on 5 occasions to to 1 September 1999, and from 8 June to 2 find missing grouse, which were subsequently September 2000, and included all locations of located from the ground. The grouse were successful females (hatched ≥1 egg) from the approached and circled within 20 m to avoid time they left the nest until brood breakup. flushing them and to minimize location error. There were no distinct changes in behavior, We obtained approximate Universal Transverse movements, or habitat use patterns of grouse Mercator (UTM) coordinates of grouse loca- between summer and fall seasons. Consequent- tions by triangulating 3–4 GPS readings using ly, these seasons were combined into a single a handheld GPS unit as we circled each bird. period (summer–fall) that ranged from 2 June The GPS unit was also used to determine the to 19 November 1999, and from 16 May to 9 elevation where the grouse was located and November 2000. This period included all loca- the direction and distance to its lek of capture. tions of males after they stopped attending Most locations (68%) were obtained between leks until they departed the breeding range. It 0930 and 1700 hours, whereas 21% of the also included all locations of females that locations were collected before 0930 and 11% hatched no eggs or lost their brood from the after 1700. time they abandoned the nest or were no Nest sites were found by monitoring females longer accompanied by chicks until they de- 2–3 times per week until they initiated incu- parted the breeding range. These females are bation. When 2 subsequent observations of a hereafter referred to as unsuccessful females. female were made at the same location, we The winter period extended from 20 November assumed she was nesting. We then circled the 1999 to 4 April 2000, and from 10 November suspected nest site at a radius of 5–10 m to 2000 until the study ended on 31 January obtain an accurate location without flushing 2001. The onset of this period was marked by the female. We recorded the UTM coordinates, obvious movements of grouse away from the distance, and compass direction to the nest CRP, MR, grassland, and shrub-steppe com- from an inconspicuously flagged location 7–10 munities on the breeding range to upland m from the nest. We continued to monitor the shrub cover types on the winter range and site until the eggs hatched or the signal indi- concluded with initiation of movements back cated the female was no longer present on the to breeding areas. nest. At this time we obtained the exact UTM We estimated home range sizes with a 95% coordinates for the nest site and determined fixed-kernal (FK) estimator (Worton 1989), using the distance to the lek of capture using the least squares cross-validation to choose kernal GPS unit. band widths. We also estimated home range We defined 5 seasons of use: breeding, nest- sizes using the Minimum Convex Polygon ing, brood-rearing, summer–fall, and winter. method (MCP; Mohr 1947) for comparison Seasons of use were defined based on changes with other studies. All home ranges were cal- in weather conditions and changes in behav- culated using the Spatial Animal Movement ior, movements, and habitat use patterns for extension (Hooge and Eichenlaub 1997) in each individual grouse. Thus, seasons of use ArcView 3.2 (ESRI 1999). We included all differed between years because of variations grouse with a minimum of 19 locations (range in weather conditions from one year to the next 19–57) during the spring–fall period in the and overlapped within years because of behav- estimation and analysis of home ranges. Win- ioral differences among individual grouse. The ter locations were not included in the home breeding season ranged from 24 April (earliest range estimate because the grouse used dis- location of radio-marked grouse) to 29 May tinctly different areas and cover types during 1999, and from 22 March to 31 May 2000, and winter (Boisvert 2002). Although we monitored included all locations of males and females the grouse during winter, we did not collect during the lekking period until they no longer enough locations to delineate their winter home consistently attended the lek (males) or initi- range due to difficulties in regularly accessing ated incubation (females). The nesting season the birds in remote areas. Likewise, we did ranged from 19 May to 8 July 1999, and from not collect enough locations to separately de- 7 May to 14 July 2000, and encompassed the lineate home ranges for breeding, brood-rear- period when females were incubating eggs. The ing, and summer–fall periods. We compared 2005] SHARP-TAILED GROUSE HOME RANGE AND MOVEMENTS 39

TABLE 1. Spring–fall home range estimates (ha) of Columbian Sharp-tailed Grouse associated with Conservation Reserve Program (CRP) and mine reclamation (MR) lands in northwestern Colorado, 1999–2000. Estimates are based on ≥19 locations per grouse.

______95% fixed-kernal ______Minimum convex polygon Category n Median Mean Range Median Mean Range Male 18 79 120 39–642 61 81 28–438 Female 36 87 170 37–777 60 108 19–581 CRP 20 112 186 39–642 64 98 19–304 MR 34 75 134 37–777 59 91 20–581 All Grouse 54 86 153 37–777 61 99 19–581

home range sizes based on age (females only), sion of the median test. Significance for all gender, and breeding habitat (CRP or MR). tests was set at 0.05. Multiple locations within seasons were used to calculate movements of individual grouse RESULTS from their lek of capture during breeding, summer–fall, and winter seasons, and from Home range size did not differ between their nest site to brood-rearing areas. Only 1 year, gender, or age class (P = 0.382–0.784); location corresponding to the nest site was thus, data were pooled by gender and age for used to calculate the distance from the lek of both years. The 95% median FK spring–fall capture for the nesting season. We first calcu- home range size estimated for 54 grouse was lated mean and median movements for indi- 86 ha (Table 1). Median home range size of vidual grouse for each season, excluding the grouse captured in CRP (112 ha) was larger nesting season. We then derived overall mean than for grouse captured in MR (75 ha), but the difference was not statistically significant and median movements for each season from (T = −1.407, P = 0.085). Although some home mean and median movements of the individ- range estimates were extremely large (range ual grouse. For the nesting season, mean and = 37–777 ha), most grouse (72%) occupied median movements were calculated from a home ranges that were <75 ha. The MCP home single estimate for each female that nested. range estimate was 61 ha, which was smaller Movements were categorized by (1) gender, (T = −18.913, P < 0.001) than the 95% FK esti- breeding habitat (CRP and MR), gender with- mate for the same 54 grouse (Table 1). in breeding habitat, and age (females only) Columbian Sharp-tailed Grouse occupied 2 for the breeding and summer–fall periods; distinct ranges corresponding to the spring– (2) breeding habitat and age for the nesting fall and winter periods. Analysis of 1775 telem- and brood-rearing periods; and (3) gender and etry locations collected during the spring–fall breeding habitat for the winter period. period indicated 85% were within 2.0 km of We made all statistical comparisons with the lek of capture (Fig. 1). In comparison, all multiresponse permutation procedures (MRPP; (n = 100) winter locations were >3.0 km from Mielke and Berry 2001) conducted in the pro- the lek of capture (Fig. 1). Directional orienta- gram BLOSSOM (Cade and Richards 2000). tion of movements from lek of capture to win- We report the standardized test statistic as T. tering areas was nonrandom (T = 3.757, P < We compared elevation change between breed- 0.001), with 16 of 30 (53%) grouse moving ing and wintering areas and differences between WSW (254°) to WNW (299°). There was a sig- FK and MCP home range estimators by using nificant positive elevation gain (T = −17.415, the MRPP test for matched pairs. The permu- P < 0.001) associated with all movements to tation version of Rao’s spacing test (Rao 1976) wintering areas. The median elevation change was used to test for nonrandom directional between spring–fall (2076–2280 m) and win- orientation of movements from leks of capture tering (2202–2593 m) areas was 102 m (range to wintering areas. For comparisons of move- 5–383 m). ments and home range sizes by gender, age, Within breeding habitat types, movements and breeding habitat, we used the MRPP ver- of grouse from their leks of capture did not 40 WESTERN NORTH AMERICAN NATURALIST [Volume 65

Fig. 1. Distribution of Columbian Sharp-tailed Grouse telemetry locations from leks of capture for the spring–fall (n = 1775 locations) and winter (n = 100 locations) periods, northwestern Colorado, 1999–2000. differ between years within seasons for either Most females (96%) raised their broods with- gender or any age category (P = 0.071–1.000). in 1.4 km of where they nested. The single ex- Therefore, data were pooled by gender and ception was a subadult female that moved her age for both years within seasons. brood 2.28 km from the nest during brood- During the breeding season, males and fe- rearing. Despite this longer movement, adult males captured in CRP moved similar distances and subadult females raised their broods within compared with their counterparts captured in similar (T = 0.739, P = 0.807) distances of MR (males: T = 0.444, P = 0.545; females: T their nest sites (Table 2). Again, there was no = 0.673, P = 0.805; Table 2). Although both difference (T = −1.082, P = 0.122) in move- sexes remained relatively close to their lek of ments between CRP and MR females during capture during the breeding season, males the brood-rearing season. The median distance moved significantly shorter distances (T = that females moved from their nest during the –11.958, P < 0.001) and exhibited less varia- brood-rearing season was 0.40 km (Table 2). tion in their movements than females (Table Males continued to show fidelity to their lek 2). No males ventured >1.3 km from their lek site throughout the summer–fall period with of capture during the breeding season. In com- 91% found within 1.5 km and 96% within 2.0 parison, 78% of the females remained within km of their lek of capture. This applied to both 1.3 km of their lek of capture during the CRP and MR males, with no difference (T = breeding season, and 85% remained within 2.5 0.689, P = 0.800) detected in their summer– km. We found no difference (T = 0.655, P = fall movements (Table 2). The only male found 0.734) in movements of adult and subadult >2.0 km from its lek during the summer–fall females during the breeding season (Table 2). period ventured 4.28 km during midsummer, Fifty (86%) of 58 females located on nests but returned to within 500 m of the lek in the were found within 2.0 km of their lek of cap- fall. ture and 52 (90%) nested within 2.5 km. We Unsuccessful females moved significantly detected no differences in movements to nest farther (T = –6.504, P < 0.001) from their lek sites between adult and subadult females (T = of capture than males during the summer–fall 0.689, P = 0.766) or between CRP and MR period (Table 2). However, 77% still remained females (T = 0.617, P = 0.709). The median within 2.0 km of their lek. There was no dif- distance moved to nest sites for all females ference in movements between age classes (T was 0.63 km (Table 2). = 0.579, P = 0.655) or between unsuccessful 2005] SHARP-TAILED GROUSE HOME RANGE AND MOVEMENTS 41

TABLE 2. Seasonal movements (km) of Columbian Sharp-tailed Grouse from leks of capture in Conservation Reserve Program (CRP) and mine reclamation (MR) lands to breeding, nesting, summer–fall, and winter use areas, and from nest sites to brood-rearing areas in northwestern Colorado, 1999–2000. Category n Median Mean Range

BREEDING AREA CRP Male 18 0.32 0.38 0.07–1.23 CRP Female 36 0.60 1.21 0.21–7.68 MR Male 24 0.36 0.44 0.08–1.27 MR Female 36 0.64 1.47 0.13–10.10 Adult Female 61 0.63 1.18 0.13–7.68 Subadult Female 11 0.56 2.20 0.20–10.10 All Males 42 0.32 0.41 0.07–1.27 All Females 72 0.63 1.34 0.13–10.10

NEST SITE CRP Female 22 0.65 1.46 0.09–8.17 MR Female 36 0.62 1.21 0.12–11.30 Adult Female 49 0.63 1.15 0.09–8.17 Subadult Female 9 0.47 2.16 0.12–11.30 All Females 58 0.63 1.30 0.09–11.30

BROOD-REARING AREA CRP Female 5 0.48 0.53 0.30–0.81 MR Female 20 0.38 0.53 0.10–2.28 Adult Female 19 0.40 0.44 0.10–1.19 Subadult Female 6 0.55 0.81 0.28–2.28 All Females 25 0.40 0.53 0.10–2.28

SUMMER–FALL AREA CRP Male 11 0.40 0.77 0.21–4.28 CRP Female 13 0.85 2.29 0.22–7.50 MR Male 12 0.38 0.52 0.27–1.53 MR Female 28 0.82 1.41 0.25–10.27 Adult Female 34 0.82 1.39 0.22–7.50 Subadult Female 7 0.89 3.18 0.25–10.27 All Males 23 0.40 0.64 0.21–4.28 All Females 41 0.84 1.69 0.22–10.27

WINTER AREA All Males 13 21.50 20.01 4.18–36.50 All Females 17 21.40 22.14 3.14–41.50 CRP Grouse 5 21.50 23.65 3.14–36.50 MR Grouse 25 21.40 20.73 4.18–41.50 All Grouse 30 21.50 21.30 3.14–41.50 females from CRP and MR (T = 0.586, P = females due to small sample sizes of subadult 0.668) during the summer–fall period (Table 2). females (n = 3). Males and females were found farther from In 1999 all grouse remained on the breed- lek sites during winter than during any other ing range until 14 November but moved to season of the year (Table 2). The closest any wintering areas by 28 December. During fall grouse was located to its lek of capture during 2000, movements away from the breeding range winter was 3.14 km; 87% of the radio-marked were documented as early as 25 October, and grouse wintered >10.0 km (median = 21.50 most birds (84%) moved to wintering areas by km) from where they were trapped. Although mid-November. In both years females left the movements to wintering areas were highly breeding range before males. Grouse remained variable (Table 2), there were no differences on the winter range through mid- to late March. between males and females (T = 0.718, P = No radio-marked grouse were observed in the 0.868) or between grouse from CRP and those same flock during winter, but in 6 instances 2 from MR (T = 0.675, P = 0.727). We did not or more grouse were found wintering in the test for differences between age classes of same general area <1 km apart. Males returned 42 WESTERN NORTH AMERICAN NATURALIST [Volume 65 to the breeding range 22–28 March, whereas Despite similarities between our findings the females returned 11–17 April. Although and those of other studies, we documented few grouse (6 males, 10 females) trapped on some atypical movements. Previously, the leks in the spring survived the entire study longest movement to a nest site reported in period, those that did were all relocated on or the literature was 7.04 km (McDonald 1998). near the leks where they were captured. In Nine females in our study moved >7.0 km addition, 4 grouse monitored both winters used from their lek of capture to nest. We excluded the same wintering area, and 6 females moni- the movements of 2 of these females because tored through 2 consecutive nesting seasons we had evidence they were nonresident females nested within 250 m of their previous year’s trapped while moving through the area from nests. winter ranges. One female moved 23.1 km and the other moved 9.5 km from where they were DISCUSSION trapped. Both females were adults trapped early in the breeding season, and both local- We found no published studies that report ized their movements and nested within 1.0 home range sizes for Sharp-tailed Grouse km of other known leks. Neither female re- associated with CRP or MR lands. Our spring– turned the following year to the CRP lek where fall home range estimates, however, were it was captured; instead, both returned to the smaller than estimates reported for Columbian vicinity of the leks where they nested the pre- Sharp-tailed Grouse occupying native habitats. vious year. Four of the remaining 7 females Using the MCP method, Marks and Marks moving >7 km were adults. We are uncertain (1987) and Giesen (1997) calculated mean if they also were nonresident females because spring–fall home range sizes of 110 ha and 187 we were unable to monitor them through 2 ha, respectively, for Columbian Sharp-tailed consecutive breeding seasons to confirm their Grouse occupying native habitats in north- status. Therefore, they were included in the western Colorado and western Idaho. analyses. Our results compare favorably with other Giesen and Connelly (1993:327) stated, descriptions of seasonal movements of Colum- “Columbian Sharp-tailed Grouse seem to move farther to wintering habitats in regions lacking bian Sharp-tailed Grouse in native and non- a broad distribution of winter food resources.” native habitats (Marks and Marks 1987, Meints Results from other studies (Ulliman 1995, 1991, Giesen 1997, McDonald 1998). Median McDonald 1998), including ours, contradict spring–fall movements from leks of capture for ≤ this statement. Northwestern Colorado has all grouse in our study were 1.6 km. Seasonal not suffered from the large-scale habitat con- movements did not differ between grouse versions that have taken place in many other from CRP and those from MR but did differ regions within the range of Columbian Sharp- between gender. Approximately 85% of the tailed Grouse (McDonald and Reese 1998, grouse monitored in this study remained with- Schroeder et al. 2000, Hoffman 2001). Conse- in 2.0 km of their lek of capture throughout quently, the landscapes, particularly the upland the spring–fall period; however, males clearly shrub and aspen cover types used for winter displayed a stronger fidelity to lek sites than habitat (Boisvert 2002), have remained intact, females. comprising >25% of available cover types in Most females (86%) in our study were able this region (Hoffman 2001). Additionally, these to find suitable nest sites within 2.0 km of cover types occur in abundance within 2 km of their lek of capture whether they were trapped all leks trapped in this study, and we frequent- on CRP or MR leks. Successful females subse- ly observed unmarked grouse using these areas quently raised their broods in close proximity during winter. Yet, the closest any grouse win- (<1.4 km) to where they nested, suggesting tered to its lek of capture was 3.14 km, the they selected nest sites within or near suitable median distance being 21.50 km, and the brood-rearing habitat. Giesen (1997) reported longest movement 41.50 km. Before our study that 92% of the females he monitored in native the longest movement documented to a win- habitats nested within 2.0 km (median = 1.4 km) tering area was approximately 20 km (Meints of their lek of capture. Similarly, Meints (1991), 1991). Apa (1998), and McDonald (1998) all reported Our findings do not support the hypothesis average movements to nests of <2.0 km. proposed by Ulliman (1995:92) that the reason 2005] SHARP-TAILED GROUSE HOME RANGE AND MOVEMENTS 43 grouse do not use the closest suitable winter we documented for grouse breeding in CRP habitat is that females move farther to avoid and MR. harassment and competition for food with males on winter habitats near leks. Not only MANAGEMENT IMPLICATIONS did males and females move similar distances in our study, but they also were found in simi- Our findings suggest that the 2.0-km radius lar geographic locations. The only difference used in the Columbian Sharp-tailed Grouse we noted in their movement patterns was the HSI model (Meints et al. 1992) for assessing timing. Males remained on the breeding range nest/brood cover around leks is biologically longer and returned earlier than females. Other relevant when applied to grouse breeding in investigators, in addition to Ulliman (1995), MR and CRP. However, the 6.5-km radius have reported longer movements by females used for assessing winter cover (i.e., shrub- to wintering areas (Giesen 1997, McDonald dominated cover types) may be less relevant 1998). However, results of these studies were for grouse in Colorado. This does not mean based on small samples of grouse trapped from that shrub-dominated cover types near leks in few leks. MR and CRP are not important, because they One reason grouse may disperse throughout are (Boisvert 2002), but their value as winter the available winter range is to reduce their habitat may not be as critical as the model vulnerability to predators. During winter grouse implies. Where cover types used during win- feed in the upper branches of deciduous shrubs, ter are abundant, but not necessarily in close such as serviceberry, where they are exposed proximity to quality breeding, nesting, and and possibly more susceptible to avian preda- brood-rearing areas, the radius could be in- tion. Large concentrations of grouse in this sit- creased to 10 km or even 15 km. uation may attract predators and increase their Our findings also have implications in eval- mortality rates. Conversely, if they are dispersed uating and selecting sites for the translocation over a broad range of suitable habitats, their of Columbian Sharp-tailed Grouse. In the past chances of survival are greater. However, for the availability of suitable winter cover within this to be true, the survival advantage gained 6.5 km was an important factor in the selec- by this behavior must outweigh the increased tion of release sites. However, Gardner (1997) risk of moving long distances. According to found that by moving the release site farther Hamerstrom and Hamerstrom (1951), large- from aspen and tall shrub-dominated cover scale movements were historically common for types that support higher densities of nesting Plains (T. p. jamesi) and Prairie (T. p. campestris) raptors, the post-release survival of transplanted Sharp-tailed Grouse under pristine conditions. birds was enhanced. Ideally, releases should Thus, longer movements should not be inter- be made within 6.5 km of suitable winter habi- preted as indicative of areas having low suit- tat; however, we believe successful releases ability for Sharp-tailed Grouse. into quality nesting and brood-rearing habitats Another reason for longer movements may can be made 10–20 km from abundant winter be the lack of quality breeding, nesting, and cover. brood-rearing areas in northwestern Colorado. In searching for grouse during winter, we The introduction of CRP and MR lands into discovered that large expanses of the upland the northwestern Colorado landscape may be shrub and aspen cover types were not inhab- partially compensating for degradation and ited by grouse. We consistently found grouse in loss of native grassland and shrub-steppe cover the same areas each winter. Topographically, types used for breeding, nesting, and brood- areas used by grouse during winter tended to rearing. However, CRP and MR lands account be on north slopes with deep, soft snow. These for only about 4% of available cover types slopes were near or along ridge tops rather within the occupied range of Columbian Sharp- than on side slopes or in draws. The few grouse tailed Grouse in northwestern Colorado (Hoff- that we monitored both winters returned to man 2001). Sharp-tailed Grouse apparently will the sites they used the previous winter. Also, move longer distances to use this limited re- grouse captured in 2000 moved to the same source. Columbian Sharp-tailed Grouse breed- general wintering areas as grouse captured in ing in shrub-steppe habitats in Colorado (Giesen 1999. Finally, although suitable winter habitat 1997) and Idaho (Marks and Marks 1987) moved occurred in all directions from the breeding shorter distances to wintering areas than what range, the majority of grouse moved WSW to 44 WESTERN NORTH AMERICAN NATURALIST [Volume 65

WNW. These observations suggest that grouse and distribution. American Midland Naturalist 46: may use traditional wintering areas. Thus, 174–226. HENDERSON, F.R., F.W. BROOKS, R.E. WOOD, AND R.B. additional studies are needed to ascertain why DAHLGREN. 1967. Gendering of Prairie Grouse by grouse used these specific areas during winter crown feather patterns. Journal of Wildlife Manage- when other areas closer to the leks appeared ment 31:764–769. equally suitable. Meanwhile, efforts should be HOFFMAN, R.W. 2001. Northwest Colorado Columbian Sharp-tailed Grouse conservation plan. Northwest made to identify and protect areas where Colorado Columbian Sharp-tailed Grouse Work Group Columbian Sharp-tailed Grouse are known to and Colorado Division of Wildlife, Fort Collins. winter. HOOGE, P.N., AND B. EICHENLAUB. 1997. Animal movement extension to Arcview, version 1.1. Alaska Bio- ACKNOWLEDGMENTS logical Science Center, United States Geological Survey, Anchorage. MARKS, J.S., AND V.A. MARKS. 1987. Habitat selection by Funding was provided by the Colorado Columbian Sharp-tailed Grouse in west central Idaho. Division of Wildlife, Twentymile Coal Com- U.S. Department of the Interior, Bureau of Land pany, Seneca Coal, Trapper Mine, Inc., and Management, Boise, ID. Colowyo Coal Company. The Colorado Divi- MCDONALD, M.W. 1998. Ecology of Columbian Sharp- tailed Grouse in eastern Washington. Master’s the- sion of Wildlife also provided field equipment sis, University of Idaho, Moscow. and logistical support. Many individuals pro- MCDONALD, M.W., AND K. P. REESE. 1998. Landscape vided help with trapping, for which we are changes within the historical distribution of Colum- extremely grateful. We are also grateful to the bian Sharp-tailed Grouse in eastern Washington: Is numerous landowners who granted us access there hope? Northwest Science 72:34–41. MEINTS, D.R. 1991. Seasonal movements, habitat use, and during the study. We thank C.G. Lowman and productivity of Columbian Sharp-tailed Grouse in S.A. Ford for their contributions in the field southeastern Idaho. Master’s thesis, University of and M.L Cowardin for help with the GIS Idaho, Moscow. analyses. This is contribution 959 of the Uni- MEINTS, D.R., J.W. CONNELLY, K.P. REESE, A.R. SANDS, AND T.P. H EMKER. 1992. Habitat suitability index versity of Idaho, Forest, Wildlife, and Range (HSI) procedure for Columbian Sharp-tailed Grouse. Experiment Station. University of Idaho Forest, Wildlife, and Range Ex- perimental Station Bulletin 55, Moscow. LITERATURE CITED MIELKE, P.W., AND K.J. BERRY. 2001. Permutation methods: a distance function approach. Springer-Verlag, New AMMANN, G.A. 1944. Determining the age of Pinnated and York. Sharp-tailed Grouse. Journal of Wildlife Manage- MOHR, C.O. 1947. Table of equivalent populations of North ment 8:170–171. American small mammals. American Midland Natu- APA, A.D. 1998. Habitat use and movements of sympatric ralist 37:223–249. Sage and Columbian Sharp-tailed Grouse in south- RAO, J.S. 1976. Some tests based on arc lengths for the cir- eastern Idaho. Doctoral dissertation, University of cle. Sankhya Series Bulletin 33:1–10. Idaho, Moscow. SCHROEDER, M.A., AND C.E. BRAUN. 1991. Walk-in traps for capturing Greater Prairie Chickens on leks. Journal BOISVERT, J.H. 2002. Ecology of Columbian Sharp-tailed Grouse associated with Conservation Reserve Pro- of Ornithology 62:378–385. SCHROEDER, M.A., D.W. HAYS, M.A. MURPHY, AND D.J. gram and reclaimed surface mined lands in north- PIERCE. 2000. Changes in the distribution and abun- western Colorado. Master’s thesis, University of dance of Columbian Sharp-tailed Grouse in Wash- Idaho, Moscow. ington. Northwestern Naturalist 81:95–103. CADE, B.S., AND J.D. RICHARDS. 2000. User manual for SIROTNAK, J.M., K.P. REESE, J.W. CONNELLY, AND K. RAD- BLOSSOM statistical software. United States Geo- FORD. 1991. Characteristics of Conservation Reserve logical Survey, Fort Collins, CO. Program fields in southeastern Idaho associated with ENVIRONMENTAL SYSTEMS RESEARCH INSTITUTE. 1999. Arc- upland game bird and big game habitat use. Com- View GIS, version 3.2. Redlands, CA. pletion Report, Project W-160-R, Idaho Department GARDNER, S.C. 1997. Movements, survival, productivity, of Fish and Game, Boise. and test of a habitat suitability index model for rein- SPRINGER, J.T. 1979. Some sources of bias and sampling troduced Columbian Sharp-tailed Grouse. Master’s error in radio triangulation. Journal of Wildlife Man- thesis, University of Idaho, Moscow. agement 43:926–935. GIESEN, K.M. 1997. Seasonal movements, home ranges, ULLIMAN, M.J. 1995. Winter habitat ecology of Columbian and habitat use by Columbian Sharp-tailed Grouse Sharp-tailed Grouse in southeastern Idaho. Master’s in Colorado. Colorado Division of Wildlife, Special thesis, University of Idaho, Moscow. Report 72. WORTON, B.J. 1989. Kernal methods for estimating the uti- GIESEN, K.M., AND J.W. CONNELLY. 1993. Guidelines for lization distribution in home range studies. Ecology management of Columbian Sharp-tailed Grouse 70:164–168. habitats. Wildlife Society Bulletin 21:325–333. HAMERSTROM, F.N., AND F. H AMERSTROM. 1951. Mobility Received 30 December 2003 of the Sharp-tailed Grouse in relation to its ecology Accepted 12 April 2004 Western North American Naturalist 65(1), © 2005, pp. 45–52

WINTER SITE FIDELITY AND BODY CONDITION OF THREE RIPARIAN SONGBIRD SPECIES FOLLOWING A FIRE

Ivan A. Samuels1,2, Thomas Gardali1, Diana L. Humple1, and Geoffrey R. Geupel1

ABSTRACT.—The effects of fire on nonbreeding songbird species in riparian habitat have not been studied. We compared body condition, within-year site fidelity, and between-year site fidelity of 3 songbird species (Passerella iliaca, Fox Sparrow; Catharus guttatus, Hermit Thrush; and Regulus calendula, Ruby-crowned Kinglet) at 2 coastal riparian sites. Wildfire, which is rare in this habitat, had occurred at 1 of the sites before data collection. A significantly larger proportion of Passerella iliaca was recaptured in subsequent winters at the unburned site than at the burned site, but little difference was found between sites for Catharus guttatus or Regulus calendula. Body mass of all 3 species declined during winter at the burned site, but differences between sites were not significant. Similarly, body mass indices of new captures were lower at the burned site than the unburned site for all 3 species, but these differences were not significant. The within-year recapture rate for all 3 species combined declined at the burned site over the course of the study, possibly due to changes in vegetation structure caused by the fire. Overall, our data suggest that wintering songbirds were resilient to this disturbance, but that response to the post-fire environment differed among foraging guilds. Well-replicated studies that include pre-burn data are needed to evaluate the effects of this disturbance in riparian systems.

Key words: riparian, fire, winter songbirds, site fidelity, body condition, Passerella iliaca, Regulus calendula, Catha- rus guttatus.

Winter site fidelity of Neotropical–Nearctic foliage-gleaning insectivores declined. A decline migrant songbirds is well documented (Holmes in foliage-gleaning insectivores after fire was and Sherry 1992, Wunderle and Latta 2000, also noted by Bock and Lynch (1970), but re- Sandercock and Jaramillo 2002), and variation growth of brush supported more ground brush– in winter site fidelity may indicate habitat qual- foraging species. In general, woodpeckers, fly- ity. While the quality of breeding habitats may catchers, and seedeaters often benefit from influence nest success, the quality of nonbreed- the habitats created by large-scale fires (Hutto ing habitat during winter also has fitness con- 1995). The suitability of burned habitat is also sequences for migratory birds (Marra et al. 1998, likely to vary for different bird species with Norris et al. 2004), and large-scale disturbances time since a fire (Raphael et al. 1987). Under- may play an important role in determining standing the development of winter site fidelity habitat quality (Rotenberry et al. 1995, Brawn by birds in burned plant communities must et al. 2001). therefore include knowledge of fire history Fire is one form of disturbance that may in- and avian foraging behavior. fluence patterns of winter habitat use by birds. Most studies of songbird response to fire in For example, Kreisel and Stein (1999) found North America during winter have focused on that the loss of canopy cover and shrub under- coniferous forests, whereas study of riparian growth made burned forests more suitable for bird communities is lacking. This bias may be trunk- and branch-foraging species. Blake (1982) due to the absence of fire as a management also found certain species restricted to burned tool or as a frequent disturbance event in ripar- or unburned sites in a nonbreeding bird com- ian communities (DeBano and Neary 1996). In munity. However, while fire created suitable contrast, prescribed burning is common in con- habitat for bark-drilling species such as wood- iferous woodland where fire is often viewed peckers, bark-gleaning species preferred un- as a natural disturbance. Most studies have burned sites. The open canopy of burned sites also used point-counts or other census meth- also supported more aerial insectivores, while ods that prevent analysis of site fidelity and

1Point Reyes Bird Observatory, 4990 Shoreline Highway, Stinson Beach, CA 94970. 2Present address: 282 31st Avenue, San Francisco, CA 94121.

45 46 WESTERN NORTH AMERICAN NATURALIST [Volume 65 demography. Marked populations of birds allow aged in unburned areas during a winter period. the researcher to identify individuals that Our unburned study site is at Pine Gulch return to the same site each season or persist (PI) near the town of Bolinas, Marin County within sites through a specified period of time. (37°92′N, 122°69′W). This site lies just outside If fire has a large impact on riparian habitat, the PRNS, approximately 20 km south of MH. some migratory species may be less likely to Both the burned and unburned study sites return in subsequent seasons or be less likely contain riparian forest dominated by a canopy to persist at a single site through a season. A of red alder (Alnus rubra) and arroyo willow better understanding of the effects of fire on (Salix lasiolepis). Understory plants common riparian songbird communities is needed, even at both sites include red elderberry (Sambucus if fire disturbance is rare in this habitat. recemosa), California blackberry (Rubus ursi- In this post hoc analysis, we used 6 years of nus), stinging nettle (Urtica dioica), and poison mist-net data from 1 burned and 1 unburned hemlock (Conium maculatum). A few plant riparian site to investigate the possible effects species were dominant at only one site, e.g., of wildfire on 3 migratory songbird species coyote brush (Baccharis pilularis) and velvet during the nonbreeding season. Our objectives grass (Holcus lanatus) at MH, and gumplant were to (1) compare between-year site fidelity (Grindelia stricta) at PI. In addition, PI is in the burned versus unburned sites using re- partly bordered by a lagoon on one side and capture data across different winter seasons, mixed evergreen forest on another, while MH (2) compare within-year site fidelity between is surrounded mostly by coastal scrub. PI was the 2 sites using within-season recapture data, chosen as the unburned riparian study site and (3) determine if body condition differed because data collection had already com- between sites or changed from early to late menced at this site before the fire at MH. winter. To our knowledge, this is the 1st study examining the ecology of nonbreeding song- Study Species birds within burned riparian habitat. Three wintering bird species were selected for these analyses. Ruby-crowned Kinglet METHODS (Regulus calendula), Hermit Thrush (Catharus Study Area guttatus), and Fox Sparrow (Passerella iliaca) were selected because of sufficient capture We selected 2 riparian study sites, 1 burned and 1 unburned. The burned study site is rates; >100 individuals of each species were located at Muddy Hollow (MH) in the Point captured per site, with >200 individual P. ili- Reyes National Seashore (PRNS), Marin County, aca and R. calendula captured per site. In California (38°02′N, 122°48′W). The 30,364- addition to being common winter residents at ha PRNS is dominated by a Mediterranean both study sites, these 3 species occupy fairly climate with wet winters and moderate, dry distinct foraging guilds. Regulus calendula summers that receive moisture in the form of represents a shrub tree–foraging guild with a coastal fog. Vegetation types occurring in PRNS winter diet primarily of spiders, insects, and include mixed evergreen forest, coastal scrub, small amounts of vegetative matter (Ingold grassland, and riparian and wetland habitats and Wallace 1994). Populations of R. calendula (Shuford and Timossi 1989). On 3 October 1995 within the study area are female-biased an unintentional human-ignited fire started at (Humple et al. 2001). Catharus guttatus is a the seashore and ultimately burned 5263 ha terrestrial or bush-gleaning omnivore and thus (17.3%) of the park over a period of 3 days represents a ground tree–foraging guild (Jones (PRNS 1997). This fire swept through the en- and Donovan 1996). And Passerella iliaca occu- tire MH study area, including the riparian pies a ground-foraging guild where it locates habitat present at this site. The rapid passing food by scraping dirt and leaf litter (Rogers of the fire through MH resulted in a relatively 1987). Fire may have a differential effect on undamaged riparian canopy, while the under- resource abundance and distribution for dif- story was completely burned. At least 5 km of ferent guilds (Raphael et al. 1987). The 3 spe- habitat was burned in all directions surround- cies chosen for these analyses should therefore ing MH. Thus, there was little chance that provide a stronger overall measure of songbird wintering birds chosen for these analyses for- response to fire at this site. 2005] WINTER SONGBIRD ECOLOGY FOLLOWING FIRE 47

Data Collection for P. iliaca and C. guttatus. Since R. calendula is the only one of the 3 species reliably sexed On 21 October 1995 (18 days after the start during winter (Pyle 1997), 2-way ANOVA was of the fire), the Point Reyes Bird Observatory used with sex added as a main effect. Compar- (PRBO) began a constant effort of mist-net isons of mass across seasons or between sites capturing birds within riparian vegetation at can be confounded with diurnal variation in MH. Fieldwork at PI had begun several months mass, with a gradual gain in mass over the earlier, but only captures within the same date course of a day (Graedel and Loveland 1995). periods as MH were used in these analyses. To verify whether this variation biased our Ten mist-nets at each site (12 m, 30-mm mesh) analyses of mass, we used t tests to compare were operated once every 10 days for 6 hours, differences in mean capture times for a ran- starting 15 minutes after sunrise. We confined domly selected group of birds from each site our analyses to the winter season (November– during early and late winter. February) for winters 1995/96 through 2001/02, Birds that were banded and subsequently except 1997/98. The total period of time nets recaptured ≥ 2 weeks later were assumed to were open (expressed as net-hours [nh]) was have spent that winter season within the vicin- similar between sites (PI = 4032 nh, MH = ity of the study site and thus demonstrated 3780 nh), thus reducing bias in recapture prob- within-year site fidelity. The winter season as ability due to variation in effort. Nets were defined above was extended by 2 weeks on operated at the same locations and in the same both ends (15 October–15 March) to include orientations throughout the study. Nets were birds that arrived during migratory periods closed early during periods of inclement but remained within the study areas over the weather. All birds were banded with a USFWS winter as indicated by recapture dates. This aluminum band before measurement and re- improved our sample size for this analysis but lease. We recorded mass with an electronic was not appropriate for the earlier analyses of balance to 0.1 g. body mass because body condition may reflect Statistical Analysis migratory behavior and resource availability at other sites. Also, recapture rates may be com- Individuals initially banded during a winter plicated by birds that are faithful to the same season and recaptured during any subsequent migratory paths (Cantos and Tellería 1994, winter season demonstrated between-year site Merom et al. 2000). To compare within-year fidelity. For each species, chi-square tests were recapture rates, we used binary logistic regres- used to compare the frequency of banded birds sion (PROC GENMOD; SAS Institute, Inc. recaptured versus not recaptured between sites. 1999), which evaluated the impact of year and To assess whether burned and unburned sites site on the probability of recapture within a differed in quality for birds, we first used winter season. The model included site as nom- t tests (SPSS, Inc. 2001) to compare changes inal data, year (6 winter seasons) as continuous in body mass of birds that were captured dur- data, and the interaction term. Because the ing early winter (November–December) and number of recaptures satisfying the above cri- recaptured during late winter (January–Feb- teria was low for some species in some seasons, ≥ ruary) with 1 month between captures. We we pooled data for all bird species. This runs also used a body mass index for a larger sam- the risk of obscuring species-specific responses ple of new captures between November and to predictor variables but provides a more February to compare body condition between robust measure of overwinter site persistence sites. Because body mass of different individu- at each site. We verified that data met para- als is confounded with body size, mass was metric assumptions of normality and homo- expressed relative to the structural size of an ± geneity of variance. Means are reported sx–, individual (Piersma and Davidson 1991, Brown with α = 0.05. 1996). Unflattened wing chord was regressed against body mass, and the residual of this RESULTS analysis (hereafter referred to as the body mass index) was used as the response variable A significantly higher frequency of banded in the following tests. We used t tests to com- P. iliaca was recaptured in subsequent winter pare mean body mass indices between sites seasons at PI (27 recaptured of 235 banded) 48 WESTERN NORTH AMERICAN NATURALIST [Volume 65

Fig. 1. Mean change in body mass from early (November–December) to late (January–February) winter for 3 species (A), and mean body mass index for new captures during winter (B) at post-fire (MH) and unburned (PI) riparian sites, Marin County, CA. Passerella iliaca = FOSP, Catharus guttatus = HETH, and Regulus calendula = RCKI. Error bars represent sx–, with sample sizes shown above/below bars.

χ2 than at MH (7 of 252; 1 = 14.21, P < 0.001). Differences in mean capture times (expressed The frequency of C. guttatus recaptured at PI as minutes after sunrise) were similar for the (15 of 171) was not significantly different from burned (184 ± 14) and unburned (185 ± 13) sites the proportion recaptured at MH (7 of 110; (t120 = –0.05, P = 0.96). Furthermore, differ- χ2 1 = 0.54, P = 0.46). Similarly, the between- ences in mean capture times across the winter year recapture rate for R. calendula did not season were similar for early (189 ± 12) and ± differ significantly between PI (30 of 377) and late (176 16) winter captures (t95 = 0.69, P χ2 MH (17 of 253; 1 = 0.34, P = 0.56). = 0.49). Thus, no attempt was made to correct 2005] WINTER SONGBIRD ECOLOGY FOLLOWING FIRE 49

TABLE 1. Within-year recapture rates for Passerella iliaca, Catharus guttatus, and Regulus calendula at post-fire (MH) and unburned (PI) riparian sites in Marin County, CA, for winter seasons 1995/96 through 2001/02. N is the number of new captures per 100 net-hours between 15 October and 15 March; R is the number of birds recaptured per 100 net- hours, and %R is the percent of N recaptured.

______Muddy Hollow (MH) ______Pine Gulch (PI) Species Yeara NRb %R N Rb %R Passerella iliaca 95/96 1.59 0.23 14.5 1.31 0.28 21.4 96/97 5.93 1.69 28.5 3.74 0.77 20.6 98/99 3.04 0.23 7.6 6.60 1.97 29.8 99/00 0.77 0.22 28.6 0.64 0.21 32.8 00/01 2.13 0.13 6.1 3.39 0.58 17.1 01/02 2.20 0.14 6.4 1.25 0.25 20.0 Mean (sx–) 95–02 2.61 (0.73) 0.44 (0.25) 15.3 (4.4) 2.82 (0.91) 0.68 (0.27) 23.6 (2.5) Catharus guttatus 95/96 1.47 0.23 15.6 3.28 1.22 37.2 96/97 2.18 0.12 5.5 4.40 0.99 22.5 98/99 1.29 0.47 36.4 2.55 0.69 27.1 99/00 0.55 0.11 20.0 1.17 0.32 27.4 00/01 1.59 0.53 33.3 1.40 0.47 33.6 01/02 2.20 0.82 37.3 1.62 0.50 30.9 Mean (sx–) 95–02 1.55 (0.25) 0.38 (0.11) 24.7 (5.3) 2.40 (0.51) 0.70 (0.14) 29.8 (2.1) Regulus calendula 95/96 2.72 0.91 33.5 3.57 0.75 21.0 96/97 9.43 2.66 28.2 6.71 1.65 24.6 98/99 5.49 1.29 23.5 10.54 2.78 26.4 99/00 1.54 0.33 21.4 3.30 0.85 25.8 00/01 2.92 0.13 4.5 4.20 1.40 33.3 01/02 4.26 0.55 12.9 10.61 3.87 36.5 Mean (sx–) 95–02 4.39 (1.15) 0.98 (0.38) 20.7 (4.3) 6.49 (1.38) 1.88 (0.50) 27.9 (2.4) aNo capture data are available for winter 1997/98. bRecaptured no less than 2 weeks after being banded, and within the date period 15 October–15 March. for diurnal changes in body mass. Mean species combined declined at MH, while at PI change in body mass from early to late winter the rate remained relatively constant, increasing was negative for all 3 species at MH and slightly slightly by the end of the study (Fig. 2). This positive for P. iliaca and C. guttatus at PI (Fig. interaction between site and year was signifi- 1A). However, differences between sites were cant (Table 2). not significant for any species (P. iliaca: t26 = DISCUSSION –1.26, P = 0.22; C. guttatus: t20 = –1.84, P = 0.08; R. calendula: t64 = 0.30, P = 0.76). Mean The significantly smaller proportion of P. body mass indices were slightly negative for iliaca recaptured in subsequent winter sea- all 3 species at MH and slightly positive for all sons at MH may reflect the effect of fire on 3 species at PI (Fig. 1B). However, these dif- guilds of birds that forage exclusively on open ferences between sites were also not signifi- ground. Catharus guttatus also utilizes terres- cant (P. iliaca: t136 = –0.19, P = 0.85; C. gut- trial food sources but readily occupies dense tatus: t83 = –0.63, P = 0.53; R. calendula: understory regrowth. As the understory began F1,321 = 0.55, P = 0.46). For R. calendula, to recover after the fire, there was a rapid loss there was a significant effect of sex (F1,321 = of open ground where P. iliaca prefers to feed. 21.95, P < 0.001), with males having a larger It is likely this species lost suitable foraging mean body mass index than females. This dif- habitat as dense vegetation began to grow over ference between sexes was consistent between riparian areas that had burned. Since the win- sites. ter diet of P. iliaca includes many kinds of seeds The mean percent of birds recaptured ≥2 (Rising 1996), a loss of preferred seed type weeks after initial capture but within the same might also result in a smaller number of birds winter season was lower at MH for all 3 species returning to this site in subsequent winters. (Table 1). During the course of the study, the Although we found little evidence the fire rate of recapture within a winter season for all adversely impacted body condition of birds 50 WESTERN NORTH AMERICAN NATURALIST [Volume 65

TABLE 2. Logistic regression model estimating the impacts of site and year on within-year recapture rate for 3 species during 6 winter seasons at Muddy Hollow (MH) and Pine Gulch (PI), Marin County, CA, 1995–2002. χ2 Variables df sx– P Intercept 1 –1.15 0.16 50.40 <0.0001 Site 1 0.12 0.25 0.24 0.62 Year 1 0.06 0.04 2.09 0.15 Site × Year 1 –0.17 0.07 5.43 0.02

(Gardali et al. 2003). Passerella iliaca contributed to this decline, however, further suggesting a decrease in appropriate foraging habitat for this species as the understory became dense in the years after the fire. In contrast, the canopy tree Alnus rubra initially survived the fire but then began to die after 3–4 years. Widespread mortality of this dominant tree species might have reduced the number of birds that initially fixed on this site after fall arrival and remained there through a winter season. Site fixation by sparrows was found to occur primarily in late winter by Ralph and Fig. 2. Proportion of birds recaptured during winter for Mewaldt (1975), although evidence is lacking 3 species combined (Passerella iliaca, Catharus guttatus, for a distinct sensitive period when site fixa- and Regulus calendula) as predicted from the logistic tion occurs (Ketterson and Nolan 1990). Dur- model (logit[π(x)] = –1.0305 – 0.1123x) for burned (MH = solid line) and (logit[π (x)] = –1.1526 + 0.060x) for ing later seasons of the study, birds initially unburned (PI = dashed line) riparian sites, Marin County, caught in early winter may have moved out of CA, 1995/96 through 2001/02. The actual proportions of the MH study area by late winter. Alterna- birds recaptured ≥ 2 weeks after initial capture during a tively, birds may have started foraging over a winter season are shown as solid circles for MH and open wider area within the burned zone. circles for PI. No capture data are available for 1997/98. The fire burned fast and hot, scorching the riparian understory but leaving the canopy intact. This may explain why foliage-gleaning wintering at MH, there was a significant de- species such as R. calendula were present crease over time in within-year recapture rates almost immediately after the fire here, while at MH. During the 1st winter after the fire, Blake (1982) observed this species in conifer- very little vegetation had grown back within ous woodland during winter only in unburned the burned area of the park, even within the sites following fire. The higher abundance of riparian habitat at MH. Since all 3 bird species foliage-gleaning species in unburned conifer- breed elsewhere, site-faithful individuals return- ous forests has also been reported during the ing to nonbreeding areas as well as immature breeding season (Bock and Lynch 1970, birds arriving for the first time in fall would Apfelbaum and Haney 1981, Raphael et al. have found a landscape dramatically altered 1987). As Alnus rubra gradually began to fall and structurally much simpler than compara- down at MH, suitability of this site for foliage- ble unburned habitat. Yet the small difference gleaning insectivores likely declined, and R. between sites in within-year recapture rate calendula may have switched to foraging in this 1st winter suggests that, contrary to expec- nonriparian habitats. tations, sufficient resources were available for Our study was limited in several important birds to remain within the study area through ways that deserve further discussion. First, we the winter. The decline over time in within-year were unable to establish long-term field sites recapture rates at MH was somewhat surpris- in a well-replicated study design, partially being, since this site witnessed a net increase in cause of the limited number of riparian sys- understory vegetation over the study period tems that burned in the fire. Our conclusions 2005] WINTER SONGBIRD ECOLOGY FOLLOWING FIRE 51 are therefore based on single burned and un- (Gardali et al. 2003), suggest that the effects of burned sites, which makes it difficult to attribute fire on songbirds are complex and will differ the changes we observed to the fire rather than among species. In general, however, songbirds to other site-specific differences. Attempts to at this riparian site appear to be resilient to generalize these results to other riparian sys- fire during the breeding season (Gardali et al. tems should be made with caution and should 2003) and the nonbreeding season. address the potentially different effects of natural and human-ignited fires. Second, it is ACKNOWLEDGMENTS important to consider the habitat that borders riparian sites when examining the effects of This work would not have been possible fire. Riparian surrounded by coniferous wood- without the assistance of over 40 interns that land, for example, might contain a heavier fuel helped catch and band birds at both study sites. load in the form of woody debris, and the In particular, we thank Benjamin Michael Parmeter for volunteering on this project and accumulation of leaf litter and woody debris for assisting with plant identification. Both may affect fire severity and the fate of riparian Nathaniel Seavy and Li Zhang helped with trees (Ellis 2001). The riparian at MH is sur- statistical analyses. We thank Grant Ballard, rounded by a mosaic of habitats classified as Richard Hutto, Steve Latta, Nathaniel Seavy, coastal scrub and mixed evergreen forest and an anonymous referee for helpful com- (Shuford and Timossi 1989). This habitat also ments on previous drafts of the manuscript. burned, and the intact riparian canopy at MH We are very grateful to Sarah Allen of the Point was the only green vegetation present imme- Reyes National Seashore and to the Marin diately afterward. This may have contributed County Open Space District for their support to the abundance of all 3 bird species at the of our songbird monitoring efforts. We also MH site soon after the fire. Passage of a fire thank the members and board of directors of through the riparian at PI would be different, PRBO for supporting all components of the since this site is bordered by a lagoon on one work. Additional funding was provided by the side. Thus, the effects of fire are confounded Bernard Osher Foundation, David and Lucille by landscape structure. Finally, the uninten- Packard Foundation, Marin Community Foun- tional nature of this fire prevented us from ob- dation, San Francisco Foundation, Ms. Doris taining valuable pre-fire data. Researchers inter- Leonard Conservation Associates, and Mrs. ested in the effects of fire on avian ecology Dorothy Hunt. This is PRBO contribution 1177. should consider prescribed burns as experimental treatments, backed by at least 5 years LITERATURE CITED of pre-fire data to help interpret population fluctuations. Even without pre-fire data, how- APFELBAUM, S., AND A. HANEY. 1981. Bird populations before and after wildfire in a Great Lakes pine forest. ever, small-scale studies such as this can be Condor 83:347–354. used in meta-analyses that combine results of BLAKE, J.G. 1982. Influence of fire and logging on non- many individual studies. breeding bird communities of ponderosa pine forests. The PRNS has recently conducted research Journal of Wildlife Management 46:404–415. BOCK, C.E., AND J.F. LYNCH. 1970. Breeding bird popula- to determine how prescribed burns will play a tions of burned and unburned conifer forest in the role in future ecosystem management within Sierra Nevada. Condor 72:182–189. the park. While human-ignited fires histori- BRAWN, J.D., S.K. ROBINSON, AND F.R. THOMPSON III. 2001. cally occurred at mean intervals of 7.7 to 8.5 The role of disturbance in the ecology and conservation of birds. Annual Review of Ecology and System- years in the Douglas-fir (Pseudotsuga menziesii) atics 32:251–276. forests of PRNS during the 18th and 19th cen- BROWN, M.E. 1996. Assessing body condition in birds. turies, there is little evidence of fire in recent Pages 67–121 in V. Nolan, Jr., and E.D. Ketterson, edi- decades (Brown et al. 1999). The historical tors, Current ornithology. Volume 13. Plenum Press, New York. pattern of fire within riparian communities at BROWN, P.M., M.W. KAYE, AND D. BUCKLEY. 1999. Fire his- this site is unknown, but recovery of vegeta- tory in Douglas-fir and coast redwood forests at Point tion at MH has been rapid; young Alnus rubra Reyes National Seashore, California. Northwest Sci- are now abundant and a thick understory of ence 73:205–216. CANTOS, F.J., AND J.L. TELLERÍA. 1994. Stopover site fidelity shrubs and herbs is present. Our data, together of four migrant warblers in the Iberian Peninsula. with breeding season data from these sites Journal of Avian Biology 25:131–134. 52 WESTERN NORTH AMERICAN NATURALIST [Volume 65

DEBANO, L.F., AND D.G. NEARY. 1996. Effects of fire on MARRA, P.P., K.A. HOBSON, AND R.T. HOLMES. 1998. Link- riparian systems. Pages 69–76 in P. F. Ffolliott, L.F. ing winter and summer events in a migratory bird by DeBano, M.B. Baker et al., technical coordinators, using stable-carbon isotopes. Science 282:1884–1886. Effects of fire on Madrean Province ecosystems— MEROM, K., Y. YOM-TOV, AND R. MCCLERY. 2000. Philopatry symposium proceedings. General Technical Report to stopover site and body condition of transient Reed RMGTR-289, U.S. Department of Agriculture, Forest Warblers during autumn migration through Israel. Service, Rocky Mountain Forest and Range Experi- Condor 102:441–444. ment Station, Fort Collins, CO. NORRIS, D.R., P.P. MARRA, T.K. KYSER, T.W. SHERRY, AND ELLIS, L.M. 2001. Short-term response of woody plants to L.M. RATCLIFFE. 2004. Tropical winter habitat limits fire in a Rio Grande riparian forest, central New Mex- reproductive success on the temperate breeding ico, USA. Biological Conservation 97:159–170. grounds in a migratory bird. Proceedings of the Royal GARDALI, T., G.R. GEUPEL, A.L. HOLMES, AND D.L. HUMPLE. Society of London, Series B, Biological Sciences 2003. Songbird response to wildfire in a coastal ripar- 271:59–64. ian system. Pages 57–64 in P. M. Faber, editor, Cali- PIERSMA, T., AND N.C. DAVIDSON. 1991. Confusions of mass fornia riparian systems: processes and floodplain man- and size. Auk 108:441–444. agement, ecology, and restoration. Proceedings of POINT REYES NATIONAL SEASHORE. 1997. Fire: burning the 2001 riparian habitat and floodplains conference. issues in fire management. Point Reyes National Sea- Riparian Habitat Joint Venture, Sacramento, CA. shore, CA. GRAEDEL, S.K., AND R.E. LOVELAND. 1995. Seasonal and PYLE, P. 1997. Identification guide to North American diurnal mass variation in Black-capped Chickadees birds. Part I (Columbidae to Ploceidae). Slate Creek and White-throated Sparrows. Wilson Bulletin 107: Press, Bolinas, CA. 723–727. RALPH, C.J., AND L.R. MEWALDT. 1975. Timing of site fixa- HOLMES, R.T., AND T.W. S HERRY. 1992. Site fidelity of migration upon the wintering grounds in sparrows. Auk tory warblers in temperate breeding and Neotrop- 92:698–705. ical wintering areas: implications for population RAPHAEL, M.G., M.L. MORRISON, AND M.P. YODER- dynamics, habitat selection, and conservation. Pages WILLIAMS. 1987. Breeding bird populations during 563–578 in J.M. Hagan III and D.W. Johnson, editors, twenty-five years of postfire succession in the Sierra Ecology and conservation of Neotropical migrant Nevada. Condor 89:614–626. landbirds. Smithsonian Institution Press, Washing- RISING, J.D. 1996. A guide to the identification and nat- ton, DC. ural history of the sparrows of the United States and HUMPLE, D.L., N. NUR, G.R. GEUPEL, AND M.P. LYNES. Canada. Academic Press, San Diego, CA. 2001. Female-biased sex ratio in a wintering popula- ROGERS, C.M. 1987. Predation risk and fasting capacity: tion of Ruby-crowned Kinglets. Wilson Bulletin 113: do wintering birds maintain optimal body mass? 419–424. Ecology 68:1051–1061. HUTTO, R.L. 1995. Composition of bird communities fol- ROTENBERRY, J.T., R.J. COOPER, J.M. WUNDERLE, AND K.S. lowing stand-replacement fires in northern Rocky SMITH. 1995. When and how are populations lim- Mountain (U.S.A.) conifer forests. Conservation Biol- ited: the roles of insect outbreaks, fires, and other ogy 9:1041–1058. natural disturbances. Pages 55–84 in T.E. Martin and INGOLD, J.L., AND G.E. WALLACE. 1994. Ruby-crowned D.M. Finch, editors, Ecology and management of Kinglet (Regulus calendula). In: A. Poole and F. Gill, Neotropical migratory birds. Oxford University Press, editors, The birds of North America, No. 119. The New York. Academy of Natural Sciences, Philadelphia, PA; and SANDERCOCK, B.K., AND A. JARAMILLO. 2002. Annual sur- The American Ornithologists’ Union, Washington, vival rates of wintering sparrows: assessing demo- DC. graphic consequences of migration. Auk 119:149–165. JONES, P.W., AND T.M. D ONOVAN. 1996. Hermit Thrush SAS INSTITUTE, INC. 1999. The SAS system for windows, (Catharus guttatus). In: A. Poole and F. Gill, editors, release 8.2. SAS Institute, Inc., Cary, NC. The birds of North America, No. 261. The Academy SHUFORD, W.D., AND I.C. TIMOSSI. 1989. Plant communi- of Natural Sciences, Philadelphia, PA; and The ties of Marin County. California Native Plant Soci- American Ornithologists’ Union, Washington, DC. ety, Special Publication 10. KETTERSON, E.D., AND V. N OLAN, JR. 1990. Site attachment SPSS, INC. 2001. Version 11.0. SPSS, Inc., Chicago, IL. and site fidelity in migratory birds: experimental evi- WUNDERLE, J.M., JR., AND S.C. LATTA. 2000. Winter site dence from the field and analogies from neurobiology. fidelity of Nearctic migrants in shade coffee planta- Pages 117–129 in E. Gwinner, editor, Bird migration: tions of different sizes in the Dominican Republic. physiology and ecophysiology. Springer-Verlag, Berlin. Auk 117:596–614. KREISEL, K.J., AND S.J. STEIN. 1999. Bird use of burned and unburned coniferous forests during winter. Wil- Received 5 August 2003 son Bulletin 111:243–250. Accepted 14 April 2004 Western North American Naturalist 65(1), © 2005, pp. 53–63

FISH ASSEMBLAGE STRUCTURE FOLLOWING IMPOUNDMENT OF A GREAT PLAINS RIVER

Michael C. Quist1,3, Wayne A. Hubert1, and Frank J. Rahel2

ABSTRACT.—Understanding the upstream and downstream effect of impoundments on stream fish assemblages is important in managing fish populations and predicting the effects of future human activities on stream ecosystems. We used information collected over a 41-year period (1960–2001) to assess changes in fish assemblage structure resulting from impoundment of the Laramie River by Grayrocks Reservoir. Prior to impoundment (i.e., 1960–1979), fish assemblages were dominated by native catostomids and cyprinids. After impoundment several exotic species (e.g., smallmouth bass [Micropterus dolomieu], walleye [Sander vitreus; formerly Stizostedion vitreum], yellow perch [Perca flavescens], brown trout [Salmo trutta]) were sampled from reaches upstream and downstream of the reservoir. Suckermouth minnows (Phenacobius mirabilis) were apparently extirpated, and hornyhead chubs (Nocomis biguttatus) and common shiners (Luxilus cornutus) became rare upstream of Grayrocks Reservoir. The lower Laramie River downstream from Gray- rocks Reservoir near its mouth retains habitat characteristics similar to those prior to impoundment (e.g., shallow, braided channel morphology) and is the only downstream area where several sensitive species persist, including suckermouth minnows, hornyhead chubs, and bigmouth shiners (Notropis dorsalis). Grayrocks Reservoir serves as a source of exotic piscivores to both upstream and downstream reaches and has altered downstream habitat characteristics. These impacts have had a substantial influence on native fish assemblages. Our results suggest that upstream and downstream effects of impoundment on fish assemblage structure are similar and that downstream reaches which retain habitat characteristics similar to pre-impoundment conditions may serve as areas of refuge for native species.

Key words: impoundment, conservation, exotic species, Great Plains, Wyoming.

Reservoirs are regarded as one of the most peak flows during spring and augmented flows significant threats to aquatic biodiversity at during summer and winter (Ward and Stanford global and regional scales (Richter et al. 1997, 1979, Pringle et al. 2000). Discharge from reser- Rosenberg et al. 1997). Over 39,000 large dams voirs can alter thermal regimes, with colder (≥15 m in height) throughout the world (Dyne- summer temperatures and warmer winter sius and Nilsson 1994) and approximately 5500 temperatures than occurred prior to impound- large dams and 75,000 smaller dams (<15 m ment (Rosenberg et al. 1997, Poff and Hart in height) in the United States have been con- 2002). Water velocity is reduced upon enter- structed to provide agricultural, hydropower, ing a reservoir, and this allows sediment to flood control, and recreational benefits (Rosen- settle from the water column. Thus, water dis- berg et al. 2000). Although the rate of reser- charged from reservoirs is relatively free of voir construction has declined during the last sediment. Reduced sediment transport, coupled 20 years (Postel et al. 1996, Rosenberg et al. with regulated flow regimes, can alter channel 2000), projects such as the Three Gorges de- morphology and substrate characteristics. For velopment in China and various small projects example, reduced peak flows and continual in northern Canada indicate continued inter- discharge of sediment-free water commonly est in reservoir construction (Rosenberg et al. result in reduced width and increased depth 1997). (i.e., channel incision), and selective transport Extensive development of water resources of smaller sediments causes armoring of the has instigated a variety of studies on the effects substrate (Ward and Stanford 1979, Poff and of impoundments to lotic ecosystems. Regu- Hart 2002). All of these factors, in isolation or lated discharge from impoundments results in in concert, influence the composition of native an altered hydrograph, often with reduced fish assemblages, especially with regard to

1U.S. Geological Survey, Wyoming Cooperative Fish and Wildlife Research Unit, University of Wyoming, Laramie, WY 82071-3166. 2Department of Zoology and Physiology, University of Wyoming, Laramie, WY 82071-3166. 3Corresponding author.

53 54 WESTERN NORTH AMERICAN NATURALIST [Volume 64 species adapted to dynamic river systems char- Reservoir, located 40 km east of Wheatland, acteristic of the western Great Plains (Cross Wyoming, was constructed in 1980 by the and Moss 1987, Fausch and Bestgen 1997). Missouri River Basin Power Company. The Most studies that have investigated the reservoir was built to provide cooling water influence of reservoirs on biotic communities for a coal-fired power-generation facility. No have focused on anadromous species. Reser- heated effluent is returned to Grayrocks Reser- voirs may act as barriers to movements and voir or the Laramie River. At capacity the contribute to the decline of species dependent reservoir has a surface area of 1435 ha and a on extensive spawning migrations, such as maximum depth of approximately 23 m. The Pacific salmonids (Li et al. 1987, Gregory et al. fish community in the reservoir includes native 2002). Like anadromous species, many resi- species (e.g., channel catfish [Ictalurus puncta- dent fishes are dependent on upstream or tus], river carpsucker [Carpiodes carpio]) and a downstream movements for spawning or for variety of exotic species introduced to provide an escape from harsh environmental condi- recreational angling opportunities (e.g., small- tions (Geist et al. 1996, Matthews 1998). In mouth bass [Micropterus dolomieu], walleye addition to blocking movements, altered habi- [Sander vitreus; formerly Stizostedion vitreum]) tat conditions, loss of channel-floodplain con- and prey for sport fishes (e.g., gizzard shad nectivity, and introduction of exotic species may [Dorosoma cepedianum]; Hubert and O’Shea influence fish assemblage structure. Under- 1991). standing how impoundments influence fish Prior to settlement by Europeans, the assemblages is important for both managing Laramie River was characteristic of other Great stream fish populations and predicting the Plains rivers having well-developed pool and effects of future human activities (e.g., dam riffle habitats in middle reaches (e.g., near construction or removal; Hart et al. 2002). Chugwater Creek, North Laramie River) and Although the effects of impoundments on fish a shallow, braided channel with dynamic sub- assemblages have received detailed investiga- strate in lower reaches (Patton and Hubert tion in the northwestern (Gregory et al. 2002) 1993, Baxter and Stone 1995). The Laramie and southwestern (Carlson and Muth 1989) River from Grayrocks Reservoir to its conflu- United States, the effects of reservoirs on ence with the North Platte River (approxi- stream systems in the Great Plains are not mately 20 km) has changed from a braided well documented. Therefore, the purpose of channel to a single, incised channel with few this study was to assess the upstream and remaining side-channel and backwater habi- downstream effects of an impoundment on the tats (Patton and Hubert 1993). fish assemblage of a Great Plains river. Specif- ically, we examined changes in fish assemblage We used historic data collected during 1960– structure upstream and downstream of Gray- 1979 by University of Wyoming (UW) and rocks Reservoir on the Laramie River, Wyo- Wyoming Game and Fish Department (WGFD) ming, using information collected before and personnel to characterize the fish assemblage after reservoir construction. prior to construction of Grayrocks Reservoir (Fig. 1). Fish assemblage structure after reservoir METHODS construction was determined using information collected by UW, WGFD, and Montana The Laramie River, one of the largest tribu- State University personnel during 1980–2001. taries to the North Platte River, originates in All fish were sampled using either electrofish- the mountains of northern Colorado and flows ing or seining. The reaches sampled during northeasterly to its confluence with the North 1960–1979 were the same as those we sampled Platte River at Fort Laramie, Wyoming (Fig. 1). during 1980–2001, allowing us to compare The Laramie River has experienced extensive changes in fish assemblages before and after water development since the early 1900s, construction of Grayrocks Reservoir. Twelve including diversion of water for agricultural additional reaches (4 upstream and 8 down- use and construction of large storage reservoirs. stream of Grayrocks Reservoir) were also sam- Two large reservoirs have been constructed on pled during 1980–2001 to provide further the Laramie River: Wheatland Reservoir Num- evidence of changes to the fish assemblage fol- ber 2 and Grayrocks Reservoir. Grayrocks lowing impoundment. 2004] EFFECTS OF IMPOUNDMENT ON NATIVE FISHES 55

Fig. 1. Map of reaches sampled in the Laramie River drainage near Grayrocks Reservoir, Wyoming. Solid symbols represent reaches sampled both before (1960–1979) and after (1980–2001) impoundment, and open symbols represent additional reaches sampled after impoundment.

Similarity in fish assemblage structure among then we used the most recent year of data in reaches was evaluated using Jaccard’s index of the analysis. Differences in the number of assemblage similarity (Jongman et al. 1995). species by family were examined using a Jaccard’s index values were calculated for all paired t test (Ott 1993). Because the effects of possible pairs of upstream and downstream impoundment may differ depending on the reaches sampled before and after impoundment. location of the reach (i.e., upstream versus The resulting matrix of similarity indices was downstream), tests were conducted on reaches clustered using the unweighted pair-group upstream and downstream of Grayrocks Reser- method (UPGMA; Jongman et al. 1995, voir. Furthermore, a Bonferroni adjustment was Matthews 1998) to produce a dendrogram used to avoid type-I errors associated with depicting clusters of reaches with similar fish multiple statistical tests (Ott 1993). Paired t assemblage structure. The similarity analysis tests were performed using SAS (SAS Insti- was conducted using NTSYS (Rohlf 1990). tute, Inc. 1996). Using the number of species by family (i.e., In addition to examining changes in assem- clupeids, salmonids, native cyprinids, catosto- blage structure across reaches, we obtained a mids, ictalurids, fundulids, centrarchids, and chronology of changes in the fish assemblage percids), we assessed changes in fish assem- from a reach downstream from Grayrocks blage structure sampled from reaches before Reservoir. The sampling reach, 8 km down- and after impoundment. If a reach was sam- stream from Grayrocks Reservoir, was sampled more than once (i.e., post-impoundment), pled repeatedly from 1979 to 1991. Similar 56 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Fig. 2. Dendrogram depicting fish assemblage similarity among reaches sampled in the Laramie River, Wyoming, before and after construction of Grayrocks Reservoir. Reach labels that begin with the same number indicate that the same reach was sampled before (bold text) and after (normal text) impoundment. Additional reaches that were sampled only after impoundment are labeled with an “A”. Reach labels also designate whether the reach was located upstream (U) or downstream (D) of Grayrocks Reservoir. data were unavailable for reaches upstream of been recently collected. Within the cluster of the reservoir. reaches sampled prior to impoundment, some clustering of the downstream reaches was RESULTS observed (i.e., reaches 8D, 9D, and 10D in bold; Fig. 2). Conversely, upstream and downstream Cluster analysis of similarity index values reaches within the cluster of post-impound- illustrated changes in fish assemblage struc- ment samples did not exhibit any evident pat- ture following construction of Grayrocks Reser- terns or consistent clustering. This suggests voir (Fig. 2). For example, 2 broad clusters were identified (separating at a Jaccard’s index that fish assemblages were different before and value of approximately 0.28; Fig. 2), with after impoundment and that after impoundment, reaches sampled prior to impoundment in 1 fish assemblages were similar in upstream and cluster and reaches sampled after impound- downstream reaches. ment in the other. Two reaches (i.e., reaches Prior to construction of Grayrocks Reservoir, A5D and A6D in Fig. 2) sampled after im- fish assemblages were dominated by native poundment clustered with those sampled before cyprinids and catostomids (Table 1). Native impoundment. Both reaches were located cyprinids were sampled from all reaches, and near the confluence with the North Platte catostomids were found at 80% of upstream and River and had assemblages similar to those 67% of downstream reaches. However, after found prior to construction of the reservoir. In completion of Grayrocks Reservoir, salmonids, addition, these 2 reaches were the only down- centrarchids, catostomids, and percids became stream ones where sensitive species (e.g., more frequent both upstream and downstream suckermouth minnow, hornyhead chub) have of the reservoir. The number of species in 2004] EFFECTS OF IMPOUNDMENT ON NATIVE FISHES 57

Fig. 3. Change in the number of species sampled by taxa (CLUP = clupeidae, SALM = salmonidae, NCYP = native cyprinidae, CATO = catostomidae, ICTA = ictaluridae, FUND = fundulidae, CENT = centrarchidae, PERC = perci- dae, NATIVE = all native species, and EXPISC = exotic piscivorous species) from reaches following construction of Grayrocks Reservoir, Wyoming. Positive values represent the addition of species to a taxonomic group, while negative values represent the loss of species. An asterisk represents a significant change in the mean number of species (P < 0.001; paired t test with a Bonferroni adjustment). nearly all families increased following impound- after impoundment. Several species were in- ment except for native cyprinids where 3 troduced to the system following impoundment fewer species were generally collected follow- including gizzard shad, channel catfish, rain- ing impoundment (Fig. 3). Most changes in bow trout, smallmouth bass, and walleye. the number of species were similar between Temporal trends in the fish assemblage were reaches sampled upstream and downstream of observed for a reach downstream from Gray- the reservoir, but for some families (e.g., native rocks Reservoir (Fig. 4). Prior to construction cyprinids, ictalurids) downstream changes were of Grayrocks Reservoir in 1980, over 90% of greater (Fig. 3). Furthermore, significantly fewer the fish collected were native catostomids and native and more exotic species were sampled cyprinids. One year after the reservoir was following impoundment from both upstream completed, brown trout, rainbow trout, and and downstream reaches. yellow perch were sampled in the reach. Species-specific occurrences at reaches Walleyes were relatively abundant in the reach sampled before and after impoundment iden- 5 years after impoundment, and smallmouth tified further changes to fish assemblages. The bass were first sampled 11 years after impound- decline of native cyprinids upstream of Gray- ment. Catostomids and cyprinids were com- rocks Reservoir was associated with reduced mon in the reach after impoundment, but frequency of occurrence of bigmouth shiners, brassy minnows, common shiners, fathead min- creek chubs, common shiners, hornyhead nows, longnose dace, and longnose suckers chubs, and suckermouth minnows (Table 1). were not collected after 1981. Although the frequency of occurrence of several species declined in downstream reaches, DISCUSSION other cyprinids (e.g., creek chubs, central stonerollers) were more common following impound- The most notable change in the fish assem- ment. Catostomids were common in all reaches blage upstream of Grayrocks Reservoir was 58 WESTERN NORTH AMERICAN NATURALIST [Volume 64 = 8) N = 4) ( N ed before and after completion of = 3) ( N Same reaches reaches Additional = 7) ( N ( Upstream Downstream Upstream Downstream 1111 01 00 10 00 1242 0112 10 10 53 13 32 41 13 23 34 21 23 31 33 12 22 10 23 21 33 00 43 32 41 10 32 34 20 31 11 01 11 2210 11 33 22 00 02 44 12 00 23 05 02 48 11 02 22 13 10 1022 00 00 01 34 ______uxilus cornutus undulus sciadicus undulus zebrinus F Hybognathus hankinsoni Catostomus catostomus Ictalurus punctatus F Etheostoma exile Notropis dorsalis Campostoma anomalum Semotilus atromaculatus L Pimephales promelas Nocomis biguttatus Rhinichthys cataractae Cyprinella lutrensis Notropis stramineus Phenacobius mirabilis Carpiodes cyprinus Carpiodes carpio macrolepidotum Moxostoma Catostomus commersoni Noturus flavus Etheostoma nigrum 1. Number of reaches where each species was sampled upstream and downstream of Grayrocks Reservoir, Wyoming. Reaches were sampl Wyoming. 1. Number of reaches where each species was sampled upstream and downstream Grayrocks Reservoir, athead minnow OSTOMIDAE Plains topminnow Brassy minnow Longnose sucker Channel catfish Plains killifish Iowa darter Bigmouth shiner Central stoneroller Creek chub Common shiner F Hornyhead chub Longnose dace Red shiner Sand shiner Suckermouth minnow Quillback River carpsucker Shorthead redhorse White sucker Stonecat Johnny darter YPRINIDAE AT UNDULIDAE ERCIDAE ABLE CTALURIDAE P T C C I F Native species the reservoir in 1980. Common name Scientific name Before After Before After After After 2004] EFFECTS OF IMPOUNDMENT ON NATIVE FISHES 59 = 8) N = 4) ( N = 3) ( N Same reaches reaches Additional = 7) ( N ( Upstream Downstream Upstream Downstream 01 12 05 2400 03 03 23 00 05 02 01 01 0034 00 23 27 0200 02 01 12 01 ______epomis cyanellus erca flavescens Salmo trutta L Micropterus dolomieu Sander vitreus P Dorosoma cepedianum Cyprinus carpio Oncorhynchus mykiss 1. Continued. alleye ellow perch Brown trout Green sunfish Smallmouth bass W Y Gizzard shad Common carp Rainbow trout UPEIDAE ENTRARCHIDAE L YPRINIDAE ABLE ERCIDAE ALMONIDAE C T S C P C Non-piscivorous exotic species Non-piscivorous exotic Piscivorous exotic species Piscivorous exotic Common name Scientific name Before After Before After After After 60 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Fig. 4. Composition of species (BNT = brown trout, RBT = rainbow trout, BMN = brassy minnow, CKC = creek chub, CRP = common carp, CSH = common shiner, FHM = fathead minnow, LND = longnose dace, RDS = red shiner, LNS = longnose sucker, QBK = quillback, SRH = shorthead redhorse, WHS = white sucker, CCF = channel catfish, STC = stonecat, GSF = green sunfish, SMB = smallmouth bass, WAE = walleye, and YEP = yellow perch) sampled from a reach 8 km downstream of Grayrocks Reservoir on the Laramie River, Wyoming. Samples collected after 1980 represent post-impoundment fish assemblages. the reduced frequency of occurrence or loss of other fishes (Moyle 1986). Gido et al. (2002) native cyprinids and the occurrence of exotic suggested that Tuttle Creek Reservoir, an im- piscivores after impoundment. Many introduced poundment on the Big Blue River, Kansas, has species are thought to be highly successful in served as a source for the dispersal and subse- reservoir systems because of the relatively sta- quent establishment of introduced species to ble lentic habitat and limited competition with upstream reaches. Although Gido et al. (2002) 2004] EFFECTS OF IMPOUNDMENT ON NATIVE FISHES 61 found that only a single species, speckled and side-channel and backwater habitats were chub (Macrhybopsis aestivalis), had been abundant (Patton and Hubert 1993, Baxter extirpated upstream of Tuttle Creek Reservoir, and Stone 1995). Patton and Hubert (1993) other studies have reported substantial found that regulated, sediment-free flows from changes in upstream fish assemblages result- Grayrocks Reservoir have resulted in an in- ing from impoundment and introduction of cised channel (approximately 1.5 m), a domi- exotic species. Taylor et al. (2001) reported nance of run habitat, reduced availability of that the fish assemblage in an Illinois stream side-channel and backwater habitats, and re- upstream from a reservoir shifted from a duced fine substrate. Not only are the habitat cyprinid-dominated to a centrarchid-domi- changes (i.e., stable substrate and flows) con- nated assemblage following impoundment. ducive to the establishment of exotic species, Winston et al. (1991) reported similar results but the reservoir also provides a continual wherein the construction of a small impound- source of exotic species to reaches downstream ment on an Oklahoma stream resulted in the of the reservoir. The direct effect of exotic extirpation of native cyprinids from upstream species on native fish assemblages is unknown, reaches. The decline or extirpation of species but their role as predators, especially small- upstream from reservoirs could be due to a mouth bass and walleye, on fish assemblages variety of mechanisms, including the inability downstream of impoundments has been well of fish to move to downstream refuge areas documented throughout the western United during environmental stress, predation by intro- States (e.g., McMahon and Bennett 1996). In duced piscivores that move upstream from the addition to providing suitable habitat for exotic reservoir, predation on eggs and larvae that species, habitat changes associated with im- drift into the reservoir, or disruption of recolo- poundments may be beneficial for some native nization dynamics from downstream source species. For example, catostomids and central populations (Winston et al. 1991, Luttrell et al. stonerollers were more common in down- 1999). stream reaches following reservoir construc- We did not determine the specific mecha- tion. Because these species are most common nism for changes in fish assemblage structure in run habitats with large substrate (Aadland upstream of Grayrocks Reservoir, but the 1993, Pflieger 1997), their increased frequen- presence of exotic piscivores may explain some cies of occurrence are likely due to reduced of the observed trends. For example, several fine sediment transport and regulated flows species such as suckermouth minnows, horny- (i.e., larger substrate, increased run habitat). head chubs, and common shiners are known Despite changes in habitat for most of the to be sensitive to biotic interactions (Baxter Laramie River downstream of Grayrocks Reser- and Stone 1995, Pflieger 1997) and are now voir, the lower Laramie River near its conflu- absent or rare in upstream reaches. In addi- ence with the North Platte River maintains tion, invasive species in reaches upstream of some of its historical characteristics, having a Grayrocks Reservoir (i.e., green sunfish, yellow shallow, braided channel with sand and gravel perch, brown trout) have been shown to be substrate (Patton and Hubert 1993). Exotic important predators when introduced to new species are generally absent from this lower systems (Tabor and Wurtsbaugh 1991, Winston segment, and this is the only downstream area et al. 1991, Johnson and Hines 1999). For in- where suckermouth minnows, bigmouth shin- stance, Lohr and Fausch (1996) reported that ers, and hornyhead chubs have been collected introduced green sunfish have had a negative recently. Therefore, downstream reaches that influence on stream fish assemblages in the retain some historical habitat characteristics Great Plains region of Colorado. Green sun- may “reset” ecological conditions (Bain et al. fish were present in a few upstream reaches 1988, Stanford et al. 1996) and provide a refuge prior to construction of Grayrocks Reservoir, for native species sensitive to biotic interactions but were more frequent following completion with exotic species. of the reservoir. Prior to settlement by Europeans, rivers and The effect of Grayrocks Reservoir on down- streams in the Great Plains were not only fre- stream habitat has been substantial. Before quently intermittent, with extreme fluctuations settlement by Europeans, downstream reaches in flow, temperature, and dissolved oxygen, of the Laramie River were shallow and braided, but they also had a heterogeneous channel 62 WESTERN NORTH AMERICAN NATURALIST [Volume 64 morphology with dynamic substrate char- CROSS, G.B., AND R.E. MOSS. 1987. Historic changes in fish acteristics (Matthews 1988, Fausch and Bestgen communities and aquatic habitats in plains stream of Kansas. Pages 155–165 in W. J. Matthews and D.C. 1997). Because many species were unable to Heins, editors, Community and evolutionary ecology tolerate the harsh environmental conditions or of North American stream fishes. University of Okla- could not reproduce due to the shifting sub- homa Press, Norman. strate (e.g., nest-building centrarchids), fish DYNESIUS, M., AND C. NILSSON. 1994. Fragmentation and assemblages were depauperate and had little flow regulation of river systems in the northern third of the world. Science 266:753–762. or no history of co-occurrence with piscivores FAUSCH, K.D., AND K.R. BESTGEN. 1997. Ecology of fishes (Fausch and Bestgen 1997). Reservoirs have indigenous to the central and southwestern Great changed habitat characteristics in downstream Plains. Pages 131–166 in F.L. Knopf and F.B. Samson, reaches such that exotic species are able to editors, Ecology and conservation of Great Plains survive, and the reservoirs provide a continual vertebrates. Springer-Verlag, New York. GEIST, D.R., L.W. VAIL, AND D.J. EPSTEIN. 1996. Analysis source of exotic species to upstream and down- of potential impacts to resident fish from Columbia stream reaches, even if these species cannot River system operation alternatives. Environmental maintain naturalized populations. Management 20:275–288. Understanding the mechanism responsible GIDO, K.B., C.S. GUY, T.R. STRAKOSH, R.J. BERNOT, K.J. HASE, AND M.A. SHAW. 2002. Long-term changes in (i.e., physical habitat alterations or biotic inter- the fish assemblages of the Big Blue River basin 40 actions) for changes in native fish assemblages years after the construction of Tuttle Creek Reser- is difficult. Regardless, our results suggest that voir. Transactions of the Kansas Academy of Science fish assemblage structure has changed in the 105:193–208. Laramie River following construction of Gray- GREGORY, S., H. LI, AND J. LI. 2002. The conceptual basis for ecological responses to dam removal. BioScience rocks Reservoir. Compared to other reservoirs 52:713–723. in the Great Plains, Grayrocks Reservoir is HART, D.D., T.E. JOHNSON, K.L. BUSHAW-NEWTON, R.J. relatively new, but changes in the fish assem- HORWITZ, A.T. BEDNAREK, D.F. CHARLES, D.A. blage upstream and downstream of the reser- KREEGER, AND D.J. VELINSKY. 2002. Dam removal: voir are evident. The long-term effects on fish challenges and opportunities for ecological research and river restoration. BioScience 52:669–681. assemblage structure are unknown; thus, con- HUBERT, W.A., AND D.T. O’SHEA. 1991. Reproduction by tinued monitoring of this system is important fishes in a headwater stream flowing into Grayrocks to better understand the effects of stream Reservoir, Wyoming. Prairie Naturalist 23:61–68. impoundment in the western Great Plains and JOHNSON, J.E., AND R.T. HINES. 1999. Effect of suspended sediment on vulnerability of young razorback suck- to develop strategies for managing stream fish ers to predation. Transactions of the American Fish- assemblages. eries Society 128:648–655. JONGMAN, R.H.G., C.J.F. TER BRAAK, AND O.F.R. VAN TON- ACKNOWLEDGMENTS GEREN. 1995. Data analysis in community and landscape ecology. Cambridge University Press, New York. We thank Dirk Miller for assistance in ob- LI, H.W., C.B. SCHRECK, C.E. BOND, AND E. REXSTAD. 1987. Factors influencing changes in fish assem- taining fish assemblage information. Comments blages of Pacific Northwest streams. Pages 193–202 by M. Belk, D. Shiozawa, L. Thel, and R. in W. J. Matthews and D.C. Heins, editors, Community Williams greatly improved the quality of the and evolutionary ecology of North American stream manuscript. Funding was provided by the fishes. University of Oklahoma Press, Norman. LOHR, S.C., AND K.D. FAUSCH. 1996. Effects of green sun- Wyoming Game and Fish Department. fish (Lepomis cyanellus) predation on survival and habitat use of plains killifish (Fundulus zebrinus). LITERATURE CITED Southwestern Naturalist 41:155–160. LUTTRELL, G.R., A.A. ECHELLE, W.L. FISER, AND D.J. AADLAND, L.P. 1993. Stream habitat types: their fish EISENHOUR. 1999. Declining status of two species of assemblages and relationship to flow. North Ameri- the Macrhybopsis aestivalis complex (Teleostei: Cy- can Journal of Fisheries Management 13:790–806. prinidae) in the Arkansas River basin and related BAIN, M.B., J.T. FINN, AND H.E. BROOKE. 1988. Stream- effects of reservoirs as barriers to dispersal. Copeia flow regulation and fish community structure. Ecol- 1999:981–989. ogy 69:382–392. MATTHEWS, W.J. 1988. North American prairie streams as BAXTER, G.T., AND M.D. STONE. 1995. Fishes of Wyoming. a system for ecological study. Journal of the North Wyoming Game and Fish Department, Cheyenne. American Benthological Society 7:387–409. CARLSON, C.A., AND R.T. MUTH. 1989. The Colorado River: ______. 1998. Patterns in freshwater fish ecology. Chapman lifeline of the American Southwest. Pages 220–229 and Hall, New York. in D.P. Dodge, editor, Proceedings of the Interna- MCMAHON, T.E., AND D.H. BENNETT. 1996. Walleye and tional Large Rivers Symposium. Canadian Special northern pike: boost or bane to Northwest fisheries? Publication of Fisheries and Aquatic Sciences 106. Fisheries 21(8):6–13. 2004] EFFECTS OF IMPOUNDMENT ON NATIVE FISHES 63

MOYLE, P.B. 1986. Fish introduction into North America: impacts of hydroelectric development. Environmen- patterns and ecological impact. Pages 27–43 in H.A. tal Reviews 5:27–54. Mooney and J.A. Drake, editors, Ecology of biologi- ROSENBERG, D.M., P. MCCULLY, AND C.M. PRINGLE.2000. cal invasions of North America and Hawaii. Springer- Global-scale environmental effects of hydrological Verlag, New York. alterations: introduction. BioScience 50:746–751. OTT, R.L. 1993. An introduction to statistical methods and SAS INSTITUTE, INC. 1996. SAS/STAT user’s guide for per- data analysis. Duxbury Press, Belmont, CA. sonal computers. Version 6. SAS Institute, Inc., Cary, PATTON, T.M., AND W.A. HUBERT. 1993. Reservoirs on a NC. Great Plains stream affect downstream habitat and STANFORD, J.A., J.V. WARD, W.J. LISS, C.A. FRISSELL, R.N. fish assemblages. Journal of Freshwater Ecology 8: WILLIAMS, J.A. LICHATOWICH, AND C.C. COUTANT. 279–285. 1996. A general protocol for restoration of regulated PFLIEGER, W.L. 1997. The fishes of Missouri. Revised edi- rivers. Regulated Rivers 12:391–413. tion. Missouri Department of Conservation, Jeffer- TABOR, R.A., AND W.A. WURTSBAUGH. 1991. Predation risk son City. and the importance of cover for juvenile rainbow POFF, N.L., AND D.D. HART. 2002. How dams vary and trout in lentic systems. Transactions of the American why it matters for the emerging science of dam re- Fisheries Society 120:728–738. moval. BioScience 47:659–668. TAYLOR, C.A., J.H. KNOUFT, AND T.M. H ILAND. 2001. Con- POSTEL, S.L., G.C. DAILY, AND P.R. EHRLICH. 1996. Human sequences of stream impoundment on fish communi- appropriation of renewable fresh water. Science 271: ties in a small North American drainage. Regulated 785–788. Rivers: Research and Management 17:687–698. PRINGLE, C.M., M.C. FREEMAN, AND B.J. FREEMAN. 2000. WARD, J.V., AND A. STANFORD. 1979. The ecology of regu- Regional effects of hydrologic alterations on riverine lated streams. Plenum Publishing, New York. macrobiotia in the New World: tropical-temperate WINSTON, M.R., C.M. TAYLOR, AND J. PIGG. 1991. Upstream comparisons. BioScience 50:807–823. extirpation of four minnow species due to damming RICHTER, B.D., D.P. BRAUN, M.A. MENDELSON, AND L.L. of a prairie stream. Transactions of the American MASTER. 1997. Threats to imperiled freshwater fauna. Fisheries Society 120:98–105. Conservation Biology 11:1081–1093. ROHLF, F.J. 1990. NTSYS-pc: numerical taxonomy and Received 18 March 2003 multivariate analysis system. Exeter Software, Accepted 29 December 2003 Setauket, NY. ROSENBERG, D.M, F. BERKES, R.A. BODALY, R.E. HECKY, C.A. KELLY, AND J.W.M. RUDD. 1997. Large-scale Western North American Naturalist 65(1), © 2005, pp. 64–69

EFFECT OF PERCH SITES ON MOURNING DOVE NEST DISTRIBUTION

Paul M. Meyers1,2, Michael R. Conover1, and John A. Bissonette3

ABSTRACT.—We examined the effect of natural and artificial perch sites on Western Mourning Dove (Zenaida macroura marginella) nest density in shrubby habitat. Nest density in shrubs was strongly correlated with the density of natural perch sites. This relationship occurred in 2 vegetation types with differing shrub composition. The correlation was stronger in areas with a homogenous shrub layer. Nest density was also higher in plots with artificial perch sites than in adjacent control plots. Nest density increased between years in plots where artificial perch sites were constructed but remained the same in adjacent control plots. Knowledge of perch-site effects has practical management applications in areas where doves nest in shrubby habitat. The possibility exists that perch sites could be managed for nest density in these habitats.

Key words: artificial perches, Mourning Dove, nest density, nest distribution, nesting, perch sites, Zenaida macroura marginella.

Mourning Doves are one of the most abun- area show that Mourning Dove nests are invari- dant bird species in the U.S. (Peterjohn et al. ably placed near prominent perches. 1994). Historically, populations have prospered Several researchers have examined perch- in areas of human presence, but over the last site selection for grassland birds (Castrale 1983, 37 years, Mourning Dove populations in the Witter and Cuthill 1992), and a few studies have western U.S. have gradually declined (Dolton looked at perch-site effects on bird presence and Holmes 2002). Ostrand et al. (1998) docu- or density during the breeding season (Lack mented a population decline in the Fillmore, 1933, Lack and Venables 1939, Harrison and Utah, area between the early 1950s and the Brewer 1979, Knodel-Montz 1981). However, early 1990s. Meyers (1994) documented a con- we found no studies that have tested the effect current decrease in nest density for the same of perch sites on nest densities. We used 2 area. For both time periods, nesting near Fill- methods to examine whether a relationship more occurred primarily in shrubby vegetation existed between perch sites and nest density (Dahlgren 1955, Meyers 1994). The quantity in our study area. First, we correlated nest of shrubby habitat has remained stable in this density with the density of naturally occurring area, but the number of trees has declined perch sites. Second, we introduced artificial (Meyers 1994). Although trees were not impor- perch sites into the nesting habitat and recorded tant for nest sites, we hypothesized that they changes in nest density. may have been important for perching areas. Many researchers have speculated on the STUDY AREA AND METHODS importance of perch sites in avian habitat (Kendeigh 1941, Hilden 1965, Zimmerman The study area was located about 2 km 1971, Wiens 1973). Territorial birds in general northwest of Fillmore, Utah, in an area called rely heavily on singing for territorial defense the Old Fields. The site was at the eastern (Welty 1982:280). Mourning Doves use perch edge of an arid basin (Pahvant Valley), with an sites for 2 reasons: (1) mate attraction before annual rainfall of 37.9 cm. Chalk Creek, which nest-site selection and (2) territory mainte- runs through the site, was diverted into irriga- nance afterward (Frankel and Baskett 1961, tion canals, and the doves nested in these Jackson and Baskett 1964). Our observations riparian areas. The concentrated area of re- from systematic nest surveys in the Fillmore search consisted of approximately 12.6 km of

1Department of Fisheries and Wildlife, Utah State University, Logan, UT 84322-5210. 2Present address: USDA Forest Service, Cordova Ranger District, Box 280, Cordova, AK 99574. 3Utah Cooperative Fish and Wildlife Research Unit, Utah State University, Logan, UT 84322-5290.

64 2005] PERCH SITES AND NEST DISTRIBUTION 65 riparian vegetation (8.1-km irrigation canal and Artificial Perch Sites 4.5-km creek) branching through approxi- and Nest Density mately 570 ha of farmland. This riparian vege- We constructed artificial perches in 16 tation consisted of a continuous line of shrubs riparian plots from 4 to 13 May 1993. Perches interspersed with trees. A few small areas along consisted of two 5.1 × 5.1-cm stakes 4.3 m high the irrigation canals contained a 2nd-story, with a 30-m length of rope running between closed canopy. them. Each plot contained 2 sets of stakes with Vegetation differed between the creek and the rope running lengthwise on either side of the irrigation canals. Predominant vegetation the riparian vegetation, offset 15 m (Fig. 1). along the irrigation canals included willow Plots extended 15 m beyond each end of the (Salix spp.), squawbush (Rhus trilobata), wild perch sites and were 75 m long. Perches were rose (Rosa sp.), and golden currant (Ribes placed on the outer edge of the shrubby vege- aureum). Vegetation along Chalk Creek largely tation (12.5 m ± 5.6 m across) and protruded consisted of squawbush, occasionally inter- 1.5–3 m above the shrub canopy. Experimen- spersed with single, tall (≥5 m) trees, such as tal plots (i.e., those containing artificial perches) willow, cottonwood (Populus spp.), locust and control plots were placed randomly in a (Robinia spp.), and boxelder (Acer negundo). paired design with a 50-m buffer zone between them. Plots were placed at least 50 m Natural Perch Sites away from trees or power lines. and Nest Density Nest searches in the plots and buffers were We searched for Mourning Dove nests once conducted once per week over the entire breed- per week 15 April–5 September 1992 and ing season the year before the study and the 1 May–5 September 1993 (i.e., the entire breed- year of the study. Only active Mourning Dove ing season). We found nests by walking the nests were recorded. Comparisons of nests in riparian corridors and flushing nesting doves experimental plots with nests in control plots by agitating the vegetation with an aluminum were made with a Poisson regression employ- pole. We measured all vegetation and struc- ing a repeated function (SAS Institute, Cary, tures (e.g., trees, telephone poles) ≥5 m tall NC). This technique is analogous to a paired with a clinometer. These were defined as perch t test but is used for count data with a Poisson sites. All nests and perch sites were plotted distribution. onto aerial photographs and then transferred Doves are indeterminate nesters (i.e., a sin- to orthophotos. gle pair will produce many nests in sequence All riparian areas were partitioned into con- throughout the season), and subsequent nests in the same plot may be from the same pair tiguous plots. Although most of the habitat that previously nested there; therefore, subse- was linear, some shrubs were scattered out- quent nests may not be independent. We side the vegetation line. To capture these areas assumed that any subsequent nest produced with a consistent distance from plot center, we in the same plot was a renesting attempt, and used circular plots. Numbers of perch sites comparisons were made with and without these and nests in each plot were compared with a additional nests. To calculate power, we used Pearson’s correlation. This design created some the formula for a paired t test. A Poisson regres- effective overlap because perch sites near plot sion has similar or slightly less power (SAS boundaries may have affected neighboring Institute, Cary, NC). plots. Larger plots created fewer boundary effects, so the analysis was run using both 200- RESULTS m and 400-m-diameter plots to examine the effect of plot size. We also analyzed the data Natural Perch Sites after removing plots devoid of nests. Doing so and Nest Density removed plots that were unsuitable for nesting We recorded 90 nests, 78% of which occurred based on reasons other than lack of perch sites. in shrubs or on the ground. Mourning Dove Plots that contained closed-canopy tree cover nest density and perch site density were cor- >25% were excluded because of 2 added vari- related (r = 0.72, n = 33, P < 0.001; Fig. 2). ables: (1) cover and (2) an additional type of The relationship was stronger along Chalk nest substrate. Creek (r = 0.85, n = 11, P = 0.001) than 66 WESTERN NORTH AMERICAN NATURALIST [Volume 65

Fig. 1. Experimental design for introduction of artificial perch sites into Mourning Dove nesting habitat, Fillmore, Utah, 1993.

along the irrigation canals (r = 0.67, n = 22, P imental plots. Power for these data was low (β = 0.001). The correlation remained significant < 0.23 at α = 0.05), so statistical significance when plots barren of nests (r = 0.65, n = 24, was set at P = 0.10 to reduce the chance of P < 0.001) were omitted, and strengthened type II errors. when plot size was increased to 400 m in Of 11 sites surveyed in the pretreatment diameter (r = 0.81, n = 11, P = 0.003). The year, 2 nests occurred in the experimental plots correlation between number of nests and and 4 in the control plots. Following perch- perch sites remained significant (r = 0.61, n = site construction in the 2nd year, nests in 33, P < 0.001) with all tree-borne nests removed experimental plots increased to 7 (Z = 2.175, from analysis. P = 0.03) while nests in control plots de- The highest percentage of nests occurred creased to 2 (Z = –1.132, P = 0.26). For the in squawbush, which accounted for 86% of year of the study only, significantly more nests nests along the creek and 13% along the irri- occurred in experimental plots than in control gation canals. Willow was the next most im- plots (Z = 2.025, P = 0.043) when renests portant plant type, accounting for 20% of nests were excluded. When these nests were in- along irrigation canals. Mean nest height was cluded, no significant difference occurred (Z 1.7 m ± 1.2 m. = 0.279, P = 0.12). All nests occurred in or around shrubby Artificial Perch Sites habitat. Thirteen nests occurred in squaw- and Nest Density bush, 2 in sagebrush, 2 in willow shrub, and 1 We found 9 Mourning Dove nests in exper- on a fenceline overgrown with willow and imental plots, 4 in control plots, and 5 in bedstraw (Galium triflorum). Mean nest height buffer areas. Highest nest density occurred in was 1.0 m ± 0.3 m. Paired experimental and a buffer area situated between adjacent exper- control plots had similar vegetation type and 2005] PERCH SITES AND NEST DISTRIBUTION 67

Fig. 2. Relationship between perch sites and Mourning Dove nests, Fillmore, Utah, 1993. Solid markers indicate multiple occurrences. structure. Because pairs were close together, relation may have been higher than what we the riparian vegetation generally did not change measured. As noted, the correlation between along the length of the experimental unit. nests and perch sites increased with increasing plot size, supporting this hypothesis. DISCUSSION The correlation remained strong after excluding tree-borne nests, suggesting that trees were Natural Perch Sites important mainly as perch sites rather than and Nest Density nest substrates. The correlation also remained The correlation between nests and perches strong after we excluded plots barren of nests. was significant for both creek and canal habi- This test removed any plots that may have tat, suggesting that natural perch sites positively been unsuitable for nesting due to reasons other affected nest density. Because plots were con- than a lack of perch sites. tiguous and plot boundaries were somewhat Chalk Creek showed the strongest relation- arbitrary, there was potential for lack of inde- ship between perches and nest density. As pendence along plot borders. This effect would noted, vegetation differed between the creek tend to lower the level of correlation, however, and the irrigation canals. The creek contained because areas near plot borders may have a homogenous vegetation structure, mainly a affected neighboring plots and obscured squawbush-lined bank, which was lightly influ- within-plot effects. As a result, the actual cor- enced by human manipulation. Conversely, 68 WESTERN NORTH AMERICAN NATURALIST [Volume 65 irrigation canals were predominantly willow perch-site design. A better understanding of but also contained more diverse shrub species. the role of perch sites on nesting doves could In addition, canals often displayed radical veg- lead to practical management tools for nesting etation shifts at property borders. The lower doves in shrubby landscape. correlation along irrigation canals may have reflected this variation. That is, many irriga- ACKNOWLEDGMENTS tion canal plots may have contained marginal nesting habitat, whereas all creek plots con- We thank the Utah Department of Wildlife tained similar vegetation. Chalk Creek, there- Resources for their generous support, espe- fore, should represent the clearest picture of cially J.A. Roberson and D.C. Larsen. We also perch-site effects. The results show that the thank S.L. Durham for statistical advice, J.A. differing vegetation between the creek and Gessaman for manuscript reviews, T.D. Cook irrigation canals probably affected nest density, for help in data collection, W.D. Ostrand for but that perch sites appeared to be a larger valuable input, and T.R. Schroeder for help in factor in determining nest-site selection. perch construction.

Effect of Artificial Perch Sites LITERATURE CITED on Nest Density CASTRALE, J.S. 1983. Selection of song perches by sage- Overall nest density in the experimental and brush-grassland birds. Wilson Bulletin 95:647–655. control plots was low. Despite the low power CHERRY, S. 1998. Statistical tests in the publications of the that resulted, between-year comparisons showed Wildlife Society. Wildlife Society Bulletin 26:947–953. positive effects of artificial perches. Same-year DAHLGREN, R.B. 1955. Factors affecting Mourning Dove comparisons were somewhat conflicting, but populations in Utah. Master’s thesis, Utah State Uni- the low power greatly increased the chance of versity, Logan. DOLTON, D.D., AND R.D. HOLMES. 2002. Mourning Dove failing to detect a difference. In these instances population status, 2002. U.S. Fish Wildlife Service, the absence of statistical significance must be Laurel, MD. 30 pp. interpreted with caution (Cherry 1998, Johnson FRANKEL, A.I., AND T.S. B ASKETT. 1961. The effect of pair- 1999). Taken together in consideration of sta- ing on cooing of penned Mourning Doves. Journal of Wildlife Management 25:372–384. tistical power, the between-year and within- HARRISON, K.G., AND R. BREWER. 1979. The role of ele- year results suggest that artificial perches pos- vated perch sites and mowing in the distribution of itively affected nest density at our study site. grassland and savanna birds. Jack-Pine Warbler 57: Additional evidence for perch-site effects 179–183. exists in the buffer data. Buffer areas produced HILDEN, O. 1965. Habitat selection in birds. Annales Zoo- logici Fennica 2:53–75. more total nests than control areas even though JACKSON, G.L., AND T.S. B ASKETT. 1964. Perch-cooing and buffer areas were smaller (50 m vs. 75 m). Buffer other aspects of breeding behavior in Mourning areas were adjacent to or between perch-site Doves. Journal of Wildlife Management 28:293–307. plots, and perch sites may have influenced these JOHNSON, D.H. 1999. The insignificance of statistical significance testing. Journal of Wildlife Management areas. Larger experimental plots may be nec- 63:763–772. essary in future experiments to cover these KENDEIGH, S.C. 1941. Birds of a prairie community. Con- adjacent areas. dor 43:165–174. In summary, data from the natural correla- KNODEL-MONTZ, J.J. 1981. Use of artificial perches on tion indicate a positive relationship between burned and unburned tallgrass prairie. Wilson Bul- letin 93:547–548. perch sites and nest density. In addition, LACK, D. 1933. Habitat selection in birds with special ref- between-year comparisons from the artificial erence to the effects of afforestation on the Breck- perch sites suggest a positive perch-site effect. land avifauna. Journal of Animal Ecology 2:239–262. Same-year comparisons from the artificial perch LACK, D., AND L.S.V. VENABLES. 1939. The habitat distri- sites were conflicting but marginally signifi- bution of British woodland birds. Journal of Animal Ecology 8:39–71. cant as well. Overall, we conclude that perch MEYERS, P.M. 1994. Assessing Mourning Dove population sites appear to have a positive effect on nest declines: changes in nesting dynamics and the role density, and we suggest that they are an impor- of perch sites. Master’s thesis, Utah State University, tant habitat component for nest-site selection Logan. for Mourning Doves within our study area. OSTRAND, W.D., P.M. MEYERS, J.A. BISSONETTE, AND M.R. CONOVER. 1998. Changes in land use as a possible We suggest further investigation with larger factor in Mourning Dove population decline in cen- experimental plots and a larger and sturdier tral Utah. Journal of Field Ornithology 69:192–200. 2005] PERCH SITES AND NEST DISTRIBUTION 69

PETERJOHN, B.G., J.R. SAUER, AND W.A. LINK. 1994. The WITTER, M.S., AND I.C. CUTHILL. 1992. Strategic perch 1992 and 1993 summary of the North American Breed- choice for bill-wiping. Animal Behaviour 43:1056– ing Bird Survey. Bird Populations 2:46–61. 1058. WELTY, J.C. 1982. The life of birds. 3rd edition. CBS Col- ZIMMERMAN, J.L. 1971. The territory and its density de- lege Publishing. New York. pendent effect in Spiza Americana. Auk 88:591–612. WIENS, J.A. 1973. Interterritorial habitat variation in Grasshopper and Savannah Sparrows. Ecology 54: Received 30 December 2003 877–884. Accepted 12 April 2004 Western North American Naturalist 65(1), © 2005, pp. 70–79

ATTACK AND BROOD PRODUCTION BY THE DOUGLAS-FIR BEETLE (COLEOPTERA: SCOLYTIDAE) IN DOUGLAS-FIR, PSEUDOTSUGA MENZIESII VAR. GLAUCA (PINACEAE), FOLLOWING A WILDFIRE

Catherine A. Cunningham1, Michael J. Jenkins1, and David W. Roberts1

ABSTRACT.—In 1994 ground fire ignited in forests of Douglas-fir, Pseudotsuga menziesii var. glauca (Mirb.) Franco, on Beaver Mountain, Utah. The Douglas-fir beetle, Dendroctonus pseudotsugae Hopkins, attacked a range of moderately fire-injured host conifers in 1995. Logistic regression models run for 1995 data illustrated that 1 year after the fire event the Douglas-fir beetle selected and attacked large-diameter Douglas-fir with 60%–80% bole char, 60%–80% crown volume scorch, and 50%–70% probability of mortality due to fire. In 1996 beetle preference shifted to smaller- diameter trees with lighter fire injury, because most large, fire-damaged conifers were colonized by beetles in 1995. Although beetle populations did not reach outbreak proportions outside the fire boundary, host selection shifted to green trees in 1997 along the burn perimeter. Log linear analysis indicated that increased brood production was conditioned by increased diameter and moderate fire damage to the trees.

Key words: Douglas-fir, bark beetles, wildfire, disturbance ecology.

Relatively few empirical studies concerning to inhibit beetle colonization, allowed success- the interaction of wildland fire and bark beetle ful brood development and emergence. Fire colonization have been conducted (Weatherby damage of especially large-diameter Douglas- et al. 1993, Ryan and Amman 1996, Bebi et al. fir may affect host defense mechanisms and 2003). Yet, multiple disturbance agents are permit greater colonization. critical to the functioning of most dynamic, On 16 August 1994 a lightning-ignited crown sustainable forest ecosystems. Interior Douglas- fire spread in a forest of subalpine fir, Abies fir is a prominent conifer of the western United lasiocarpa (Hook.) Nutt. (Pinaceae), on Beaver States that extends throughout the Rocky Moun- Mountain. The fire burned 247 ha of land from tains and into British Columbia. This species the mid-slope above Franklin Basin to the upper covers a distance of nearly 4500 km (Hermann northwestern aspect of the mountain and and Lavender 1990). Public land managers are became a surface fire as it entered the multi- concerned with Douglas-fir beetle attacking storied Douglas-fir, Pseudotsuga menziesii var. live, fire-injured trees and establishing suc- glauca (Mirb.) Franco (Pinaceae), stands along cessful brood, which later overcome green the edge of the main fire front. Prior to the Douglas-fir trees adjacent to scorched stands. fire, endemic populations of Douglas-fir beetles Understanding how fire damage contributes were detected on Beaver Mountain. During the to increasing the insect population is critical 1995 spring flight season the beetle population for preventing beetle epidemics (Jenkins 1990). increased substantially in the fire-damaged Although information about Douglas-fir beetle Douglas-fir stands. host selection may help forest managers iden- The 1st objective of this research was to tify susceptible trees, research focused on bark explore whether bark beetles prefer fire-dam- beetle brood production is also important from aged over non-fire–damaged and fire-killed host a population standpoint. Some beetle-attacked trees. The 2nd objective was to quantify brood Douglas-fir trees on Beaver Mountain, Utah, emergence in Beaver Mountain Douglas-fir were able to repel attacking beetles with resin forests for each beetle-attacked tree following flow. Conversely, other Douglas-fir trees, unable the August 1994 fire.

1Department of Forest, Range and Wildlife Sciences, Utah State University, Logan, UT 84322-5230.

70 2005] DOUGLAS-FIR BEETLE FIRE INJURY 71

MATERIALS AND METHODS ened bark surface and obvious root and cambial death exhibited by exfoliated bark (Ryan Study Area and Sample 1982a). We scheduled fire-data collection on Beaver Mountain peak (2699 m) is located Beaver Mountain for the summer of 1996 after in the Bear River Range of northern Utah. The the 2nd growth season following the fire event. study area was located on this peak between Fire-caused mortality is best observed after 2 2500 m and 2622 m in elevation (41′58″N, spring seasons while the tree attempts recov- 111′33″W) and included Douglas-fir trees on ery of its lost energy-fixing tissue and when 7 plots. Plots were selected in 7 areas where delayed cambial injury is apparent (Ryan and low-intensity, surface spot fires resulted in low Amman 1994). to moderate fire injury to Douglas-fir. Plot Scorching the stem’s dead outer bark may boundaries were delineated in Douglas-fir not necessarily damage the tree in all cases stands by observing blackened surface fuels, (Ryan 1982a). Therefore, an indirect means of charred bark, torched branches, and fire-dam- analyzing both crown and bole injury was also aged needles of individual trees. performed in the analysis. Reinhardt and On each of the 7 plots varying in size from Ryan’s (1989) revised PM fire effects equation, 2 ha to 5 ha, all Douglas-fir trees greater than 30.5 cm in diameter at breast height (dbh) [PM = 1/ (1+ e (–1.941+BF+CF))], (1) were sampled. Furniss (1962) established that beetles generally prefer to attack larger-diam- was used to calculate fire injury for each of the eter trees. Given this criterion, 997 individual 997 sample trees and to confirm results for the trees were evaluated. Each sample tree was measured fire damage variables. Components tagged with a unique number for identification of the equation calibrated for Douglas-fir were and relocation. A few individual fire-killed trees as follows: were harvested in 1995 before the study was conducted. However, the number of sample Bark factor (BF) = 6.316(1 – exp [–0.3937BT]), (2) trees in the population over the course of field research was not reduced by salvage logging. where BT = bark thickness (cm) and CF = Fire-caused Injury crown volume scorched (%). Bark thickness was calculated from dbh adjusted specifically Degree of damage caused by the 1994 fire for Douglas-fir (Monserud 1979). on individual Douglas-fir trees was quantified by measuring percent of crown volume scorch Bark Beetle Attack (CVS) and percent of bole char. Probability of Behavior mortality (PM) due to fire ranging from 0 to 1 Beginning in the summer of 1996, we in- (Reinhardt and Ryan 1989) was also calculated. spected all plot trees after adult flight for Diameter of each sample tree at breast height insect activity on the lower bole to a height of was measured to the nearest 0.1 cm because 10 m to determine bark beetle invasion (Pasek dbh is perhaps the single most important fac- 1990, Rasmussen et al. 1996). Bark samples tor for analyzing an individual’s resistance to were removed on attacked trees to confirm fire (Ryan 1982a). We estimated CVS as the that damage was caused by Douglas-fir beetle. proportion of crown foliage and buds scorched Crown fade, dried pitch, and emergence holes relative to the amount of pre-fire live photo- distinguished 1995 insect activity from 1996 synthetic tissue (Ryan 1982a, Peterson 1985). insect activity. Entrance holes were not likely Different observers are capable of consistently to be confused with emergence holes because quantifying the affected crown within 10% (Ryan females typically bore into their host under 1982a), so CVS was categorized into 10% incre- the bark furrows and cover the entrance with ments. Different stem scorch heights typically frass. Successful beetle activity in the current correlate with the intensity of the fire at the year was identified by pitchy, red/orange-col- base of individual trees, resulting in varying duff ored boring dust at the base of the host tree. consumption and root crown/stem damage Conversely, fresh and clear pitch reflected (Norum 1976). Therefore, the extent of bole unsuccessful bark beetle attempts to colonize injury from the base of the tree up to 5 m was trees during the recent flight. Beetle activity also estimated visually by the extent of black- limited to a section of the tree left other parts 72 WESTERN NORTH AMERICAN NATURALIST [Volume 65 of the bole vulnerable to further attack and of successfully colonized trees accurately re- insect colonization during the 2nd season. flected brood survival. We placed square cages measuring 900 cm2 randomly on 31 tree boles Quantifying Douglas-fir that evidenced previous beetle attack in the Brood Success 1996 flight season. These mesh enclosures were A subset of host trees attacked by Douglas- stapled to a smoothed bark surface, similar to fir bark beetles in 1995–1996 was relocated in the method described by Lessard and Schmid July 1997. The resulting sample size used for (1990). We sealed cages tightly to prevent this 2nd analysis included 343 trees exhibiting escape, and the bottom ends were fastened ≥1 emergence hole on the lower 5 m of the together to form a funnel secured to a plastic tree bole. Two sections 900 cm2 (30 cm × 30 tube. A no-pest insect strip with Vapona® as cm) on the lower stem of each beetle-selected its active lethal ingredient was placed in the Douglas-fir tree were chosen, and we counted container to kill collected insects and to prevent the number of beetle emergence holes. Dou- consumption by other collected organisms. We glas-fir beetle exit holes are aligned with pupal collected emerging adults weekly and counted chambers and were not confused with ventila- them in the laboratory. Mesh cages were re- tion holes, which are formed in the egg gallery. moved from trees and emergence holes were We also observed wood borer and smaller counted. Thick bark was smoothed to reveal ambrosia beetle pinholes on the host bark sur- all possible beetle holes against a flush surface. face. We avoided measuring obvious fire scars Statistical Analysis from previous fires, which are devoid of bark and cambium. Chi-square analyses were performed com- Generally, the north and south aspects of paring presence and absence of beetle attacks the tree were sampled. Furniss (1962) con- in fire-damaged, fire-killed, and non-fire– cluded that although beetle attack density was affected trees for each field season to deter- greatest on the northern aspect of the stem, mine annual beetle host selection. Each year many Douglas-fir were observed to have greater the same host population was measured, but brood production on the southern exposure. previously insect-affected trees were elimi- The 2-count samples on each tree were pooled nated. Alpha values were adjusted for multiple and used in the analysis, rather than compared simultaneous inference. Chi-square values with or evaluated separately. The resulting sample probabilities <0.001 were considered statisti- area for calculated brood density was 1800 cm2 cally significant. for each Douglas-fir. We used stepwise logistic regression (Hamil- Schmitz and Rudinsky (1968) concluded ton 1992) to analyze annual beetle attack be- that colonizing Douglas-fir bark beetles show havior data, relating the log of the odds of bark little preference for any particular portion of beetle attack to a linear function of dbh, and the tree from the lower to the upper bole. quadratic functions of CVS, bole char, and Therefore, breast height was selected as an PM. Terms were added to the equation if re- acceptable and efficient region to quantify brood duction in deviance was statistically significant success (Furniss 1962, 1965, Lessard and given the change in degrees of freedom, simi- Schmid 1990, Pasek 1990). Ground fire affect- lar to the calculation of adjusted R2 in multi- ing the root crown and lower tree bole also ple linear regression. possibly attracted greater numbers of bark Log linear regression (McCullagh and Nelder beetles to that section of the stem. Although 1989) was used to analyze emergence hole Douglas-fir beetles will not colonize dead density for each successfully beetle-attacked cambium, these insects generally first occupy Douglas-fir in the sample population. Terms areas immediately adjacent to scorched bark were tested for significance by calculation of (Miller and Keen 1960). adjusted reduction in deviance. Additional log linear regressions were calculated on the caged Emergence Hole Count and sections of trees to estimate number of beetles Actual Brood Production emerging as a function of emergence hole Additional analyses were conducted in density. Total brood production could then be 1997 to ensure that the number of emergence estimated as product of emergence hole den- holes counted on the larger sample (n = 343) sity and beetle emergences per hole. 2005] DOUGLAS-FIR BEETLE FIRE INJURY 73

RESULTS A Fire Damage A total of 997 potential host Douglas-fir trees were inventoried in the study area; 389 (39%) were not fire damaged, 429 (43%) were damaged but not killed by fire, and 180 (18%) were killed by fire. Annual Beetle Host Selection In the spring of 1995, beetles inhabited a wide range of Douglas-fir trees in the area (Fig. 1A). Beetle colonization was most com- ∏2 B mon in fire-damaged trees ( 2 df = 244.062, P < 0.001), but also included 74 fire-killed trees presumably because sufficient phloem resources existed the 1st spring after the fire. Beetle preference for fire-damaged trees persisted ∏2 the 2nd year ( 2df = 165.6, P < 0.001; Fig. 1B), although an increasing number of non- damaged trees were selected. By 1997 beetles selected mostly non-fire–damaged trees, as most fire-damaged trees had been colonized during the previous 2 seasons. In 1995 the probability of beetle attack was significantly correlated (P < 0.001) with dbh, C CVS, bole char, and PM (Fig. 2). Beetles primarily selected moderately fire-damaged, large Douglas-fir trees, with probabilities approaching 1 for selected trees. Beetle-affected trees from the previous year were removed from the sample for 1996, leaving 635 potential host trees. Again, the probability of attack was correlated with CVS, bole char, and PM, but effects were smaller than in 1995, with probabilities approaching 0.50 for selected trees (Fig. 3). Tree diameter was negatively associated with probability of attack in combination with bole char (Fig. 3A) or CVS (Fig. 3B) and was statistically insignificant in combination with PM (Fig. 3C), reflecting the low availability of suitable large trees after the 1st year. Beetle-affected trees were again removed from the sample for 1997. Only CVS and PM were significantly correlated with colonization for 1997, and both exhibited exponentially decreasing probabilities as fire damage increased. Neither bole Fig. 1. Distribution of Dendroctonus pseudotsugae attack char nor dbh was significant. Plot location, on Douglas-fir trees by fire damage classes in 1995 (A), simply measured as the distance in meters to 1996 (B), and 1997 (C). the main fire front, was not significant in 1995 or 1996 but emerged as an important variable in predicting colonization in 1997. 74 WESTERN NORTH AMERICAN NATURALIST [Volume 65

Fig. 2. Logistic regression model for 1995 Dendroctonus pseudotsugae attack data for bole char (A), crown volume scorch (B), and probability of mortality (C). 2005] DOUGLAS-FIR BEETLE FIRE INJURY 75

Fig. 3. Logistic regression model for 1996 Dendroctonus pseudotsugae attack data for bole char (A), crown volume scorch (B), and probability of mortality (C). 76 WESTERN NORTH AMERICAN NATURALIST [Volume 65

Fig. 4. Log linear model of Dendroctonus pseudotsugae brood emergence as indexed by density of emergence holes for bole char (A), crown volume scorch (B), and probability of mortality (C). 2005] DOUGLAS-FIR BEETLE FIRE INJURY 77

Quantifying Douglas-fir Brood Success Bole char, dbh, CVS, PM, and plot location were all statistically significant variables in predicting brood emergence, with dbh showing the most significant effects. Opposing trends in predicted emergence were observed for bole char versus CVS or PM (Fig. 4A vs. Figs. 4B, 4C). Maximum emergence is predicted for large trees with moderate CVS (45%–70%), with predicted emergence densities of nearly 80 exit holes ⋅ 1800 cm–2. Number of beetles emerging per emergence hole increased significantly with increasing emergence hole density Fig. 5. Log linear model for Dendroctonus pseudotsugae (Fig. 5). emergence holes ⋅ 1800 cm–2 and actual adult emergence (n = 31). DISCUSSION

Annual host selection was conditioned by both tree size and relative extent of fire injury 1996 confirmed the greater relative importance to Douglas-fir trees. A significant effect for of damage to the crown, rather than injury to logistic regression models of 1995 data (Fig. 2) the stem, in conditioning a beetle attack re- confirmed the aggregation of Douglas-fir bee- sponse. Primarily a function of crown damage tles on predominantly larger hosts (ranging for larger, more attractive hosts, PM also ex- between 120 cm and 140 cm). Conversely, re- hibited a highly significant effect in 1995 and corded insignificance of dbh in the PM model 1996. Heat-caused injury to the photosyn- of 1996 data suggests that beetles attacked thetic crown has been widely accepted by fire smaller trees only because most mature Dou- ecologists as the most common source of conifer glas-fir had already been colonized in 1995. injury and mortality due to fire (Wagener 1961, In 1996 the highest probability of attack Peterson 1985, Peterson and Arbaugh 1986). calculated for bole char and CVS shifted from Conclusions drawn by Furniss (1965) follow- 60%–80% in 1995 to 50%–60% in 1996 for ing the Poverty Flat Fire demonstrated that larger-diameter trees. The highest probability beetle attack densities rose as host crown of attack calculated using PM decreased to injury increased, although successful Douglas- 35%–50% (Fig. 3). Relative significance of all fir beetle colonization declined dramatically in fire-damage variables conditioning beetle completely defoliated trees. Host conifers on attack declined from 1995 to 1996. However, Beaver Mountain that evidenced crown and beetles still selected moderately fire-weak- stem injury >80% by the 1995 flight season ened trees in 1996, demonstrating their ability were not generally attacked in 1995–1997. to sense altered host condition. Amman and However, in 1995 females aggregated on the Ryan (1991) similarly concluded that female surviving stem surfaces on 74 large host conifers bark beetles selected mature Douglas-fir that were observed to have suffered delayed exhibiting ≥50% basal cambial damage (not to fire mortality. The thicker bark on large trees exceed 80%). Following the 1989 Lowman likely insulated sections of the cambium from Fire Complex, Weatherby et al. (1993) extensive heating, delayed complete drying of reported that Douglas-fir bark beetles were the stem, and protected pockets of the phloem discovered in the study area colonizing trees resource. In 1996 and 1997, fire-killed Douglas- that reported ≥48% CVS. After the Yellow- fir, exhibiting exfoliated bark and dry phloem, stone fires, bark beetles were also found to were not colonized by beetles. attack host trees with moderate crown heating By 1997 both small- and large-diameter, not greater than 80% CVS (Amman and Ryan fire-weakened Douglas-fir had already been 1991). killed by beetles, and beetle host selection on Highly significant effects for CVS in logis- Beaver Mountain shifted, making plot location tic regression models for data in 1995 and in a marginally important term in explaining 78 WESTERN NORTH AMERICAN NATURALIST [Volume 65 beetle attack preference. Therefore, emerging Actual beetle emergence showed a trend of beetles were forced to aggregate in large num- increasing emergence with the number of exit bers on fewer (n = 53) relatively vigorous host holes counted for 31 trials. All circular holes trees. Small infestation centers of green Dou- sized to this specific bark beetle were counted. glas-fir developed along the perimeter of cer- A few adult reemergence exits and ventilation tain plots on northwestern and western aspects. holes were likely measured as well. These errors Significantly more beetle-preferred, fire-dam- in sampling may have slightly overestimated aged trees were in these field locations than the number of emergence holes (Schmitz and on sites farther from the main fire source. Rudinsky 1968). Conversely, some 1800-cm2 Moderately fire-damaged and large-diame- surfaces with 20–40 emergence holes exhib- ter trees were not only highly desirable for ited nearly twice as many emerging beetles attack by Douglas-fir beetles, but they were (Fig. 5). If Douglas-fir beetle attack densities also successfully colonized at higher densities were relatively high in host trees, then intra- than other potential hosts. Results of log linear specific competition among adjacent larvae models for quantitative brood emergence dem- could have partly regulated mining patterns onstrated that both tree diameter and different (McMullen and Atkins 1961, Schmitz and degrees of fire damage influenced brood sur- Rudinsky 1968). Mountain pine beetle emer- vival. gence by >1 young adult insect from a single Regardless of fire injury on individual hosts, exit hole is possible for similarly mass-colo- the predicted density of surviving brood rose nized lodgepole pine (Cole and Amman 1980). as tree diameter increased. The largest trees Therefore, surviving Douglas-fir larvae may measuring 120–140 cm recorded 50–80 exit create closely neighboring or coalesced pupal holes ⋅ 1800 cm–2 (Fig. 5). The greatest predicted chambers and encourage overlapping brood density of emerging beetles was established emergence. If this was the case, then actual for large trees, subjected to 50%–70% crown brood production in moderately crown-dam- damage or 45%–65% PM (Fig. 4). However, aged, large-diameter host trees was signifi- the relationship between bole char and emer- cantly higher than the results reported for the emergence density log linear models. gence density contradicts the results reported Although attack preference shifted in the for the other fire-damage variables. The graph 3rd season toward green host trees, fewer trees supports the substantial increase in brood pro- were attacked. The decline in the number of duction for larger-diameter trees, yet results beetle-affected conifers in 1997 may have been show a reduction in emergence holes with partly due to greater densities of bark beetles moderately stem-injured Douglas-fir. Biologi- necessary to mass colonize healthier conifers. cally, it makes sense that non-fire–affected This paper provides information useful in stems would allow greater brood survival and selecting trees for sanitation and salvage based larger emergence densities. Conversely, it does not only on direct fire mortality, but also on the not immediately seem reasonable that large likelihood that injury will result in beetle colo- brood densities would be supported in severely nization and mortality. Early postfire removal fire-weakened tree hosts. Yet, it is possible that of non-fire–damaged, bark beetle–susceptible large-diameter trees suffering severe delayed trees may decrease colonization, brood pro- mortality effects maintained sufficient resources duction, and subsequent beetle mortality. to permit aggregations of bark beetles on limited sections of the tree bole in 1995. These ACKNOWLEDGMENTS small pockets of bark beetle attack were likely recorded on severely damaged host trees, be- The authors thank Richard Cutler, Nancy cause the sampling method purposely avoided Roberts, Jim Long, Liz Hebertson, Jim Honaker, areas of the lower bole with exposed and dry Rick Stratton, Joel Godfrey, Janet Beales, cambium, charred by surface fire effects. A tree Laurel Anderton, Andrea Brunello, Kelly with a greatly reduced crown may still have (McCloskey) Polito, James Dailey, and Rick provided ample resources for regenerating bark Mowry of Utah State University; Kevin C. beetle populations on part of its stem or on the Ryan, Judith E. Pasek, Jerry Brunner, Jesse entire bole adjacent to fire scars (Miller and Logan, Barbara Bentz, Lynn Rasmussen, Jim Keen 1960). Vandygriff, Steve Munson, John Anhold, and 2005] DOUGLAS-FIR BEETLE FIRE INJURY 79

Dawn Hansen of the USDA Forest Service. northern Idaho and northwestern Montana. USDA This research was supported by the Utah Agri- Forest Service, INT-266. 8 pp. NORUM, R.A. 1976. Fire intensity–fuel reduction relation- cultural Experiment Station, Utah State Uni- ships associated with under-burning in larch/Dou- versity, Logan, and approved as journal paper glas-fir stands. Proceedings of Tall Timbers Fire 7056. Ecology Conference 14:559–572. PASEK, J.E. 1990. Douglas-fir beetle infestation following the Clover Mist fire on the Clark Fork Ranger Dis- LITERATURE CITED trict, Shoshone National Forest, Wyoming. USDA Forest Service, Biological Evaluation R2-90-1. 10 pp. AMMAN, G.D., AND K.C. RYAN. 1991. Insect infestation of PETERSON, D.L. 1985. Crown scorch volume and scorch fire-injured trees in the Greater Yellowstone Area. height: estimates of post-fire tree condition. Cana- USDA Forest Service, Intermountain Forest and dian Journal of Forest Research 15:596–598. Range Experiment Station, Research Note INT-398. PETERSON, D.L., AND M.J. ARBAUGH. 1986. Post-fire sur- 9 pp. vival in Douglas-fir and lodgepole pine: comparing BEBI, P., D. KULAKOWSKI, AND T.T. V EBLEN. 2003. Interac- the effects of crown and bole damage. Canadian tion between fire and spruce beetles in a subalpine Journal of Forest Research 16:1175–1179. Rocky Mountain forest landscape. Ecology 84: RASMUSSEN, L.A., G.D. AMMAN, J.C. VANDYGRIFF, R.D. 362–371. OAKES, S.A. MUNSON, AND K.E. GIBSON. 1996. Bark COLE, W.E., AND G.D. AMMAN. 1980. Mountain pine bee- beetle and wood borer infestation in the Greater Yel- tle dynamics in lodgepole pine forests: course of an lowstone Area during four post-fire years. USDA infestation. I. USDA Forest Service, Intermountain Forest Service, Intermountain Forest and Range Ex- Forest and Range Experiment Station, General Tech- periment Station, Research Paper INT-RP-487. 12 pp. nical Report INT-89. 56 pp. REINHARDT, E.D., AND K.C. RYAN. 1989. Estimating tree FURNISS, M.M. 1962. Infestation patterns of Douglas-fir mortality resulting from prescribed fire. Pages 41–44 beetle in standing and windthrown trees in southern in Proceedings from Prescribed Fire in the Inter- Idaho. Journal of Economic Entomology 55:486–491. mountain Region. ______. 1965. Susceptibility of fire-injured Douglas-fir to RYAN, K.C. 1982a. Techniques for assessing fire damage to bark beetle attack in southern Idaho. Journal of trees. Pages 1–11 in Proceedings from the Inter- Forestry 63:8–11. mountain Fire Council and Rocky Mountain Fire HAMILTON, L.C. 1992. Regression with graphics: a second Council. course in applied statistics. Wadsworth Publishing RYAN, K.C., AND G.D. AMMAN. 1994. Interactions between Company, Belmont, CA. 363 pp. fire-injured trees and insects in the Greater Yellow- HERMANN, R.K., AND D.P. LAVENDER. 1990. Pseudotsuga stone Area. USDA Forest Service, Intermountain menziesii (Mirb.) Franco. In: R.M. Burns and B.H. Forest and Range Experiment Station, Technical Honkola, technical coordinators, USDA Forest Ser- Report NPS/NRYELL/NRTR-93/XX. 13 pp. vice, Agricultural Handbook 654:527–540. ______. 1996. Bark beetle activity and delayed tree mor- JENKINS, M.J. 1990. The relationship between fire and tality in the Greater Yellowstone Area following the bark beetle attack in western North American forests. 1988 fires. Pages 1–8 in Ecological implications of Page c11-1 in Proceedings of the Symposium of For- fire in Greater Yellowstone. est Fire Research. International Conference of For- SCHMITZ, R.F., AND J.A. RUDINSKY. 1968. Effect of compe- est Fire Research, Coimbra, Portugal. tition on survival in western Oregon of the Douglas- LESSARD, E.D., AND J.M. SCHMID. 1990. Emergence, attack fir beetle. USDA Forest Service, Forest Research densities, and host relationships for the Douglas-fir Laboratory, Oregon State University, Corvallis, Re- beetle (Dendroctonus pseudotsugae) in northern Col- search Paper 8. 41 pp. orado. Great Basin Naturalist 50:333–338. WAGENER, W.W. 1961. Guidelines for estimating the sur- MCCULLAGH, P., AND J.A. NELDER. 1989. Generalized lin- vival of fire-damaged trees in California. USDA For- ear models. Chapman and Hall, London. 511 pp. est Service, Pacific Southwest Forest and Range MCMULLEN, L.H., AND M.D. ATKINS. 1961. Intraspecific Experiment Station, Miscellaneous Paper 60. 11 pp. competition as a factor in the natural control of the WEATHERBY, J.C., P. MOCETTINI, AND B.R. GARDNER. 1993. Douglas-fir beetle. Forest Science 7:197–203. A biological evaluation of tree survivorship within ______. 1962. On the flight and host selection of the Dou- the Lowman Fire boundary. USDA Forest Service, glas-fir beetle, Dendroctonus pseudotsugae. Canadian Intermountain Forest and Range Experiment Sta- Entomologist 94:1309–1324. tion, Report R4-94-06. 5 pp. MILLER, J.M., AND F. P. K EEN. 1960. Biology and control of the western pine beetle. USDA Forest Service, Mis- Received 5 August 2003 cellaneous Publication 800. 381 pp. Accepted 20 April 2004 MONSERUD, R.A. 1979. Relations between inside and outside bark diameter at breast height for Douglas-fir in Western North American Naturalist 65(1), © 2005, pp. 80–86

RESPONSE OF SAGEBRUSH STEPPE SPECIES TO ELEVATED CO2 AND SOIL TEMPERATURE

Melissa S. Lucash1,3, Blake Farnsworth2, and William E. Winner1

ABSTRACT.—Elevated atmospheric CO2 may cause long-term changes in the productivity and species composition of the sagebrush steppe. Few studies, however, have evaluated the effects of increased CO2 on growth and physiology of species important to this ecosystem. Since the response of plants to elevated CO2 may be limited by environmental factors, soil temperature was also examined to determine if low soil temperatures limit CO2 response. To determine how CO2 and soil temperature affect the growth of species native to the sagebrush steppe, bottlebrush squirreltail [Elymus elymoides (Raf.) Swezey], Thurber needlegrass (Stipa thurberiana Piper), and Wyoming big sagebrush (Artemisia triden- –1 –1 ° tata ssp. wyomingensis Beetle) were grown in ambient (374 mL L ) or high (567 mL L ) CO2 and low (13 C) or high (18°C) soil temperature for approximately 4 months. Although soil temperature affected the growth of squirreltail and needlegrass, temperature did not modify their response to elevated CO2. Total biomass of sagebrush was consistent across soil temperature and CO2 treatments, reflecting its slow-growing strategy. All 3 species had higher leaf water-use efficiency at elevated CO2 due to higher net photosynthesis and lower transpiration rates. We conclude that elevated CO2 and soil warming may increase the growth of grasses more than shrubs. Field studies in the sagebrush steppe are necessary to determine if differences in biomass, resulting from changes in CO2 and soil temperature, are exhibited in the field.

Key words: elevated CO2, soil temperature, Artemisia tridentata, Stipa thurberiana, Sitanion hystrix.

Sagebrush steppe, a major vegetation type 1996). In the field elevated CO2 increased net occupying approximately 45 × 106 ha in the photosynthesis (Morgan et al. 1994) and growth western United States, has undergone long- (Morgan et al. 2001) in plants native to the term changes in species composition due to short-grass steppe. Elevated CO2 increased overgrazing and introduction of exotic plants shoot production by 50% in a desert shrub com- (West 1983). Increasing atmospheric CO2 may munity in a high-rainfall year (Smith et al. 2000). also cause long-term changes in species com- Predicting how elevated CO2 will affect the position and productivity of rangelands (Mooney sagebrush steppe may be complicated by the et al. 1991, Polley 1997, Campbell et al. 2000, 2°–5°C global increase in air and soil tempera- Smith et al. 2000, Morgan et al. 2001). ture expected by 2300 (IPCC 2001). In one Only a few studies to date have quantified study in California annual grassland, the effect how elevated CO2 affects the growth and phys- of CO2 on growth was higher in ambient than iology of grasses and shrubs common to semi- elevated temperature plots (Shaw et al. 2002), arid systems. In one study CO2 enrichment whereas in the short-grass steppe the CO2 re- increased leaf weights of 3 grasses; water-use sponse was higher at elevated temperatures efficiency was higher at elevated CO2 due to (Coughenour and Chen 1997). Most studies of reduced stomatal conductance and higher net global warming test the simultaneous effects photosynthesis (Smith et al. 1987). Elevated of increased soil and air temperature, but few CO2 increased shoot biomass of Artemisia tri- have independently tested the effects of ele- dentata (Nutt.), but effects on leaf area and vated soil temperature and CO2 on plant root:shoot ratios were inconclusive (Johnson growth. In tomato (Lycopersicon esculentum and Lincoln 1990, 1991). In another study ele- Mill.), there were interactive effects of CO2 vated CO2 increased root but not shoot bio- and root temperature on root but not shoot mass of Artemisia tridentata (Klironomos et al. biomass (Yelle et al. 1987). No interactive effects

1Department of Botany and Plant Pathology, Oregon State University, Corvallis, OR 97331. 2Campbell Scientific, Inc., Logan, UT 84321. 3Present address: c/o Ruth Yanai, Department of Forestry, SUNY-ESF, Syracuse, NY 13210. Send reprint requests to [email protected].

80 2005] ELEVATED CO2 AND SOIL TEMPERATURE 81 were found in root or shoot growth of tussock ized coarse river sand and soil obtained from sedge (Eriophorum vaginatum L.; BassiriRad the experimental range (coarse to fine sandy et al. 1996). loam Holte-Milican complex; Lentz and Simon- The effects of soil temperature on growth son 1986). Flats were placed in continuously may differ between grasses and shrubs. Soil stirred tank reactors (CSTR) with a 16-hour temperature may alter growth and competi- photoperiod and constant day (24°C) and night tion by favoring the shoot growth of grasses, (15°C) temperatures. In April we transplanted since their apical meristems are located at the 2 seedlings of each species into pots (10 × 10 × soil surface (Engels 1994). Although the effects 25 cm) containing the soil mixture. The pots of soil temperature on plant growth may depend were thinned to 1 seedling. After soil analysis on growth form, the effects of elevated CO2 in April revealed that percentage nitrogen in are also dependent on growth rates, with higher pots was lower than N at the Northern Great growth stimulation in fast- than slow-growing Basin Experimental Range, we subsequently species (Poorter 1993). This could be particu- fertilized the plants each week with 57 mM larly detrimental to degraded rangelands if nitrogen, 10 mM phosphorus, and 17 mM fast-growing invasive species such as cheat- potassium. grass are stimulated by elevated CO2 (Smith et al. 1987, 2000). Experimental Design The objective of this study was to determine This experiment was a split-plot design in how atmospheric CO2 concentration and soil which the CSTR was designated as the whole temperature affect the growth and physiology plot. At the whole-plot level, 6 CSTR cham- of 3 species native to the sagebrush steppe. To bers were randomly allocated to ambient (374 isolate the effects of soil warming, we exposed ppm) and 6 chambers to high CO2 (567 ppm) seedlings to different soil temperature treat- treatment. These conditions reflected the local ments while keeping the air temperature con- concentration of atmospheric CO2 and the stant. We hypothesized that (1) elevated CO2 predicted CO concentration in the year 2050 and elevated soil temperature would increase 2 (IPCC 2001). Within each CO treatment, 3 root, shoot, and total biomass, (2) low soil tem- 2 chambers were randomly assigned high (18°C) peratures would limit growth responses to and 3 were assigned low (13°C) soil tempera- CO , and (3) elevated CO would increase 2 2 tures. The low soil temperature treatment re- water-use efficiency of all 3 species. We also flects average soil temperature in March at a hypothesized that the fast-growing grass, bottle- sagebrush site in south central Washington brush squirreltail [Elymus elymoides (Raf.) Swezey], would be the most responsive to (Black and Mack 1986), while the high soil treatment represents a 5°C increase in tem- CO2; Thurber needlegrass (Stipa thurberiana Piper), a slow-growing grass, would be inter- perature predicted by 2300 (IPCC 2001). This mediate in its response; and the slow-growing design allowed for 3 replicates of each CO2 shrub, Wyoming big sagebrush (Artemisia tri- and soil temperature treatment combination. dentata ssp. wyomingensis Beetle), would be At the subplot level, each chamber contained the least responsive. We also expected soil tem- 36 pots (12 pots per species per chamber). peratures to have a greater effect on grasses Treatment Conditions than shrubs. During the experiment, the 12 CSTR cham- METHODS bers were maintained at a relative humidity of 48% (Humicap, HMD 20, Vaisala Sensor Sys- Seed Collection and tems, Woburn, MA) using a steam generator Propagation (EHU-500, Armstrong Machine Works, Three Seeds from Wyoming big sagebrush, bottle- Rivers, MI). Chambers were illuminated by brush squirreltail, and Thurber needlegrass 1000-W metal halide lamps that subjected seed- were collected from the Northern Great Basin lings to photon flux of 600 µmol m–2 s–1 (SB- Experimental Range (199°43′W, 4 3 °29′N) in 190 Quantum Sensors, LI-COR, Inc., Lincoln, Harney County, Oregon, in November 1993. NE) and a 16/8 hour light/dark period. Air tem- In March 1994 we initially planted seeds in peratures were maintained at 15°C during the flats. Soil medium was a 3:1 mixture of steril- dark period and at 24°C during the light period. 82 WESTERN NORTH AMERICAN NATURALIST [Volume 65

Soil temperature was regulated indepen- root weight, and leaf area, we analyzed data dently of air temperature by using a root using the generalized linear models (GLM) chiller located at the base of each chamber. In procedure in SAS with error terms dictated by addition, a foam septum within each chamber the split plot design. When species interac- limited the mixing of air between the roots tions were significant at the P < 0.05 level, we and shoots. Soil temperature was measured analyzed the effects of CO2 and temperature hourly at a depth of 10 cm. Chambers allo- separately by species. Gas exchange measure- cated to the low soil temperature treatment ments were averaged over the 3 days (24, 26, ± ° were maintained at 13 1 C (sx–), while the and 30 June) and analyzed using GLM. high temperature chambers had an average root temperature of 18 ± 2°C. RESULTS Low CO chambers were maintained at CO 2 2 Effects of Elevated CO concentrations of 374 ± 24 ppm, while high 2 ± CO2 chambers averaged 567 5 ppm. To en- Total biomass of both grasses, squirreltail and sure adequate water supply for each species, needlegrass, was higher at elevated CO2, where- we watered seedlings every other day to field as biomass was similar between CO2 treat- capacity. Two pots of each species were ran- ments for the slow-growing shrub, sagebrush domly selected and watered until water was (Fig. 1). Since squirreltail has a higher growth observed at the base. The volume of water rate than needlegrass, we expected CO2 enrich- added to each pot was averaged between the 2 ment would stimulate the growth of squirrel- plants, and the average volume of water was tail more than needlegrass. Contrary to our added to all plants of that species. hypothesis, elevated CO2 had similar effects on squirreltail and needlegrass; CO2 enrich- Gas Exchange Analysis ment increased their total growth by 14% and To compare physiological responses with 11%, respectively. CO2 and soil temperature between species, Elevated CO2 significantly increased root gas exchange measurements were taken before growth of needlegrass but not squirreltail or harvest. Photosynthesis, transpiration, and sagebrush (Fig. 1). Elevated CO2 did not sig- water-use efficiency were measured using a nificantly affect shoot growth. Needlegrass LI-COR 6200 Portable Photosynthesis System was the only species in which CO2 signifi- (LI-COR, Inc., Lincoln, NE) on the most re- cantly affected R:S ratios. At ambient CO2, cently emerged, fully expanded leaf of 9 ran- R:S ratios averaged 0.39 but were 0.55 at ele- domly selected plants from each treatment. vated CO2. Carbon dioxide concentrations had Measurements were taken 3 times daily on 20, no impact on leaf area or leaf thickness (data 26, and 30 June. A small window in the cham- not shown). ber door allowed for gas exchange measure- As expected, leaf water-use efficiency was ments within the CSTR. significantly higher at elevated CO2 in all species (Fig. 2). This increased efficiency resulted Growth Analysis from higher photosynthetic rates and lower Plants were harvested in July, approximately transpiration rates in response to high CO2, 127 days after planting. Squirreltail and needle- although these effects taken separately were grass were separated into shoots and roots, not statistically significant. and sagebrush was separated into stems, leaves, Effects of Soil and roots. We then oven-dried this material at Temperature 60°C for 72 hours. Root:shoot (R:S) ratios were calculated for each plant, and leaf area Higher soil temperature significantly in- was measured on a random sample of the pop- creased the root and total plant weights of ulation (9 plants per species per treatment) squirreltail and needlegrass and the shoot using a LI-3000 Portable Area Meter (LI-COR, weights of needlegrass (Fig. 3). Surprisingly, Inc., Lincoln, NE). soil temperature stimulated total growth of slow-growing needlegrass by 48% while increas- Statistical Analysis ing growth of slow-growing squirreltail by To determine the effects of CO2 and soil only 18%. Sagebrush was not significantly temperature on total plant weight, shoot weight, affected by soil temperature. Soil temperature 2005] ELEVATED CO2 AND SOIL TEMPERATURE 83

Fig. 1. Mean total plant, shoot, and root weights of squirreltail, needlegrass, and sagebrush exposed to ambient (370 ppm) or high (570 ppm) CO2. Error bars indicate standard Fig. 2. Mean water-use efficiency, photosynthetic rate, errors. N = 69–72 plants. and transpiration rate of squirreltail, needlegrass, and sagebrush exposed to ambient (370 ppm) or high (570 ppm) CO2. Error bars indicate standard errors. N = 6 plants. had no effect on photosynthesis, transpiration rates, or water-use efficiency (data not shown) growing squirreltail at elevated CO2 was sta- and did not significantly modify the response tistically indistinguishable from that of slow- to CO in any of the species. However, trends 2 growing needlegrass. Since CO2 stimulated indicate that growth stimulation of roots and the growth of grasses more than shrubs, in- shoots may be higher at low soil temperatures. creased CO2 may lead to changes in seedling competition between these 2 growth forms. DISCUSSION Higher overall biomass of squirreltail and needlegrass at elevated CO2 was primarily due The effects of elevated CO on growth are 2 to root growth, although CO2 increased root often dependent on potential growth rates, with biomass significantly only in needlegrass. Sage- higher stimulatory effects of CO2 in fast- than brush root biomass was similar between CO2 slow-growing species (Poorter 1993). As pre- treatments, which agrees with a previous study dicted, elevated CO2 increased biomass of (Johnson and Lincoln 1991). These results, squirreltail and needlegrass more than that of however, disagree with another study that found sagebrush. Unexpectedly, the growth of fast- sagebrush root biomass was higher at elevated 84 WESTERN NORTH AMERICAN NATURALIST [Volume 65

shoot weights of sagebrush to increase with elevated CO2 as found in a previous study with Artemisia tridentata ssp. tridentata (John- son and Lincoln 1990) grown at 650 µL L–1 for 3 months. Artemisia tridentata ssp. wyomingensis, used in this study, has slower growth rates (Bonham et al. 1991) and therefore may be less responsive to CO2 than Artemisia tridentata ssp. tridentata. Although not significant, photosynthesis was higher at elevated CO2 as previously found in many studies with range plants (e.g., Larigauderie et al. 1988). We found that elevated CO2 increased water- use efficiency of all 3 species, confirming previous studies with crop and rangeland plants (Kimball and Idso 1983, Dahlman et al. 1985, Larigauderie et al. 1988, Jackson et al. 1994). Needlegrass exhibited a 40% greater increase in water-use efficiency than sagebrush or squirreltail at elevated CO2. Since water is often limiting in rangelands, small increases in water- use efficiency of seedlings may cause relatively large changes in seedling survival and competition between species. Therefore, alterations in water-use efficiency in response to CO2 enrichment may cause relatively large changes in the structure of rangelands (Polley 1997). Soil temperature plays an important role in determining community composition of the sagebrush steppe (West 1983) and tallgrass prairie (DeLucia et al. 1992). High soil temperature increases growth rates (Benzioni and Dunstone 1988) and photosynthetic (Bassiri- Rad et al. 1993) rates of rangeland plants. As Fig. 3. Mean total plant, shoot, and root weights of squir- predicted, soil temperature increased total reltail, needlegrass, and sagebrush exposed to low (13°C) or high (18°C) soil temperature. Error bars indicate standard growth of both grasses, needlegrass and squir- errors. N = 69–72 plants. reltail. Sagebrush had consistent total and root growth across soil temperature treatments, confirming germination trials that indicate its CO2 (Klironomos et al. 1996). Elevated CO2 ability to tolerate a broad range of soil temper- affected R:S ratios of needlegrass but not atures ranging from 10° to 30°C (McDonough squirreltail or sagebrush. Among rangeland and Harniss 1974). species, there is no consistent pattern of car- We expected high soil temperatures to affect bon allocation in response to elevated CO2 shoot growth of both grasses (Larigauderie et since some species increase (Smith et al. 1987, al. 1991, Engels 1994), but soil warming signif- Larigauderie et al. 1988), decrease (Johnson icantly increased shoot biomass only of needle- and Lincoln 1991), or have no change (John- grass. Although higher soil temperatures have son and Lincoln 1990) in their R:S ratios. been reported to increase net photosynthesis Although previous studies indicate that (Duke et al. 1979, Day and Heckathorn 1991, CO2 stimulates shoot growth by 10%–150% in Vapaavuori et al. 1992) and decrease transpira- species native to the sagebrush steppe (Smith tion rates (Benzioni and Dunstone 1988), we et al. 1987, Johnson and Lincoln 1990), we found no significant changes in physiology in found no significant effects of CO2 on shoot response to soil temperature. This could be due biomass or leaf area. In particular, we expected to the small sample size in our study or the 2005] ELEVATED CO2 AND SOIL TEMPERATURE 85 small difference (5°C) between soil tempera- under controlled conditions. Physiologia Plantarum ture treatments. 74:107–112. BLACK, R.A., AND R.N. MACK. 1986. Mount St. Helens ash: These results suggest that elevated CO2 recreating its effects on the steppe environment and will stimulate the growth of squirreltail and ecophysiology. Ecology 67:1289–1302. needlegrass seedlings more than sagebrush, BONHAM, C., T. COTTRELL, AND J. MITCHELL. 1991. Infer- under conditions where water and nutrients ences for life history strategies of Artemisia tridentata subspecies. Journal of Vegetation Science 2: are not limiting. Since soil temperatures did 339–344. not affect growth responses to CO2, diurnal CAMPBELL, B.D., D.M. STAFFORD SMITH, A.J. ASH, J. FUHER, and seasonal changes in nutrient and water R.M. GIFFORD, P. HIERNAUS, S.M. HOWDEN, ET AL. availability may play a more important role in 2000. A synthesis of recent global change research regulating responses to CO . Since the plants on pasture and rangeland production: reduced un- 2 certainties and their management implications. Agri- in this study were harvested after 4 months to culture, Ecosystems and Environment 82:39–55. prevent root restriction in the pots, additional COUGHENOUR, M.B., AND D.X. CHEN. 1997. Assessment of studies are needed to assess how elevated grassland ecosystem responses to atmospheric change CO will affect mature individuals in the field. using linked plant-soil process models. Ecological 2 Applications 7:802–827. DAHLMAN, R.C., B.R. STRAIN, AND H.H. ROGERS. 1985. CONCLUSIONS Research on the response of vegetation to elevated atmospheric carbon dioxide. Journal of Environmen- This study suggests that elevated CO2 and tal Quality 14:1–8. soil temperature have the potential to alter DAY, T.A., AND S.A. HECKATHORN. 1991. Limitations of photosynthesis in Pinus taeda L. (loblolly pine) at growth and carbon partitioning of seedlings in low soil temperatures. Plant Physiology 96:1246–1254. the sagebrush steppe. In addition, elevated DELUCIA, E.H., S.A. HECKATHORN, AND T.A. DAY. 1992. CO2 and soil warming may affect grasses more Effects of soil temperature on growth, biomass allo- than shrubs. These controlled environment cation and resource acquisition of Andropogon ger- studies should pave the way for field studies ardii Vitman. New Phytologist 120:543–549. DUKE, S.H., L.E. SCHRADER, C.A. HENSON, J.C. SERVAITES, in the sagebrush steppe to determine whether R.D. VOGELZANG, AND J.W. PENDLETON. 1979. Low differences in carbon allocation, resulting from root temperature effects on soybean nitrogen metab- changes in CO2 and soil temperature, are ex- olism and photosynthesis. Plant Physiology 63: hibited in the field. Alterations in growth and 956–962. ENGELS, C. 1994. Effect of root and shoot meristem tem- carbon allocation in response to elevated CO2 perature on shoot to root dry matter partitioning and may potentially alter the competitive relation- the internal concentrations of nitrogen and carbohy- ships between species and influence succes- drates in maize and wheat. Annals of Botany 73: sional processes in the sagebrush steppe. 211–219. IPCC. 2001. Climate change 2001: the scientific basis. Contribution of Working Group 1 to the third assess- ACKNOWLEDGMENTS ment report of the Intergovernmental Panel on Cli- mate Change. Cambridge University Press, Cam- Research was funded by the U.S. National bridge, UK, and New York, USA. Biological Service (IB-OR-98-079). We wish to JACKSON, R.B., O.E. SALA, C.B. FIELD, AND H.A. MOONEY. thank Tony Svejcar, Ruth Yanai, John McNabb, 1994. CO2 alters water use, carbon gain, and yield for the dominant species in a natural grassland. and Erik Hobbie for their editorial comments; Oecologia 98:257–262. Paul Murtaugh, Mike Lasarev, and Pete Zimin- JOHNSON, R.H., AND D.E. LINCOLN. 1990. Sagebrush and ski for assistance in statistical analysis; and grasshopper responses to atmospheric carbon diox- Beth Parsons for assistance in the laboratory. ide concentration. Oecologia 84:103–110. ______. 1991. Sagebrush carbon allocation patterns and grasshopper nutrition: the influence of CO2 enrich- LITERATURE CITED ment and soil mineral limitation. Oecologia 87: 127–134. BASSIRIRAD, H., M.M. CALDWELL, AND K.A. MOTT. 1993. KIMBALL, B.A., AND S.B. IDSO. 1983. Increasing CO2: effects Effect of root cooling on photosynthesis of Artemisia on crop yield, water use and climate. Agriculture tridentata seedlings under different light levels. Water Management 7:55–72. Botanica Acta 106:223–227. KLIRONOMOS, J.N., M.C. RILLIG, AND M.F. ALLEN. 1996. BASSIRIRAD, H., D. TISSUE, J.F. REYNOLDS, AND F. S . C HAPIN Below-ground microbial and microfaunal responses III. 1996. Response of Eriophorum vaginatum to CO2 to Artemisia tridentata grown under elevated atmos- enrichment at different soil temperatures: effects on pheric CO2. Functional Ecology 10:527–534. 3– growth, root respiration and PO4 uptake kinetics. LARIGAUDERIE, A.B., B.A. ELLIS, J.N. MILLS, AND J. KUMME- New Phytologist 133:423–430. ROW. 1991. The effect of root and shoot temperatures BENZIONI, A., AND R.L. DUNSTONE. 1988. Effect of air and on growth of Ceanothus greggii seedlings. Annals of soil temperature on water balance of jojoba growing Botany 67:97–101. 86 WESTERN NORTH AMERICAN NATURALIST [Volume 65

LARIGAUDERIE, A.B., D.W. HILBERT, AND W.C. OECHEL. SHAW, M.R., E.S. ZAVALETA, N.R. CHIARIELLO, E.E. CLE- 1988. Effect of CO2 enrichment and nitrogen avail- LAND, H.A. MOONEY, AND C.B. FIELD. 2002. Grass- ability on resource acquisition and resource alloca- land responses to global environmental changes sup- tion in a grass, Bromus mollis. Oecologia 77:544–549. pressed by elevated CO2. Science 298:1987–1990. LENTZ, R.D., AND G.H. SIMONSON. 1986. A detailed soils SMITH, S.D., T.E. HUXMAN, S.F. ZITZER, T.W. CHARLET, inventory and associated vegetation of Squaw Butte D.C. HOUSMAN, J.S. COLEMAN, L.K. FENSTERMAKER, Range Experiment Station. Oregon State University ET AL. 2000. Elevated CO2 increases productivity Experiment Station, Corvallis. and invasive species success in an arid ecosystem. MCDONOUGH, W.T., AND R.O. HARNISS. 1974. Effects of Nature 408:79–82. temperature on germination in three subspecies of SMITH, S.D., B.R. STRAIN, AND T.D. S HARKEY. 1987. Effects big sagebrush. Journal of Range Management 27: of CO2 enrichment on four Great Basin grasses. 204–205. Functional Ecology 1:139–143. MOONEY, H.A., B.G. DRAKE, R.J. LUXMOORE, W.C. OECHEL, VAPAAVUORI, E.M., R. RIKALA, AND A. RYYPPO. 1992. Effects AND L.F. PITELKA. 1991. Predicting ecosystem re- of root temperature on growth and photosynthesis in sponses to elevated CO2 concentrations. BioScience conifer seedlings during shoot elongation. Tree Phys- 41:96–104. iology 10:217–230. MORGAN, J.A., H.W. HUNT, C.A. MONZ, AND D.R. LECAIN. WEST, N.E. 1983. Western intermountain sagebrush steppe. 1994. Consequences of growth at two carbon dioxide Pages 351–374 in N. West, editor, Temperature deserts concentrations and two temperatures for leaf gas and semi-deserts. Ecosystems of the world. Elsevier, exchange in Pascopyrum smithii (C3) and Bouteloua Amsterdam. gracilis (C4). Plant, Cell and Environment 17: YELLE, S., A. GOSSELIN, AND M.-J. TRUDEL. 1987. Effect 1023–1033. of atmospheric CO2 concentration and root-zone MORGAN, J.A., D.R. LECAIN, A.R. MOSIER, AND D.G. temperature on growth, mineral nutrition and nitrate MILCHUMAS. 2001. Elevated CO2 enhances water reductase activity of greenhouse tomato. Journal of relations and productivity and affects gas exchange American Society of Horticultural Science 112: in C3 and C4 grasses of the Colorado shortgrass 1036–1040. steppe. Global Change Biology 7:451–466. POLLEY, H.W. 1997. Implications of rising atmospheric Received 12 January 2004 carbon dioxide concentration for rangelands. Journal Accepted 18 June 2004 of Range Management 50:562–577. POORTER, H. 1993. Interspecific variation in the growth response of plants to an elevated ambient CO2 concentration. Vegetatio 104/105:77–97. Western North American Naturalist 65(1), © 2005, pp. 87–90

INFLUENCE OF WATER SIZE AND TYPE ON BAT CAPTURES IN THE LOWER SONORAN DESERT

Michael J. Rabe1 and Steven S. Rosenstock1

ABSTRACT.—We compared bat use by mist-netting at 4 different types of wildlife water developments in southwestern Arizona during summer 2000 and 2001. Scaling our results by netting effort, we caught bats more frequently and observed higher species diversity at tinajas (modified natural rock pools) with larger open-water area compared with “guzzler” type water developments that had less open water and more obstacles to bat flight. We caught the fewest bats at guzzlers with buried concrete vault drinkers, which impede bat access and have the smallest areas of open water. Water development designs that minimize evaporative water loss by reducing the amount of open water apparently reduce use by bats in this area.

Key words: bats, water development, Sonoran desert, guzzler, tinaja, southwestern U.S., Arizona.

Recent debates over the value of wildlife water developments more readily than other water developments (Broyles 1995) have direct types (Aldridge and Rautenbach 1987, Aldridge implications for conservation of desert bats. In and Brigham 1988). The area and configura- the arid Southwest, water developments are tion of open water vary among different types often the only free water sources available to of water developments, and these are key vari- bats (Burkett and Thompson 1994, Rosenstock ables that may affect bat use. The goal of our et al. 1999). Desert bats are attracted to water study was to assess bat diversity at and use of sources in great numbers during the hottest common types of wildlife water developments in and driest part of the summer (O’Farrel and the Sonoran Desert of southwestern Arizona. Bradley 1970, Kunz and Kurta 1988), and availability of surface drinking water may limit STUDY AREA bat distributions in these arid regions (Geluso 1978). The study area is north of Yuma in south- The desert landscape of southwestern Ari- western Arizona on Kofa National Wildlife zona is extremely arid, and natural, perennial Refuge (U.S. Fish and Wildlife Service) and surface water is rare. However, a large num- Yuma Proving Ground (U.S. Army). Terrain is ber of water developments have been built to diverse, consisting of mountains, bajadas, and benefit wildlife. Construction and maintenance broad valleys dissected by ephemeral washes. of these water sources have been high priori- Vegetation cover types within the area are ties for federal and state resource manage- approximately 25% Arizona Upland Sonoran ment agencies since the 1950s (Rosenstock et Desertscrub and 75% lower Colorado River al. 1999). While most of these water sources Sonoran Desertscrub (Brown 1994). Elevations were intended to benefit mule deer (Odocoi- within the area range from 98 m on valley bot- leus hemionus) and desert bighorn (Ovis cana- toms to 1467 m on the highest mountain peaks. densis mexicana), they are heavily used by Mean yearly rainfall (1971–2000) at the near- other wildlife including bats, birds, and mam- est weather stations (Yuma Proving Ground, malian predators. Vegetation or cliffs surround- 98-m elevation, and Kofa Mine, 542-m eleva- ing a water source, or the structure of the tion) was 9.8 cm and 18.44 cm, respectively. water development itself, can impede access Mean summer temperatures (June–August) at by bats (Kalcounis and Brigham 1995, Schmidt Yuma Proving Ground and Kofa Mine weather 1999). Because bats vary in size and maneu- stations were 33.0° and 31.9°C, respectively verability, some species may use some types of (NOAA 2001).

1Arizona Game and Fish Department, Research Branch, 2221 W. Greenway Road, Phoenix, AZ 85023.

87 88 WESTERN NORTH AMERICAN NATURALIST [Volume 65

The most common types of wildlife water Our sampling sites included 2 tinajas, 2 buried developments within the study area are tina- vaults, 2 buried troughs, and 1 aboveground jas, precipitation catchments (guzzlers), and trough (Table 1). These water developments wells. Tinajas occur in canyons and rocky mon- had been in place for a minimum of 12 years tane areas. Many have been modified by the before we began our study. We visited each addition of masonry structures (dams, diversions, location and water type twice per month on a or gabions) that increase water inflow and stor- rotating schedule. Summer 2000 sampling age capacity and reduce sedimentation during occurred from 18 July to 26 August. We cap- flooding. Surface area of tinajas varies accord- tured bats at 3 locations on 4 occasions for a ing to physical dimensions, evaporation, and total of 11 nights of sampling (1 visit to Horse water inflow. Access for flying bats may be re- Tank tinaja site was aborted because of a stricted by steep cliffs surrounding the pool and severe thunderstorm). In 2001 we increased by adjacent trees and other vegetation. our effort, sampling from 29 May through 30 Guzzlers and wells are located on bajadas August. We sampled 4 locations on 6 occasions or in valley bottoms adjacent to large washes for a total of 24 nights. and have 3 distinct types of drinkers available We captured bats with mist-nets set over to bats: buried vaults, buried troughs, and open water in configurations that made it diffi- aboveground troughs. Buried vault drinkers cult for bats to drink without encountering nets. are made of fiberglass or concrete and are We erected nets each night by 2000 hours filled by passive flow from 1 or more adjacent (approximately sunset) and removed them after storage tanks. The drinker has vertical walls 2200. Each captured bat was identified to on 3 sides and a 4th side slopes outward at a species and gender and then released. We ° 45 angle, forming a ramp that leads down to marked captured bats on the top of the head the water. When full, the drinker has an open- × with a felt-tip marker to avoid multiple count- water area that measures approximately 1 m ing of recaptures during the same night. 1 m and that is nearly level with the surround- To compare the number of bats caught ing ground surface. Water level within the among different water types and net sets, we drinker drops and surface area decreases as scaled bat captures as the number of bats cap- water is depleted from the storage tank. When tured per m2 of net area per hour of effort. the drinker is nearly empty, its water surface Since surface algae often restricted the area of measures approximately 0.1 m × 1 m and is open water available to bats, we estimated approximately 0.8 m below ground level. At open water during each netting occasion by low water levels, bats must fly down to reach removing the area covered by algae from the the water surface and then pull up quickly total water area available. We calculated Pear- while exiting to avoid contacting the walls. son bivariate correlations (Sokal and Rohlf Buried troughs are made of concrete or fiber- 1995) for the open water estimates and the glass and have surface areas similar to vault 2 drinkers. However, water level and surface area number of bats captured per m of net area are regulated by a float valve and are more or per hour of effort. less constant. Aboveground troughs are made of concrete, have similar dimensions to buried RESULTS troughs, and are equipped with float valves. The Over the 2 summers we captured 427 bats water surface is located 0.5 m above ground level and 0.1 m below the top of the trough. belonging to 6 species. Western pipistrelles Surrounding vegetation usually has little influ- (Pipistrellus hesperus) were most common (187 ence on bat access to guzzlers because these individuals, 43.8% of captures), followed by waters are located away from trees or other California myotis (Myotis californicus, 105 indi- dense natural vegetation, or vegetation is re- viduals, 24.6% of captures), pallid bats (Antro- moved to prevent roots from damaging the zous pallidus, 58 individuals, 13.6% of cap- structure. tures), big brown bats (Eptesicus fuscus, 31 individuals, 7.3% of captures), Townsend’s big- METHODS eared bats (Corynorhinus townsendii, 22 individuals, 5.2% of captures), and California leaf- We captured bats at 7 wildlife water devel- nosed bats (Macrotus californicus, 18 individuals, opments in summer 2000 and summer 2001. 4.2% of captures). 2005] WATER TYPE AND BAT CAPTURES 89

TABLE 1. Results of mist-netting effort at selected wildlife water developments in southwestern Arizona during summer 2000 and 2001. No. of 2-hr Open water Bats ⋅ m–2 net Total netting surface area (m2), area ⋅ hour–1, Location Water type captures occasions mean (s)a mean (s) High Tank 7 Tinaja 207 7 6.1 (1.26) 0.41 (0.198) Scott’s Well Buried trough 79 4 1.4 (0) 0.41 (0.258) Horse Tank Tinaja 65 3 6.6 (3.43) 0.34 (0.329) Guzzler 534 Aboveground trough 44 6 1.1 (0.56) 0.15 (0.066) Guzzler 736 Buried vault 16 6 1.1 (0.24) 0.06 (0.034) Guzzler 531 Buried vault 12 6 0.9 (0.22) 0.04 (0.026) Guzzler 967 Buried trough 4 6 0.6 (0.32) 0.04 (0) aOpen water equals total water surface area minus algae cover surface area.

Bat diversity varied among water types. We despite our best efforts. Thus, our results likely found the highest diversity at tinajas, where underestimated bat use of these sites. we captured 6 species, 3 of which were not Given the concentration of bats frequently captured at other types of waters (big brown found around waters in desert habitats (O’Far- bat, Townsend’s big-eared bat, and California rel and Bradley 1970, Schmidt 1999) and the leaf-nosed bat). The fewest bat species were additional water stress that arid environments captured at buried vaults, which had only west- impose on insectivorous bats (Geluso 1978, Bas- ern pipistrelles and California myotis. We had set 1986, Happold and Happold 1988), many, if no bat captures on 3 sampling occasions each not all, species probably require free water on at buried vaults and buried troughs. occasion. Lactation imposes additional demands Scaled by netting effort, the most bats (all on insectivorous bats, and the requirement for species combined) were captured at tinajas free water is most acute for lactating females and the fewest at buried vaults (Table 1). (Kurta et al. 1989). Although no studies have ⋅ –2 Mean number of bats captured (bats m net shown that desert-dwelling bats require water, ⋅ –1 area hour ) at the 2 tinajas was 0.39 (s = non-desert bats have been shown to require 0.23, n = 10 trapping occasions). In descend- drinking water. For example, during lactation, ing order this was followed by buried troughs – drinking water may account for 23%–25% of (x = 0.23, s = 0.26, n = 8), the aboveground the daily water influx for little brown bats drinker (x– = 0.15, s = 0.07, n = 6), and buried – (Myotis lucifugus; Kunz and Kurta 1988, Kurta vaults (x = 0.05, s = 0.03, n = 12). et al. 1989). Captures were highest at wildlife waters with In our study area bat diversity and bat use the most open water and lowest in locations at wildlife waters were positively correlated with the least open water available to drinking with surface area of open water. Some bat bats. The Pearson bivariate correlation between species (e.g., western pipistrelles and Califor- open water and numbers of bats captured for nia myotis) are quite maneuverable and can all capture occasions (in bats ⋅ m–2 net area ⋅ drink from small water surfaces below ground hour–1) was 0.600 (P < 0.001, n = 36). level, such as those found at buried vaults. DISCUSSION However, other less maneuverable species (e.g., big brown and pallid bats) may be excluded Mist-net captures may provide biased esti- from these waters, and they may rely more on mates of bat use in some cases, but we believe tinajas and other larger bodies of water. they were reasonably accurate in our study. At The higher numbers and increased diver- sites with low captures and low diversity we sity of bats caught at tinaja sites were likely a observed few bats flying or attempting to drink. combination of greater water surface area and Because of the small size of the guzzler drinkers, close proximity to roosts. Distribution of bats we could usually assure that bats could not may be constrained by availability of roost sites drink without encountering nets. Because tinajas (Humphrey 1975). Our tinaja sites were located were larger and had more flight approaches, near cliffs and rocky canyons that offered roost many bats were able to drink and avoid nets sites, which may have accounted for higher use 90 WESTERN NORTH AMERICAN NATURALIST [Volume 65 than was observed at wildlife waters located HAPPOLD, D.C., AND M. HAPPOLD. 1988. Renal form and on bajadas and valley bottoms. function in relation to the ecology of bats (Chirop- tera) from Malawi, Central Africa. Journal of Zoology To be most useful to bats, water develop- 215:629–655. ments should have large surface area and be HUMPHREY, S.R. 1975. Nursery roosts and community located close to possible roost sites such as diversity of Nearctic bats. Journal of Mammalogy 56: cliffs and rock piles. Water development designs 321–346. that minimize water surface area to reduce KALCOUNIS, M.C., AND R.M.R. BRIGHAM. 1995. Intraspe- cific variation in wing loading affects habitat use by evaporation may therefore restrict use by bats. little brown bats (Myotis lucifugus). Canadian Jour- Management agencies should keep these con- nal of Zoology 73:89–95. siderations in mind when designing new wild- KUNZ, T.H., AND A. KURTA. 1988. Capture methods and life waters in the Sonoran Desert. holding devices. Pages 1–29 in T.H. Kunz, editor, Ecological and behavioral methods for the study of bats. Smithsonian Institution Press, Washington, DC. LITERATURE CITED KURTA, A., G.P. BELL, K.A. NAGY, AND T.H. KUNZ. 1989. Water balance in free-ranging little brown bats ALDRIDGE, H.D., AND R.M. BRIGHAM. 1988. Load carrying (Myotis lucifugus) during pregnancy and lactation. and maneuverability in an insectivorous bat: a test of Canadian Journal of Zoology 67:2468–2472. the 5% “rule” of radio telemetry. Journal of Mammal- NATIONAL OCEANIC AND ATMOSPHERIC ADMINISTRATION. ogy 69:379–382. 2001. Climatological data—Arizona. United States ALDRIDGE, H.D., AND I.L. RAUTENBACH. 1987. Morphology, Climatic Data Center, Asheville, NC. echolocation and resource partitioning in insectivo- O’FARREL, M.J., AND W. G. B RADLEY. 1970. Activity pat- rous bats. Journal of Animal Ecology 56:763–778. terns of bats over a desert spring. Journal of Mam- BASSET, J.E. 1986. Habitat aridity and urine concentrating malogy 51:18–26. ability of Nearctic, insectivorous bats. Comparative ROSENSTOCK, S.S., W.B. BALLARD, AND J.C. DEVOS, JR. 1999. Biochemistry and Physiology 53A:125–131. Viewpoint: benefits and impacts of wildlife water BROWN, D.E. 1994. Biotic communities of the southwest- developments. Journal of Range Management 52: ern United States and northwestern Mexico. Univer- 302–311. sity of Utah Press, Salt Lake City. 334 pp. SCHMIDT, S.L. 1999. Activity patterns of California leaf- BROYLES, B. 1995. Desert wildlife water developments: nosed and other bats at wildlife water developments questioning use in the Southwest. Wildlife Society in the Sonoran Desert. Master’s thesis, University of Bulletin 23:663–675. Arizona, Tucson. BURKETT, D.W., AND B.C. THOMPSON. 1994. Wildlife asso- SOKAL, R.R., AND F. J . R OHLF. 1995. Biometry. 3rd edition. ciation with human-altered water sources in semi- Freeman, New York. 887 pp. arid vegetation communities. Conservation Biology 8:682–690. Received 11 March 2003 GELUSO, K.N. 1978. Urine concentration ability and renal Accepted 3 September 2003 structure of insectivorous bats. Journal of Mammal- ogy 59:312–323. Western North American Naturalist 65(1), © 2005, pp. 91–102

EFFECTS OF GRAZING EXCLUSION ON RANGELAND VEGETATION AND SOILS, EAST CENTRAL IDAHO

Jeffrey J. Yeo1

ABSTRACT.—Nineteen exclosures on sagebrush steppe and shadscale rangelands, varying in age from 18 to 38 years, were sampled for plant species richness, plant composition, indicators of soil erosion, ground cover, vegetative cover, and herb–low shrub layer screening cover. Features within the exclosures were compared with adjacent sites of the same size that were open to grazing by livestock and wildlife. Species richness typically was slightly greater inside exclosures than in adjacent grazed sites (median = 2 more species inside exclosures), but the difference was not significant (P = 0.16). Similarity of plant community composition between exclosures and adjacent grazed sites ranged from 45% to 82%. Evidences of soil movement, soil pedestals, and soil flow patterns were all more pronounced outside exclosures than inside (P ≤ 0.02), even though many sites were on flat to mild slopes (median slope = 12%). Meta-analysis of the 19 exclosure sites indicated that grazing exclusion resulted in less bare ground cover compared with adjacent grazed sites (P ≤ 0.05). The effect of grazing exclusion on visible soil surface cryptogams was significant (P ≤ 0.05), with generally greater cover inside exclosures. Cryptogam cover differences between grazed sites and exclosures tended to increase with the number of years of grazing exclusion (r = 0.64, P = 0.046). Pseudoroegneria spicata, a principal livestock forage, averaged greater basal cover inside exclosures than outside on 4 of 10 sites where it occurred, although no exclosure sites had greater P. spicata cover on grazed sites. Meta-analysis of the 10 sites indicated that grazing exclusion resulted in greater P. spicata cover compared with adjacent grazed areas (P ≤ 0.05). Poa secunda, a short-growing grass that initiates growth early in the spring and is not important livestock forage, averaged greater basal cover outside exclosures on 5 of 15 sites where it occurred. Meta-analysis of the 15 sites indicated a significant treatment effect (P ≤ 0.05), with greater Poa secunda basal cover outside exclosures. Grazing exclusion resulted in greater screening cover in the herb–low shrub layer (0–0.5 m height; P ≤ 0.05). These results indicate that despite improved livestock grazing management over the past half century, livestock grazing still can limit the potential of native plant communities in sagebrush steppe ecosystems, and that the health of semiarid ecosystems can improve with livestock exclusion in the absence of other disturbances. A few exclosure sites were similar for the measured parameters, suggesting that these sites were ecologically stable and that exclusion of livestock grazing was not sufficient to move succession toward more pristine conditions, at least within the time periods studied. Managed disturbance such as fire or mechanical brush treatments may be necessary to restore herb productivity on these ecologically stable sites.

Key words: grazing exclosures, long-term vegetation change, erosion, vegetative cover, screening cover, cryptogams, wildlife.

An issue central to rangeland management stock may not result in rangeland improvement, concerns the ability of plant communities to at least in arid and semiarid ecosystems such respond to changing livestock management as sagebrush steppe (West et al. 1984, Bork et practices. Range managers recognize the his- al. 1998, West 2000). toric impacts of overgrazing prior to passage of Traditional models of plant community suc- the Taylor Grazing Act in 1934 but assume cession (Clements 1916, Dyksterhuis 1949, that recent practices of grazing management Huschle and Hironaka 1980) postulate gener- (e.g., rest-rotation, deferred rotation, short-dura- ally linear pathways of succession to a pre- tion grazing) are adequate to protect range- dictable climax state following disturbances land resources given appropriate stocking rates such as fire and grazing. More recent commu- (Laycock 1994). Removal of livestock from west- nity succession models suggest that vegetation ern arid and semiarid rangelands has been can exist in multiple quasi-stable states depend- advocated because of widespread evidence of ing on the history of the site, and that transi- overgrazing and the impacts on biodiversity tion between states requires some biotic or (Fleischner 1994, Noss 1994, Donahue 1999). abiotic force to move the community beyond Yet, reduction in numbers or removal of live- a threshold (Noy-Meir 1975, Hanley 1979,

1The Nature Conservancy, 116 First Avenue North, Hailey, ID 83333.

91 92 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Westoby et al. 1989, Tausch et al. 1993). The observed differences were a result of continued first 10–15 years of heavy grazing by livestock grazing or its absence. My expectation was may have the greatest impact on vegetation that sites inside exclosures would have greater (Hull 1976). Once vegetation degrades to some cover of palatable grasses, perennial forbs, threshold, cessation of grazing may not halt cryptogams, and screening cover, and less bare continued decline or at least may not allow the ground than adjacent sites outside exclosures. community to improve (Whitford 1995). Cycles I also expected that evidence of soil erosion of drought and soil changes may preclude re- would be less pronounced within exclosures covery of plant communities to pregrazing con- than outside. ditions. The availability of native plant seed, availability of nutrients, and invasion of exotic STUDY AREA plants may alter the succession pathways available. Reduction of fine fuels because of live- The exclosures were situated on high-ele- stock grazing and fire-suppression policies has vation (1585–2315 m), federally managed pub- resulted in a change in the frequency and in- lic rangelands in east central Idaho (Table 1). tensity of natural fire, which historically played Vegetation varied from lower-elevation xeric an important role in rangeland dynamics. communities of Atriplex confertifolia and Atri- Ecological advantages reported for livestock plex spinosa commonly growing on alkaline grazing include enhanced seeding/germination soils, or Artemisia tridentata spp. wyomingen- by trampling seed into the ground, fertilization sis on well-drained soils, to more mesic com- from feces and urine, and plant growth stimu- munities of Artemisia tripartita or Artemisia lation from grazing (see reviews by Briske and tridentata spp. vaseyana sometimes mixed Richards 1994, Pieper 1994). However, over- with Purshia tridentata bordering Pseudotsuga grazing can result in the reduction or extinc- menziesii at higher elevations. Slope gradients tion of forage preferred by livestock, degrada- ranged from nearly flat (1%) to 20%, and aspect tion of biodiversity, introduction or dominance of the exclosure sites included most exposures. of annuals and exotic plants, reduction or elim- Exclosure sizes varied from 0.04 ha to 0.8 ha, ination of cryptogamic soil crusts that can and the time during which grazing was ex- reduce nitrogen available for plant growth, cluded spanned 18 to 38 years. increased soil erosion, and soil compaction with Climate at the exclosure sites is semiarid impeded water infiltration (Fleischner 1994, and cool with average annual precipitation Belsky and Blumenthal 1997). Factors other ranging from 128 mm to 559 mm (based on than livestock grazing can also affect vegeta- soil descriptions, Natural Resources Conser- tion dynamics. For example, cycles of drought vation Service 2002a). Average summer tem- can have pronounced effects on the composi- perature at Challis is 20°C, and winters aver- tion and structure of sagebrush steppe vegeta- age –8°C. Climate averages can be deceiving tion (Anderson and Inouye 2001). for this area because of the substantial vari- Grazing exclosures have been constructed ability due to rugged topography and localized in the Bureau of Land Management (BLM) storms. The frost-free period may range from Challis Resource Area and surroundings since 60 to 100 days (Bureau of Land Management the 1st half of the 20th century (Idaho Depart- 1998). A consistent annual weather record for ment of Fish and Game records, Salmon, ID). Challis began about 1932 (Fig. 1). Drought Yet, few of these exclosures apparently were was apparent in the 1930s and broke in 1940. sampled when constructed or the records have Drought returned in the late 1940s and con- been lost, and even fewer have been evaluated tinued through about 1963. The late 1980s in the intervening years. This research addresses through the early 1990s also experienced the question, What are the effects on vegeta- drought. In contrast, the years 1993, 1995, and tion and soils of exclusion of livestock grazing 1998 were exceptionally wet. after a long history of grazing (and often over- Precipitation in 1999, when the exclosures grazing)? I assumed in this study that sites were sampled, was generally less than the long- were in similar condition inside and outside term average. This was particularly true during the exclosures at the time of construction, and mid- to late summer when sampling occurred. that paired sites experienced similar environ- Plant growth in 1999 was initially slowed by a mental conditions since fencing so that any cool June, followed by rapid growth in July that 2004] LONG-TERM EFFECTS OF GRAZING EXCLUSION 93

TABLE 1. Exclosure site characteristics and topography. Predicted range of annual precipitation is based on soil characteristics (Natural Resources Conservation Service 2002a). Plant association: ACHY = Achnatherum hymenoides, ARAR = Artemisia arbuscula, ARFR = Artemisia frigida, ARTR4 = Artemisia tripartita, ARTRV = Artemisia tridentata vaseyana, ARTRW = Artemisia tridentata, ATCO = Atriplex confertifolia, FEID = Festuca idahoensis, HECO = Hes- perostipa comata, POSE = Poa secunda, PSSP = Pseudoroegneria spicata, SPAR = Sphaeromeria argentea, and SPCR = Sporobolus cryptandrus. Community similarity index is described in Methods. Predicted precipitation Elev. Slope Size Years Similarity Exclosure (mm) Plant association (m) Aspect (%) (ha) enclosed index (%) Eagle Rock 128–205 ATCO/ACHY-HECO 1585 NE 8 0.81 22 70 Meadow Creek 178–254 ATCO/SPCR 1804 N 3 0.4 38 80 Leaton Gulch 179–256 ATCO/SPCR 1670 W 8 0.4 38 82 Antelope Flats 203–279 SPAR-ARFR/POSE-HECO 1938 N 1 0.04 28 75 McGowan Creek 203–305 ARTRW/HECO 1987 W 12 0.08 18 73 Centennial Flats 203–305 ARTRW/PSSP 1938 E 16 0.36 18 78 Sage Creek No. 2 203–305 ARTRW/PSSP 1951 NE 1 0.08 18 60 Sage Creek No. 1 203–305 ARTRW/PSSP 1975 NE 1 0.12 18 70 East Fork Fan 203–305 ARTRW/PSSP 1743 NE 12 0.36 18 82 Boneyard Gulch 203–305 ARTRW/PSSP 1792 E 12 0.45 29 45 Bradshaw Basin 203–305 ARTRW/PSSP 2219 NW 12 0.04 28 69 Jeff Flats No. 1 279–330 ARTRW/PSSP 1853 NE 12 0.04 28 61 Donkey Hills 203–305 ARAR/PSSP 2146 NE 2 0.4 38 78 Jeff Flats No. 2 305–406 ARTR4/FEID 1926 N 12 0.32 19 80 Broken Wagon 330–406 ARTR4/FEID 2085 N 12 0.08 18 54 Second Spring 305–406 ARTR4/FEID 2292 S 12 0.81 21 77 Martin Creek 330–406 ARTRV/FEID-PSSP 2012 SE 20 0.4 31 72 Third Spring 406–559 ARTRV/FEID 2134 S 20 0.36 22 75 Horse Heaven 406–559 ARTRV/FEID 2316 N 5 0.45 29 54

Fig. 1. Long-term annual precipitation records for Challis, Idaho. Median distance to the 19 exclosures from Challis was 23.2 km (range = 8.4–52.3 km). 94 WESTERN NORTH AMERICAN NATURALIST [Volume 64 terminated in early August with desiccation, The point-interception frame was leveled with particularly among forbs. a bubble level to maintain a consistent vertical Livestock grazing of the region began in projection on the ground. All plant species the 1860s, with overgrazing noted by the 1880s beneath each point were recorded based on and probably continuing for 40–50 years (Shoup “hits” of live vegetation of forbs and shrubs and 1935). Domestic sheep numbers began increas- basal areas of graminoids. Points were visually ing in the late 1880s, reaching >50,000 sheep projected through the overstory so that a sin- grazing Bureau of Land Management lands in gle point could include more than 1 species the Challis area by the 1920s (Bureau of Land but not ground cover. Points intercepting bare Management records, Salmon, ID). After the ground, litter, or cryptogams were recorded only 1950s cattle became the principal livestock if no vascular plants occurred above the point. grazing the area. In the 1970s planned grazing At 3 exclosure sites I used a 20 × 50-cm systems were implemented. plot frame to measure cover. I did this either because vehicle access was >1.6 km from the METHODS exclosure and the combination of equipment was too cumbersome to transport to the sites, I sampled upland grazing exclosures on or because shrubs were too tall and dense to public lands in which at least half the area use the point-interception frame effectively. inside the exclosure appeared undisturbed Cover was estimated using Daubenmire’s (1959) since its construction. Visually disturbed areas cover classes (in percentage): 0, 1–5, 6–25, either inside or outside exclosures (i.e., soil 26–50, 51–75, 76–95, and 96–100. disturbance from range improvements or ex- Screening cover is important to rangeland perimental seeding) were not sampled. I sam- wildlife such as nesting Greater Sage-Grouse pled inside each exclosure and at a nearby (Centrocercus urophasianus) and pronghorn (<30 m distant) site outside the exclosure ex- (Antilocapra americana) fawns, both for secu- hibiting similar size, vegetation, aspect, slope, rity from predators and for thermal cover (Aut- and elevation as the exclosure. Therefore, sam- enreith 1978, Connelly et al. 2000). Screening pling site selection emphasized similarities cover was estimated at each sampling point between the exclosure and outside. using a 1.5-m-tall cover pole (Griffith and Sample points, 5–15 m apart (depending on Youtie 1988). The pole was divided into three the size of the exclosure, 10–25 samples were 0.5-m sections, with each section divided into collected per treatment), were systematically five 10-cm segments. The 3 sections were des- located on a randomly positioned grid to ignated as herb–low shrub layer (0–0.5 m achieve good dispersion of points throughout height), medium shrub layer (0.5–1.0 m height), the exclosure and adjacent grazed site. A 5-m and tall shrub layer (1.0–1.5 m height). The buffer zone inside and outside the exclosure number of 10-cm segments at least 50% cov- fence was not sampled to avoid possible effects ered by live vegetation was recorded separately of the fence (e.g., livestock trailing and tram- for each of the three 0.5-m sections. Four read- pling, fertilization from birds perched on the ings at a distance of 5 m from each of the car- fence or other animals concentrating along the dinal directions were taken at each sampling fence). At each sampling point I measured point. The 2 lower layers were read from a horizontal vegetation cover, ground cover, and kneeling position to reduce parallax error. screening cover (vertical vegetation cover). Soil Analyses of screening cover comparisons are surface condition was evaluated for each treat- presented only for the herb–low shrub layer ment as a whole. because many fewer sites contained screening I measured canopy cover of shrubs and cover in the 2 taller layers. forbs, basal cover of graminoids, and ground I evaluated soil erosion using 7 indicators of cover of bare ground, litter, and cryptogams soil surface condition: evidence of soil move- using a 50 × 100-cm point-interception frame ment, surface litter, surface rock, soil pedestals, (Floyd and Anderson 1982) placed at each evidence of flow patterns, rills, and gullies sampling point. Each point-interception frame (Bureau of Land Management 1973). Soil sur- contained 36 intersection points (created by 2 face condition factors were evaluated for each superimposed grids, 15 cm apart, of 10 × 10-cm exclosure site sampled (i.e., separately for the squares of string) at which cover was recorded. entire area within the exclosure and for the 2004] LONG-TERM EFFECTS OF GRAZING EXCLUSION 95 sampled area outside the exclosure). Each ero- with differences between treatments ranging sion indicator was evaluated on an ordinal from 0 to 9 species. Twelve of 19 sites had scale with 1 indicating little or no evidence of greater species richness within exclosures than erosion and 5 indicating severe erosion. outside (median difference = 2 more species Species richness was compared between within exclosures than on adjacent grazed sites), the exclosure and the adjacent grazed area. but this difference was not significant (P = Sorensen’s community coefficient (similarity 0.16). Community similarity ranged from 45% index), weighted by % cover (Barbour et al. to 82% (median = 73%), with Boneyard Gulch 1980), was calculated for each exclosure site. having the lowest community similarity (45%) Depending on the distribution of the data for of 19 exclosure sites studied (Table 1). The each parameter estimated, I used an unpaired number of years of grazing exclusion was not t test or its nonparametric equivalent (Mann- related either to differences in species richness Whitney rank sum test) to compare the effects between treatments or to community similar- of grazing and grazing exclusion. Correlations ity (r ≤ 0.13, P ≥ 0.60). between the duration of grazing exclusion and Slope steepness at the exclosure sites ranged measured parameters were analyzed using from 1% to 20% (median = 12%; Table 1). Pearson’s correlation index. Syntheses across Exclusion of livestock grazing resulted in a exclosure sites of treatment effects for each of consistent pattern of improved soil surface the cover parameters estimated were analyzed conditions compared with areas open to graz- using meta-analytic methods (Rosenberg et al. ing. Evidence of soil movement, soil pedestals, 2000). The treatment effect size (with the area and soil flow patterns were all more pronounced open to grazing representing the control and outside exclosures than inside (n = 19; P ≤ the area excluding grazing representing the 0.02). experimental treatment) for each exclosure Cover of bare ground ranged from 3% to site was standardized as the natural log of the 51%, with differences between treatments response ratio (the ratio of the mean estimate ranging from 0% to 39%. The amount of bare within the exclosure to that of the area open to ground was greater outside exclosures at 9 of grazing). This measure of effect size estimates 19 sites (P ≤ 0.017), and there was no apparent the change resulting from grazing exclusion. A relationship between the duration of grazing random-effects model was used because the exclusion and bare ground differences between vegetation communities differed among exclo- grazed and exclosure treatments (Fig. 2). Al- sures such that the response to grazing exclu- though the correlation between slope steep- sion would not be expected to be consistent ness and the difference between excluded and among exclosure sites. Confidence intervals grazed sites was not significant (P = 0.07), about the mean effect size were calculated using there was a tendency for flatter sites to have bias-corrected bootstrapping (Rosenberg et al. greater differences between treatments than 2000). Indicators of soil erosion were com- steeper sites (r = 0.42). Meta-analysis of the pared across all exclosures using a Wilcoxon 19 exclosure sites indicated that the amount of signed rank test. Significance level was set at bare ground exposed was reduced when graz- P ≤ 0.05. Plant names follow the National ing was excluded (P ≤ 0.05). PLANTS database (Natural Resources Con- Differences between treatments for ground servation Service 2002b). cover of litter were variable but were significant at 5 exclosure sites. Two sites had greater RESULTS litter cover outside the exclosures and 3 had greater litter cover inside. There was no corre- I encountered 20 species of graminoids, 60 lation between litter cover differences between species of forbs, 20 species of shrubs, and 1 treatments and the duration of grazing exclu- tree species (Pseudotsuga menziesii) at the 19 sion (r = 0.02, P = 0.92). In addition, there exclosure sites. Nonnative plants were sparse was no consistent effect of grazing exclusion (only 6 species found) and, except for Agropy- when all exclosure sites were analyzed together. ron cristatum seeded at 3 sites, had only trace Soil cryptogams, found at 10 sites, had sig- amounts of cover at any exclosure site. nificantly greater cover inside exclosures at 6 Species richness observed at each of the sites compared with adjacent grazed areas (P ≤ exclosure sites ranged from 8 to 31 species, 0.015; Fig. 3). Cover ranged from 1% to 36%, 96 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Fig. 2. Comparison between exclosure sites and adjacent grazed sites for bare ground cover, arranged by number of years of grazing exclusion within the exclosure. N. S. groups those sites where the difference between exclosure and grazed sites was not statistically significant; P < 0.05 groups those sites where treatment differences were statistically significant. Error bars denote 95% confidence intervals.

Fig. 3. Comparison between exclosure sites and adjacent grazed sites for cover of crytogams, arranged by number of years of grazing exclusion within the exclosure. N. S. groups those sites where the difference between exclosure and grazed sites was not statistically significant; P < 0.05 groups those sites where treatment differences were statistically significant. Error bars denote 95% confidence intervals. 2004] LONG-TERM EFFECTS OF GRAZING EXCLUSION 97

Fig. 4. Comparison between exclosure sites and adjacent grazed sites for cover of Pseudoroegneria spicata, arranged by the number of years of grazing exclusion within the exclosure. N. S. groups those sites where the difference between the exclosure and grazed sites was not statistically significant; P < 0.05 groups those sites where treatment differences were statistically significant. Error bars denote 95% confidence intervals. with cover differences between treatments cover differences between treatments (r = ranging from 1% to 29%. Cryptogam cover dif- –0.01, P = 0.96). Meta-analysis of all 10 sites ferences between exclosures and adjacent showed a significant treatment effect with grazed areas tended to increase with more increased basal cover with grazing exclusion years of grazing exclusion (r = 0.64, P = (P ≤ 0.05). 0.046). Meta-analysis of all 10 sites indicated a Poa secunda, an early growing, small, fine- significant treatment effect with increased leaved grass of limited forage value, averaged cryptogam cover corresponding to exclusion of greater basal cover outside the exclosure at 5 grazing (P ≤ 0.05). of 15 sites at which it occurred (P ≤ 0.05; Fig. Graminoid basal cover ranged from 5% to 5). Poa secunda had greater basal cover within 26%. The 2 most consistently encountered the Sage Creek No. 2 exclosure than outside grasses, and generally with the largest cover, (P = 0.031). There was not an apparent rela- were Pseudoroegneria spicata and Poa secunda. tionship between duration of exclusion and Other grasses common (≥5% basal cover) on amount of Poa secunda basal cover differences some sites included Festuca idahoensis (2 sites), between grazed and excluded sites (r = –0.11, Sporobolus cryptandrus (1 site), and Agropyron P = 0.64). Poa secunda basal cover decreased cristatum (1 site). with grazing exclusion (P ≤ 0.05), based on Pseudoroegneria spicata, probably the most meta-analysis of the 15 sites. important livestock forage species in the region Average cover of perennial forbs ranged (Yeo 1981), occurred at 10 exclosure sites. Basal from trace amounts to 24% at the 19 exclosure cover of P. spicata was significantly greater sites. Only at 1 site (Boneyard Gulch) did cover within the exclosures at 4 sites (P ≤ 0.037), of perennial forbs differ significantly between with cover differences ranging from 6% to treatments, with about a threefold greater forb 10% (Fig. 4). No sites had significantly greater cover outside the exclosure (P ≤ 0.001). Phlox cover of P. spicata outside the exclosures. spp. and Lomatium spp. were the principal There was no apparent relationship between contributors to greater cover outside the Bone- the duration of grazing exclusion and basal yard Gulch exclosure. There was no apparent 98 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Fig. 5. Comparison between exclosure sites and adjacent grazed sites for cover of Poa secunda, arranged by number of years of grazing exclusion within the exclosure. N. S. groups those sites where the difference between exclosure and grazed sites was not statistically significant; P < 0.05 groups those sites where treatment differences were statistically significant. Error bars denote 95% confidence intervals.

Fig. 6. Comparison between exclosure sites and adjacent grazed sites for screening in the herb–low shrub layer, arranged by number of years of grazing exclusion within the exclosure. N. S. groups those sites where the difference between exclosure and grazed sites was not statistically significant; P < 0.05 groups those sites where treatment differences were statistically significant. Error bars denote 95% confidence intervals. 2004] LONG-TERM EFFECTS OF GRAZING EXCLUSION 99 treatment effect on perennial forb cover ana- ity and site history are integral to understand- lyzed across all exclosure sites by meta-analysis ing community responses to livestock grazing (P ≥ 0.05). or its exclusion. Screening cover in the herb–low shrub layer Despite changes in grazing management, was significantly greater inside exclosures at continued livestock grazing may hinder the 10 of 19 sites (P ≤ 0.05; Fig. 6). Jeff Flats No. 1 rate or magnitude of vegetation response com- was the only exclosure site where screening pared to livestock exclusion. For example, the cover averaged greater outside the exclosure Idaho National Engineering and Environmen- than inside. The relationship between years of tal Laboratory (INEEL), a 2315-km2 reserve grazing exclusion and differences between about 65 km south of the study area, has been treatments of herb–low shrub layer screening protected from livestock grazing for 45 years. cover was not significant (r = 0.10, P = 0.70). Shrub cover increased (during the first 25 years Meta-analysis indicated that grazing exclusion after livestock exclusion), followed by increased resulted in increased screening cover in the grass abundance, increased average species herb–low shrub layer (P ≤ 0.05). richness, and increased vegetation hetero- Artemisia tridentata occurred at 14 of 19 geneity over the next 20 years (Anderson and sites. Canopy cover was similar between exclo- Inouye 2001). Vegetation outside the INEEL, sures and adjacent areas open to grazing. Meta- which was available for livestock grazing, analysis indicated no effect on A. tridentata showed a similar although less pronounced cover due to grazing exclusion. The lack of pattern of vegetation change. treatment effect remained true when the 2 West et al. (1984), studying Artemisia tri- subspecies of A. tridentata encountered, A. t. dentata ssp. tridentata shrub steppe (annual ssp. wyomingensis and vaseyana, were ana- precipitation = 280–347 mm), reported that lyzed separately. exclusion of livestock would not necessarily improve native perennial grass biomass. They DISCUSSION stated that disturbance was “mandatory” to return these semiarid communities to domi- Each exclosure site represented a case study nance by perennial grasses. However, in the of the effects of grazing exclusion. Exclosures absence of obvious disturbance, many of the varied in size (which affected sample sizes), plant communities inside exclosures reported vegetation types (which affected types of treat- here showed improvements based on indica- ment responses possible), and number of years tors of rangeland health (National Research of exclusion (which could limit the magnitude Council 1994, Pellant et al. 2000). Valone et al. of potential treatment responses). Because of (2001), working on arid grassland sites in Ari- these differences, variability of ecological effects zona (annual precipitation = 222–376 mm), and effect sizes should be expected, and meta- suggested that there might be time lags of 20 analyses are recommended to identify treat- years or more before perennial grasses respond ment effects for these situations of variable to livestock grazing removal. For the exclosures results commonly encountered in field studies reported here, some exclosures in place for (Johnson 2002). >30 years showed no difference for principal Many differences were evident at most grass basal cover, while some exclosures <20 Challis exclosure sites that can be attributed to years old had greater grass cover inside exclo- the exclusion of livestock grazing. These sures. The lack of correlation between period include reduction of bare ground cover and of grazing exclusion and vegetation response reduction of evidence of soil erosion, increased suggests that site history and site potential principal forage cover, increased cover of cryp- may be important factors determining rates of togams, and increased screening cover. These vegetation recovery. differences are consistent with those reported Pseudoroegneria spicata is a principal live- in reviews of the ecological effects of livestock stock forage species on these rangelands com- grazing (e.g., Fleischner 1994, Belsky and Blu- prising as much as 80% of cattle forage use in menthal 1997). At least some of these differ- Artemisia tridentata wyomingensis communi- ences were evident at most, but not all, exclosure ties (Yeo 1981). Results presented here sug- sites, indicating that environmental complex- gest that cattle preference for P. spicata was 100 WESTERN NORTH AMERICAN NATURALIST [Volume 64 suppressing its recovery, a situation in which for nests and young fawns, as well as forage for Poa secunda was apparently capitalizing. Poa pronghorn and Greater Sage-Grouse and pro- secunda provides less forage and lower nutri- ductive communities of insects as food for tive quality than Pseudoroegneria spicata Greater Sage-Grouse and songbirds. (Willms and McLean 1978) and, because of its These differences indicate that despite im- low growth form, affords less cover for wildlife. proved livestock grazing management in the Independent evidence for the Challis area past half century, continued livestock grazing also suggests that livestock grazing practices has limited the potential of some of these have not allowed P. spicata to recover from native rangeland communities, or at least slowed past overgrazing. Range trend monitoring tran- their recovery relative to grazing exclusion. sects were implemented in the BLM’s Challis Greater productivity (particularly of the herba- Resource Area in the 1950s. Reexamination of ceous understory), greater species richness, those that could be found in 2001 (n = 11 tran- greater extent of cryptogamic soil crusts with sect clusters) suggested that between the 1950s less bare ground and less evidence of soil ero- and 1970, vegetation cover remained sparse sion all are signs of better rangeland health (Yeo 2001). Even though average vegetation (National Research Council 1994, Pellant et al. cover was 26% greater in 2001 than in the 2000) and benefit not only wildlife with in- 1950s and 1970, P. spicata, typically confined creased forage and cover but also livestock to protection within shrub canopies in the with increased forage availability. These dif- 1950s, showed little change in cover by 2001 ferences were not evident at all sites, and and was still confined to shrub canopies. How- range managers should be cautious in their ever, periods of rest between livestock grazing expectations of ecosystem responses to changes may result in improvements of P. spicata cover in grazing management. Plant species respond in the Challis area, at least within the drier individually in a nonlinear fashion and at dif- vegetation types. Within 10 years of imple- ferent rates to disturbance or ecosystem stres- mentation of a rest-rotation grazing system sors, such as drought (Anderson and Inouye and with rest periods of ≥3 years, P. spicata 2001), so land managers should expect unpre- cover increased 42% in Artemisia tridentata dictable variability to management actions at a ssp. wyomingensis/Pseudoroegneria spicata local level. The results reported here, coupled communities (Yeo et al. 1990). with other evidence reported in the literature, Cryptogams decline under livestock forag- clearly indicate the need for monitoring to ing and trampling (Rice and Westoby 1978, guide land management and the worth of con- Anderson et al. 1982). Anderson et al. (1982) trastive experiments such as exclosures to act reported that cryptogams could recover within as controls or references as part of that moni- 20 years if protected from livestock grazing toring (Ford 2000). and trampling. This study’s results indicate that cryptogams are impacted by livestock graz- ACKNOWLEDGMENTS ing and that recovery is time related. Cryp- The research was funded through a Bureau togams are an important component of xeric of Land Management Challenge–Cost Share landscapes of the West, contribute to soil sta- agreement with the Challis Field Office. J. bilization and soil moisture retention, influ- Gregson, W. Diage, W. Osborne, and E. ence nitrogen cycling, and may aid seedling Williams of the BLM Challis Field Office pro- establishment (West 1990, Belnap 2000). vided help throughout the study. K. Rodgers, Grazing exclusion resulted in greater screen- R. Camper, and K. Gallogly, Salmon-Challis ing cover in the herb–low shrub layer that has National Forest, found records and soils infor- implications for wildlife. Guidelines for habitat mation for the 2 exclosures on U.S. National management for Greater Sage-Grouse (Con- Forest Service lands. M. Olson, Natural Re- nelly et al. 2000), pronghorn (Autenreith 1978, sources Conservation Service, Challis Office, Allen et al. 1984), and sagebrush-dependent and M. Scott, Idaho Department of Fish and songbirds (Paige and Ritter 1999) recommend Game, Salmon Office, gave access to their ex- mosaics of native sagebrush communities with closure records. I thank J. Gregson, W. Diage, productive herbaceous understories. These and 2 anonymous reviewers for many helpful communities afford thermal and security cover comments on earlier drafts. 2004] LONG-TERM EFFECTS OF GRAZING EXCLUSION 101

LITERATURE CITED FLOYD, D.A., AND J.E. ANDERSON. 1982. A new point interception frame for estimating cover of vegetation. ALLEN, A.W., J.G. COOK, AND M.J. ARMBRUSTER. 1984. Vegetatio 50:185–186. Habitat suitability index models: pronghorn. U.S. Fish FORD, E.D. 2000. Scientific method for ecological research. and Wildlife Service, FWS/0BS-82/10.65. Cambridge University Press, Cambridge, UK. ANDERSON, D.C., K.T. HARPER, AND S.R. RUSHFORTH. 1982. GRIFFITH, D.B., AND B.A. YOUTIE. 1988. Two devices for Recovery of cryptogamic soil crusts from grazing on estimating foliage density and deer hiding cover. Wild- Utah winter ranges. Journal of Range Management life Society Bulletin 16:206–210. 35:355–359. HANLEY, T.A. 1979. Application of an herbivore-plant model ANDERSON, J.E., AND R.S. INOUYE. 2001. Landscape-scale to rest-rotation grazing management on shrub-steppe changes in plant species abundance and biodiversity vegetation. Journal of Range Management 32:115–118. of a sagebrush steppe over 45 years. Ecological Mono- HULL, A.C. 1976. Rangeland use and management in the graphs 71:531–556. Mormon West. In: Symposium on Agriculture, Food and Man—Century of Progress. Brigham Young Uni- AUTENRIETH, R., EDITOR. 1978. Guidelines for the management of pronghorn antelope. Pages 473–526 in versity, Provo, UT. Proceedings of the Eighth Biennial Pronghorn Ante- HUSCHLE, G., AND M. HIRONAKA. 1980. Classification and lope Workshop, 2–4 May 1978, Jasper, Alberta. ordination of seral plant communities. Journal of Range Management 33:179–182. BARBOUR, M.G., J.H. BURK, AND W. D. P ITTS. 1980. Terres- JOHNSON, D.H. 2002. The importance of replication in trial plant ecology. Benjamin/Cummings Publishing wildlife research. Journal of Wildlife Management Company, Inc., Menlo Park, CA. 66:919–932. BELNAP, J. 2000. Structure and function of biological soil LAYCOCK, W.A. 1994. Implications of grazing vs. no grazing crusts. Pages 55–62 in P. G. Entwistle, A.M. DeBolt, on today’s rangelands. Pages 250–280 in M. Vavra, J.H. Kaltenecker, and K. Steenhof, compilers, Pro- W.A. Laycock, and R.D. Pieper, editors, Ecological ceedings: Sagebrush Steppe Ecosystems Sympo- implications of livestock herbivory in the West. Soci- sium. Bureau of Land Management Publication ety of Range Management, Denver, CO. BLM/ID/PT-001001+1150, Boise, ID. NATIONAL RESEARCH COUNCIL. 1994. Rangeland health: BELSKY, A.J., AND D.M. BLUMENTHAL. 1997. Effects of live- new methods to classify, inventory, and monitor range- stock grazing on stand dynamics and soils in upland lands. National Academy Press, Washington, DC. forests of the interior west. Conservation Biology 11: NATURAL RESOURCES CONSERVATION SERVICE. 2002a. 315–327. Custer-Lemhi Counties, Idaho, soil survey. USDA BORK, E.W., N.E. WEST, AND J.W. WALKER. 1998. Three- Natural Resources Conservation Service, Boise, ID. tip sagebrush steppe responses to long-term seasonal ______. 2002b. The PLANTS database. Version 3.5 (http:// sheep grazing. Journal of Range Management 51: plants.usda.gov). U.S. Department of Agriculture 293–300. National Plant Data Center, Baton Rouge, LA. BRISKE, D.D., AND J.H. RICHARDS. 1994. Physiological NOSS, R.F. 1994. Cows and conservation biology. Conser- responses of individual plants to grazing: current vation Biology 8:613–616. status and ecological significance. Pages 147–176 in NOY-MEIR, I. 1975. Stability of grazing systems: an appli- M. Vavra, W.A. Laycock, and R.D. Pieper, editors, cation of predator-prey graphs. Journal of Ecology Ecological implications of livestock herbivory in the 63:459–481. west. Society of Range Management, Denver, CO. PAIGE, C., AND S.A. RITTER. 1999. Birds in a sagebrush sea: BUREAU OF LAND MANAGEMENT. 1973. Determination of managing sagebrush habitats for bird communities. erosion condition class, Form 7310-12. U.S. Depart- Partners in Flight Western Working Group, Boise, ID. ment of the Interior, Washington, DC. PELLANT, M., P. SHAVER, D.A. PYKE, AND J.E. HERRICK. 2000. ______. 1998. Challis Resource Area. Proposed Resource Interpreting indicators of rangeland health. Version Management Plan and Environmental Impact State- 3. USDI Bureau of Land Management Technical ment. U.S. Department of Interior Bureau of Land Reference 1734-6, Publication BLM/WO/ST-00/001 Management, Salmon, ID. +1734. CLEMENTS, F.E. 1916. Plant succession: an analysis of the PIEPER, R.D. 1994. Ecological implications of livestock development of vegetation. Carnegie Institution of grazing. Pages 177–211 in M. Vavra, W.A. Laycock, Washington, Publication 242, Washington, DC. and R.D. Pieper, editors, Ecological implications of CONNELLY, J.W., M.A. SCHROEDER, A.R. SANDS, AND C.E. livestock herbivory in the west. Society of Range BRAUN. 2000. Guidelines to manage Sage Grouse Management, Denver, CO. populations and their habitats. Wildlife Society Bul- RICE, B., AND M. WESTOBY. 1978. Vegetative responses of letin 28:967–985. some Great Basin shrub communities protected DAUBENMIRE, R.F.1959. Canopy coverage method of veg- against jackrabbits or domestic stock. Journal of etation analysis. Northwest Science 33:43–64. Range Management 31:28–34. DONAHUE, D.L. 1999. The western range revisited: re- ROSENBERG, M.S., D.C. ADAMS, AND J. BUREVITCH. 2000. moving livestock from public lands to conserve native MetaWin: statistical software for meta-analysis. Ver- biodiversity. University of Oklahoma Press, Norman. sion 2. Sinauer Associates, Sunderland, MA. DYKSTERHUIS, E.J. 1949. Condition and management of SHOUP, G.E. 1935. Ranges of the Salmon River. Unpub- range land based on quantitative ecology. Journal of lished manuscript, Bureau of Land Management Range Management 2:104–115. records, Salmon, ID. FLEISCHNER, T.L. 1994. Ecological costs of livestock graz- TAUSCH, R.J., P.E. WIGAND, AND J.W. BURKHARDT. 1993. ing in western North America. Conservation Biology Plant community thresholds, multiple steady states, 8:629–644. and multiple successional pathways: legacy of the 102 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Quaternary? Journal of Range Management 46: WHITFORD, W.G. 1995. Desertification: implications and 439–447. limitations of the ecosystem health metaphor. In: VALONE, T.J., M. MEYER, J.H. BROWN, AND R.M. CHEW. D.J. Rapport, C.L. Gaudet, and P. Calow, editors, 2001. Timescale of perennial grass recovery in deser- Evaluating and monitoring the health of large-scale tified arid grasslands following livestock removal. ecosystems. NATO ASI Series, Volume 28. Springer- Conservation Biology 16:995–1002. Verlag, Berlin. WEST, N.E. 1990. Structure and function of soil microphytic WILLMS, W., AND A. MCLEAN. 1978. Spring forage selec- crusts in wildland ecosystems of arid and semiarid tion by tame mule deer on big sagebrush range, regions. Advances in Ecological Research 20:179–223. British Columbia. Journal of Range Management 31: ______. 2000. Synecology and disturbance regimes of 192–199. sagebrush steppe ecosystems. Pages 15–26 in P. G . YEO, J.J. 1981. The effects of rest-rotation grazing on mule Entwistle, A.M. DeBolt, J.H. Kaltenecker, and K. deer and elk populations inhabiting the Herd Creek Steenhof, compilers, Proceedings: Sagebrush Steppe allotment, East Fork Salmon River, Idaho. Master’s Ecosystems Symposium. Bureau of Land Manage- thesis, University of Idaho, Moscow. ment Publication BLM/ID/PT-001001+1150, Boise, ______. 2001. Long-term vegetation trend in the BLM ID. Challis Resource Area, east central Idaho. Unpub- WEST, N.E., F.D. PROVENZA, P.S. JOHNSON, AND M.K. lished report to the Challis Field Office, Bureau of OWENS. 1984. Vegetation change after 13 years of Land Management, Challis, ID. livestock exclusion on sagebrush semidesert in west YEO, J.J., W.T. WITTINGER, AND J.M. PEEK. 1990. Vegeta- central Utah. Journal of Range Management 37: tion changes on a rest-rotation grazing system. 262–264. Rangelands 12:220–225. WESTOBY, M., B. WALKER, AND I. NOY-MEIR. 1989. Oppor- tunistic management of rangelands not at equilib- Received 25 March 2003 rium. Journal of Range Management 42:266–274. Accepted 8 March 2004 Western North American Naturalist 65(1), © 2005, pp. 103–111

NOTES ON SIGNIFICANT COLLECTIONS AND ADDITIONS TO THE FLORA OF GLEN CANYON NATIONAL RECREATION AREA, UTAH AND ARIZONA, BETWEEN 1992 AND 2004

John R. Spence1

ABSTRACT.—Symphyotrichum expansum (Puepp ex Spreng.) Nesom is reported new to Utah from the Escalante River drainage. A major range extension is reported for Aralia racemosa L. in the Escalante drainage, and additional populations are reported of the rare species Imperata brevifolia Vasey in Utah, including the 1st record for the Grand Staircase–Escalante National Monument. Heterotheca grandiflora Nutt. is reported new to north central Arizona. New locations and notes on an additional 22 rare species in Glen Canyon National Recreation Area are listed.

Key words: flora, Glen Canyon National Recreation Area, Utah, Arizona, relicts, dispersal.

Intensive fieldwork in Glen Canyon National emerge at the Navajo-Kayenta interface. In Recreation Area (NRA) was conducted on addition to these records, new records for the riparian communities in side canyons around Colorado River below Glen Canyon Dam in Lake Powell between 1991 and 2002 (cf. Spence northern Arizona are also discussed. 1996, 2005). This work has added numerous Below, the collection locality, habitat, and species to the NRA, as well as 1 species new to significance for the new Utah State record are Utah. In this paper significant collections are presented; then other records are listed with reported, including species that are rare in the families and genera arranged alphabetically. region or that represent new range extensions. Nomenclature follows Welsh et al. (2003) unless Glen Canyon NRA comprises 508,000 ha in otherwise noted. In a few cases the current south central Utah and north central Arizona, accepted name in the USDA Plants database 13% (66,000 ha) of which is occupied by Lake (http://plants.usda.gov) is used instead of Welsh. Powell. Over 440,000 ha of arid and semiarid Each species is represented by 1 or more col- vegetation along the Colorado River drainage lections, although specimens were not collected system occurs within the NRA, much of it for some species at all newly reported localities. rugged and inaccessible. Currently, ca. 800 Specimens are deposited in the Glen Canyon species have been collected or are known NRA herbarium and Northern Arizona Uni- (Spence and Zimmerman 1996), while an addi- versity (AST). Universal transmercator (UTM) tional 100 species are known from adjacent Bureau of Land Management, Navajo Nation, coordinates are based on the NAD27 datum. and National Park Service lands. The flora is Duplicates of Perityle specuicola from the San based primarily on inventories completed in Juan River are located at BYU. The ecological the 1980s (Welsh 1984, Schulz et al. 1987). setting of many of the species is reported else- As part of a riparian vegetation survey of where (Spence 1996), while the distribution selected side canyons around Lake Powell, and ecology of several rare species found in plant collections were made of rare or other- relict stands of Pseudotsuga menziesii are re- wise interesting species (Spence 1996). Most ported in Spence (1995). The status, distribu- work was conducted in canyons draining into tion, and ecology of 3 additional rare species the lake incised through the Triassic–Jurassic in Utah, Cladium californicum (Wats.) O’Neill, Glen Canyon group, comprising from youngest Cycladenia jonesii Eastwood, and Platanthera to oldest the Navajo, Kayenta, and Wingate For- zothecina (Higgins & Welsh) Kartesz & Gandhi, mations. Many species were associated with will be reported elsewhere (Spence in pre- springs, common in these canyons, which paration).

1National Park Service, Resource Management Division, Glen Canyon National Recreation Area, Box 1507, Page, AZ 86040.

103 104 WESTERN NORTH AMERICAN NATURALIST [Volume 65

NEW TO UTAH orado Plateau generally along streams and around springs, the new locations for this ASTERACEAE species are at unusually low elevations. The 2 Symphyotrichum expansum (Puepp ex Spreng.) Nesom. populations in Cow and Fence Canyons UTAH: Kane County. Cow Canyon, main fork, Esca- occurred only 30–90 m above the high-water lante Arm of Lake Powell. Along stream in lower portion of canyon, on disturbed, moist sand. Associated elevation of Lake Powell at 1130 m. These with Panicum virgatum, Thinopyrum ponticum, and populations may be relicts from the late Wis- Plantago lanceolata. Elevation 1160 m. UTM: consin when the species was more common at 12E505600N4140200. 31 July 1992. Spence 4966. lower elevations. Remarkably, macrofossils of Symphyotrichum expansum is now a com- this species have been found in late Wisconsin mon species along the Colorado River below and early Holocene deposits in canyons on the Glen Canyon Dam in Arizona, although it may west side of the Escalante Arm of Lake Powell have been rare prior to the construction of the within a few kilometers of these stands (With- dam in 1963. It probably extended well into ers and Mead 1993). Utah along the Colorado River and its tributaries, in areas now drowned by Lake Powell. ANACARDIACEAE Since it tends to flower in late summer and fall, it could have been easily overlooked dur- Rhus glabra L. UTAH: Kane County. Cow Canyon, north fork, Escalante Arm of Lake Powell. Around spring in ing the river studies conducted while Glen southeast facing alcove at base of Navajo Sandstone Canyon Dam was being constructed. A variety cliff, growing in mixed deciduous woodland of Quer- of other species, typical of riparian and spring cus gambelii and Frangula betulifolia. Elevation 1260 vegetation in the lower Grand Canyon and m. UTM: 12E507380N4143200. 30 July 1992. Spence Sonoran and Mojave Deserts, also follow this 4972. West fork of Bowns Canyon, far end in eastern- most alcove, associated with Frangula betulifolia and same pattern of extending along the Colorado Cirsium rydbergii. Elevation 1280 m. UTM: River into Utah. Other species displaying this 12E496300N4141250. 16 July 1997 (not collected). pattern include Baccharis salicifolia, Chlo- Coyote Gulch, at southeast-facing spring near Jacob racantha spinosa, Cercis occidentalis, Cladium Hamblin’s Arch. Associated with Adiantum capillus- veneris, Frangula betulifolia, Rosa woodsii, and Toxico- californicum, Imperata brevifolia, Parthenocis- dendron rydbergii. Elevation 1220 m. UTM: sus vitacea, Frangula betulifolia, and Tessaria 12E496300N4141250. 10 May 2002 (not collected). sericea. These represent the 2nd through 4th populations in Glen Canyon NRA. The only other SIGNIFICANT COLLECTIONS population is from a hanging garden at Buoy IN GLEN CANYON NATIONAL 73 Mile on Lake Powell (Welsh 1989). In RECREATION AREA southern Utah this species is rare and is most ACERACEAE common in the upper Virgin River drainage in

Acer grandidentatum Nutt. in T. & G. UTAH: Kane County. Zion National Park. Woodbury (1959) reported Cow Canyon, north fork, Escalante Arm of Lake Pow- smooth sumac as “occasional” in “hillside ell. Along stream in shaded, north-facing alcove at glens” along the Colorado River. All these base of Navajo Sandstone cliff, growing with Acer populations were drowned by Lake Powell. negundo, Quercus gambelii, and Frangula betulifolia. Elevation 1200 m. UTM: 12E510400N4145350. 30 July 1992 (not collected). Fence Canyon, Escalante ARALIACEAE Arm of Lake Powell. Along stream in shallow, north- Aralia racemosa L. UTAH: new to Kane County. Cow Can- facing alcove on steep colluvial slope at base of Navajo yon, north fork, Escalante Arm of Lake Powell. Along Sandstone cliff, growing with Acer negundo, Betula stream in shaded, north-facing alcove at base of occidentalis, Quercus gambelii, and Frangula betulifo- Navajo Sandstone cliff, growing under mixed wood- lia. Elevation 1260 m. UTM: 12E505450N4138900. 29 land of Acer grandidentatum, A. negundo, Quercus July 1993 (not collected). Millers Creek, off Halls Creek, gambelii, and Frangula betulifolia. Elevation 1200 m. Waterpocket Fold. Growing in dense shade under Dou- UTM: 12E510400N4145350. 30 July 1992. Spence glas-fir in north-facing alcove, associated with Ostrya 4976. knowltonii, Quercus gambelii, Mahonia repens, Platan- thera zothecina, and Maianthemum stellatum. Elevation In Utah, Aralia racemosa was considered 1770 m. UTM: 12E507160N4158040. 24 September restricted to narrow, shaded canyons in Zion 1992. Spence 5059. National Park and immediately adjacent areas Although Acer grandidentatum is a com- (Welsh et al. 1993). The Cow Canyon locality is mon species at higher elevations on the Col- about 180 km northeast of Zion Canyon. About 2005] ADDITIONS TO THE FLORA OF GLEN CANYON 105

50 plants were counted in 1992 and on a return known only from the type locality on Pollywog visit in 1994. The species had flowered both Bench, just upstream from the confluence of years and had set fruit in 1994. The stream, the Escalante Arm and main channel of Lake along which the plants grow, issues from a per- Powell. Surveys conducted on the west end of manent spring. The site receives no direct Pollywog Bench located ca. 100 individuals in sunlight. The Aralia was associated with a mixed May 1995. However, during an August 1995 deciduous woodland that is widespread in side trip, a large population, numbering in the canyons around Lake Powell. These wood- hundreds, was located down-lake from the lands harbor several boreal-montane disjuncts confluence on the east side of the canyon, as well as state rare species. At the spring between Buoy Markers 65 and 66 Mile. Typi- where Aralia grew, other species present in- cally, the plants grow on exposed, seepy slopes cluded Acer grandidentatum, Amelanchier alni- in the Kayenta Formation where soil has accu- folia, Carex rossii, Glyceria striata, Platanthera mulated and carbonate deposits occur at the zothecina, Mahonia repens, and Ostrya knowl- surface. The Long Canyon population showed tonii. These probably represent remnants of late some differences in morphology and habitat Wisconsin woodlands that may have occurred and may not be closely related to the other on stream bottoms and in side alcoves in these populations. The leaves of this population are canyons during glacial climates. wider and the heads somewhat larger than plants at Pollywog Bench. The habitat, on damp ASTERACEAE shaded soil at a spring, is also distinctive.

Erigeron kachinensis Welsh. UTAH: Garfield County. Clear- Heterotheca grandiflora Nutt. ARIZONA: Coconino County. water Canyon, off Cataract Canyon, just down from Junction of Highway 89 and Lakeshore Drive, south major fork, at top of talus on west-facing slope, base of entrance, along roadside in disturbed sandy soil, grow- Cedar Mesa Sandstone, in seepy area. Associated with ing with Baileya multiradiata and Machaeranthera Carex curatorum and Hedeoma drummondii. Eleva- canescens. Elevation 1170 m. UTM: 12E455930 tion 1550 m. UTM: 12E574400N4208050. 13 August N4087910. 24 October 2002. Spence 5533. 1992. Spence 5014. This is a new report for Glen Canyon NRA. Telegraph weed is a weedy native species This locality extends the range of E. kachinen- found in the southwestern deserts of North sis slightly west of its known distribution America. This is the 1st report for the region (Cronquist 1994). The central part of Clear- and a significant range extension from known water Canyon is difficult to reach as it is pro- populations in Washington County, Utah, and tected by cliffs above and below. Because of Yavapai County, Arizona. The location sug- this, it had apparently not been explored gests that it may have been brought in as part floristically before 1992. Within this protected of a seed mix, probably also including the part of the canyon, the following species (in exotic Baileya multiradiata, that was used to addition to E. kachinensis) were collected: revegetate the roadside at the site. Although this project was completed in the late 1980s, Carex rossii, Ostrya knowltonii, Perityle specu- telegraph weed is well known to produce dor- icola, and Rubus neomexicanus. mant seeds from ray flowers (Flint and Palm-

Erigeron zothecinus Welsh. UTAH: Kane County. Long blad 1978), and it is possible that fruits have Canyon, near upper end of canyon on damp, shaded been dormant in the seed bank since that soil in seep along stream. Associated with Juncus ensi- time. Alternatively, the species may have been folius and Epilobium ciliatum. Elevation 1400 m. UTM: recently and inadvertently brought to the site 12E513000N4142800. 3 September 1992. Spence 5044. through transport on tourist vehicles and San Juan County. Seepy soil slopes near Lake Powell, Buoy Marker 66 Mile, and along drainage to east. Ele- boats, some of which originate from the Las vation 1140 m. UTM: 12E512450W4124990. 16 August Vegas–Lake Mead region. 1995. T. Haberle s.n. Pectis angustifolia Torr. UTAH: Kane County. Long Canyon, The status of E. zothecinus is not well ca. 1 km from upper end, on open, east-facing, sandy understood. Although Welsh et al. (1993) con- slopes. Associated with Ipomopsis gunnisonii and Erio- sidered it a good species, Cronquist (1994) gonum palmerianum. Elevation 1400 m. UTM: placed it under E. pumilus. The species and its 12E513000N4142800. 3 September 1992. Spence habitat appear distinctive, however, compared 5050. with typical E. pumilus, and its status needs to Pectis angustifolia is a rare species in Utah, be investigated. Originally E. zothecinus was previously known from a few sites in southern 106 WESTERN NORTH AMERICAN NATURALIST [Volume 65

Kane and San Juan Counties. It is fairly com- lation also occurs in the mountains of south mon in sandy sites in Coconino County in central New Mexico and adjacent Texas. The northern Arizona. species is rare throughout most of its range. In Glen Canyon NRA there are only 7 known Perityle specuicola Welsh & Neese. UTAH: Garfield County. populations, including the 4 reported here. Clearwater Canyon, off Cataract Canyon, downstream ca. 200 m from major fork in canyon, in exposed cracks of Cedar Mesa Sandstone in dry stream channel. Asso- BRASSICACEAE ciated with Brickellia microphylla. UTM: 12E574400 Rorippa islandica (Oeder) Borbas. ARIZONA: Coconino N4208050. 13 August 1992. Spence 5013. San Juan County. Glen Canyon, rooted in mud and sand along County. San Juan River, Point Lookout Canyon, on margins of return channel marsh along Colorado River, right (north side) at river mile 62.5 at Government 6.5 miles upstream from Lees Ferry. Associated with Rapid. In hanging garden above 1st ledges. 13 Sep- Melilotus officinalis and Juncus articulatus. Elevation tember 1997. Atwood, Curtis and Melloy 23319 (BYU). 950 m. UTM: 12E450500N4080900. 5 August 1992. San Juan County, Easter Pasture Canyon overlooking Spence 4975. Cataract Canyon, on dry rock faces around hanging garden, 1st drop from top toward river. Elevation 1460 The only other record for this species along m. UTM: 12E578180N4208940. 5 August 1998 (not the Colorado River below Glen Canyon Dam collected). is at river mile 51.5 (miles downstream from The first 2 collections extend the range of Lees Ferry; 84.4 km) in Marble Canyon (Ayers Perityle specuicola in Cataract Canyon several et al. 1994). miles to the west of previously known locations in and around Dark Canyon (Cronquist CYPERACEAE 1994). The San Juan River population repre- Carex rossii F. Boott. UTAH: Garfield County. Clearwater sents a considerable range extension south Canyon, off Cataract Canyon, on soil in shade of from the species main center of distribution in Ostrya knowltonii woodland on stream terrace ca. 200 m up west fork of canyon. Associated with Penstemon and around Natural Bridges National Monu- rostriflorus, Rubus neomexicanus, and Symphoricarpos ment. longiflorus. Elevation 1450 m. UTM: 12E572900 N4209100. 13 August 1992. Spence 5018. Millers BETULACEAE Creek, off Halls Creek, Waterpocket Fold. Growing in dense shade under Douglas-fir in north-facing alcove, Ostrya knowltonii Cov. UTAH: Garfield County. Clearwa- associated with Ostrya knowltonii, Quercus gambelii, ter Canyon, off Cataract Canyon, forming woodland Mahonia repens, Platanthera zothecina, and Smilacina on stream terrace ca. 200 m up west fork of canyon. stellata. Elevation 1770 m. UTM: 12E507160 Associated with Penstemon rostriflorus, Rubus neo- N4158040 (not collected). Wayne County. Millard Can- mexicanus, and Symphoricarpos longiflorus. Elevation yon, near Hans Flat. In alcove at upper end of canyon 1450 m. UTM: 12E572900N4209100. 13 August 1992. under Douglas-fir, at base of wet detritus slope associ- Spence 5021. Northeast Fork of Cow Canyon, upper ated with Amelanchier alnifolia, Rosa woodsii, and Escalante Arm of Lake Powell, in north-facing alcove Cornus sericea. Elevation 1890 m. UTM: 12E574540 with spring. Associated with Amalanchier alnifolia, N4233100 (not collected). Aralia racemosa, Galium aparine, Mahonia repens, Pla- tanthera zothecina, Parthenocissus vitacea, and Fran- These populations are associated with gula betulifolia. Elevation 1350 m. UTM: 12E510440 patches of mixed deciduous woodland and N4145400. 30 July 1992 (not collected). South fork of Douglas-fir stands at unusually low elevations Ticaboo Canyon, in north-facing alcove with spring. for the species on the Colorado Plateau. Associated with Cercis occidentalis, Quercus gambelii, Frangula betulifolia, Symphoricarpos longiflorus, and Cyperus squarrosus L. ARIZONA: Coconino County. Glen Toxicodendron rydbergii. Elevation 1250 m. UTM: Canyon, rare in bare mud along margins of return 12E539720N4172900. 14 August 1996 (not collected). channel marsh along Colorado River, 6.5 miles Waterpocket Fold, in dense shade under Douglas-fir upstream from Lees Ferry. Elevation 950 m. UTM: in north-facing alcove, associated with Acer grandi- 12E450500N4080900. 4 October 1994. Spence 5248. dentatum, Quercus gambelii, Mahonia repens, Platan- thera zothecina, and Maianthemum stellatum. Eleva- This is a new record for the Colorado River tion 1770 m. UTM: 12E507160N4158040. 24 Septem- between Glen Canyon Dam and Lake Mead ber 1992. Spence 5059. (Phillips et al. 1987, Ayers et al. 1994). The Knowlton hophornbeam is a small tree plants were growing in an area that had been found along the Colorado River in southern thoroughly surveyed several times between Utah from near the Colorado border to the 1992 and 1994. The plants had not been seen Escalante River drainage. It reappears in the before this collection nor have they been seen Grand Canyon region (Brian and Spamer 2000) since then. Hence, this record apparently rep- and along the Mogollon Rim. A disjunct popu- resents a recent dispersal to the area, possibly 2005] ADDITIONS TO THE FLORA OF GLEN CANYON 107 by waterfowl, which are abundant along this rooted in mud. Associated with Typha domingensis, stretch of the river in winter (NPS unpub- Phragmites australis, and Leersia oryzoides. Elevation lished data, Spence and Bobowski 2003). 950 m. UTM: 12E (not collected). Lycopus americanus is a common marsh- DRYOPTERIDACEAE land and riparian species in northern and eastern Utah and Colorado. It has been reported Cystopteris utahensis Windham & Haufler. UTAH: Kane County. Cow Canyon, south fork, Escalante Arm of from a few localities at high elevations on the Lake Powell. In rock crevices near spring in perma- Navajo Reservation and White Mountains. nently shaded, north-facing alcove at base of Navajo These new locations are over 200 km west of Sandstone cliff. Elevation 1285 m. UTM: 12E509400 these localities. The Horseshoe Bend popula- N4143050. 31 July 1992. Spence 4977. Millers Creek, off Halls Creek, Waterpocket Fold. On boulders along tion was found associated with other wetland stream in dense shade under Douglas-fir in north-fac- species, including Leersia oryzoides, in a small ing alcove. Associated with Acer grandidentatum, natural marsh, the lower edges of which are Ostrya knowltonii, Quercus gambelii, Mahonia repens, trimmed by high flows of the Colorado River. Platanthera zothecina, and Maianthemum stellatum. Elevation 1770 m. UTM: 12E507160N4158040 (not The other locality was an artificial return- collected). Coyote Gulch, in north-facing alcove with channel marsh resulting from fluctuating flows spring. Associated with Galium aparine, Platanthera from Glen Canyon Dam (cf. Stevens et al. zothecina, Frangula betulifolia, Maianthemum stellatum, 1995). and Toxicodendron rydbergii. Elevation 1220 m. UTM: 12E499280N4140740. 9 May 2002 (not collected). LILIACEAE Cystopteris utahensis is generally found in Zigadenus vaginatus (Rydb.) Macbr. UTAH: Kane County. moist, shaded sites in the mountains on the At head of south fork of main north fork of Cow Colorado Plateau, including the Abajo, La Sal, Canyon, Escalante Arm of Lake Powell, in large Henry, Boulder, and Navajo Mountains. The alcove, on wet backwall of hanging garden. Associated collections from the Escalante River drainage with Aquilegia micrantha, Calamagrostis scopulorum, in Cow Canyon and Coyote Gulch are at an Carex curatorum, Cirsium rydbergii, and Mimulus eastwoodiae. Elevation 1372 m. UTM: 12E510800 unusually low elevation for the species on the N4145780. 30 July 1992 (not collected). At head of Colorado Plateau. The plants in Millers Creek main south fork of Cow Canyon, Escalante Arm of are associated with the relict stand of Dou- Lake Powell, in large alcove on wet backwall of hang- glas-fir. ing garden. Associated with Aquilegia micrantha, Calamagrostis scopulorum, Carex curatorum, Cirsium rydbergii, and Mimulus eastwoodiae. Elevation 1292 EUPHORBIACEAE m. UTM: 12E509840N4143350. 31 July 1992 (not col- Euphorbia aaron-rossii A. Holmgren and N. Holmgren. lected). At head of Fence Canyon, Escalante Arm of ARIZONA: Coconino County. On steep limestone slope Lake Powell, in large alcove on wet backwall of hang- at mouth of Cathedral Wash, above the Colorado ing garden. Associated with Adiantum capillus-veneris, River, 4.4 km downstream from Lees Ferry. Associated Aquilegia micrantha, Calamagrostis scopulorum, Cir- with Atriplex confertifolia and Ephedra torreyana. Ele- sium rydbergii, and Lobelia cardinalis. Elevation 1280 vation 975 m. UTM: 12E444860N4077850. 10 May m. UTM: 12E507050N4139780. 29 July 1992 (not col- 1994. J. Spence & J. Crawford s.n. lected). At head of unnamed canyon on Escalante Arm, 1st canyon north of Cow Canyon on east side of lake, Euphorbia aaron-rossii was described from in large alcove on wet backwall of hanging garden. collections along the Colorado River in Mar- Associated with Adiantum capillus-veneris, Aquilegia ble Canyon (Holmgren and Holmgren 1988). micrantha, Calamagrostis scopulorum, Carex curatorum, This is the 1st record for Glen Canyon NRA Cirsium rydbergii, and Mimulus eastwoodiae. Eleva- tion 1210 m. UTM: 12E504183N4141230. 15 August and extends the species distribution to above 1995 (not collected). San Juan County. In small hanging Navajo Bridge at the mouth of Cathedral Wash, garden near entrance to Ribbon Canyon, Lake Powell, about 4.4 km downstream from Lees Ferry. on detritus slopes and backwall. Associated with Rubus neomexicanus. Elevation 1130 m. UTM: 12E514300 N4123480. 9 June 1992 (not collected). LAMIACEAE Sheathed deathcamas is a rare species dis- Lycopus americanus Muhl. ex Barton. ARIZONA: Coconino County. Glen Canyon, rooted in mud and sand along tributed in hanging gardens along the Colo- margins of return channel marsh along Colorado rado River drainage in southeastern and south River, 6.5 miles upstream from Lees Ferry. Associated central Utah. Within Glen Canyon NRA the with Mentha arvensis, Juncus articulatus, and Euthamia species was known from 3 locations (Welsh occidentalis. Elevation 950 m. UTM: 12E450500 N4080900. 5 August 1992. Spence 4975. In marsh 8.8 1989). These additional locations bring the miles upstream from Lees Ferry, Horseshoe Bend, number of populations of the species in the 108 WESTERN NORTH AMERICAN NATURALIST [Volume 65

NRA to 8. In Glen Canyon the species is typi- on the banks of the Colorado River down- cally found in alcoves in large, shaded hanging stream of Glen Canyon Dam. This collection gardens on wet backwalls. Most populations fills in a significant gap in the distribution of are inaccessible, as they are usually growing rice cutgrass on the central Colorado Plateau. on seeping cliffs >100 m aboveground. The species may have been introduced to the site by waterfowl that overwinter on the river POACEAE below the dam (NPS unpublished data). Imperata brevifolia Vasey. UTAH: new to Kane County. Coyote Gulch, in sand along stream in lower portion, Sporobolus asper (Michx.) Kunth. UTAH: new to Kane associated with Baccharis emoryi, Juncus balticus, County. Long Canyon, ca. 3 km from upper end, on Equisetum hyemale, and Scirpus pungens. Elevation dry soil along stream. Elevation 1380 m. UTM: 1190 m. UTM: 12E499380N4141175. 9 May 2002 (not 12E513300N4141600. 3 September 1992. Spence collected). In sand along stream in lower portion, asso- 5052. ciated with Juncus balticus and Salix exigua. Elevation 1195 m. UTM: 12E499374N4141172. 9 May 2002 (not Tall dropseed is known from a few localities collected). In sand on terrace in upper Coyote Gulch in eastern Utah. The population in Long Canyon above confluence with Hurricane Wash, growing with represents a significant range extension west- Juncus balticus, Equisetum hyemale, Salix exigua, and ward from populations in Grand County. Scirpus pungens. Elevation 1250 m. UTM: 12E494200 N4141920. 10 May 2002 (not collected). POLEMONIACEAE Satintail grass is a distinctive species that Gilia flavocincta A. Nels. UTAH: new to Garfield County, was known previously from 3 locations in Two Mile Canyon, ca. 4 km up-canyon from Lake Utah, in Wilson Creek on the lower San Juan Powell, above major drop in canyon, in dry sand along Arm of Lake Powell, at the mouth of Forbid- dry wash. Associated with Physaria acutifolia. Eleva- ding Canyon, and from the vicinity of Rain- tion 1250 m. UTM: 12E542000N4181900. 22 April bow Bridge (Woodbury 1958). The latter 2 1992. Spence 4936. locations are presumably under Lake Powell A rare species known from a single previ- as the species has not been relocated at these ous collection in Utah, in Kane County (Welsh sites. These 3 new populations in Coyote et al. 1993), this newly discovered population Gulch increase the number of extant popula- extends the range to southern Garfield County. tions in the state to 4. The 2 lower Coyote Gulch populations are in Glen Canyon NRA, POTAMOGETONACEAE while the upper population is in the Grand Potamogeton natans L. UTAH: new to Kane County. Bowns Staircase–Escalante National Monument near Canyon, southern end of Waterpocket Fold. In still the boundary with the NRA. Even with these water of pools in lower canyon and in beaver ponds new locations, the status of this species in just up from fork. Associated with Typha domingensis. Utah remains precarious. All 3 new popula- Elevation 1320 m. UTM: 12E511900N4134950. 19 tions are small and are located in the flood August 1992. Spence 5031. zone of Coyote Gulch. The upper Coyote Potamogeton natans is a rare species dis- Gulch population is within 50 m of a livestock tributed in ponds, lakes, and slow-moving fence, and the area had been grazed until the streams in northern Utah, in Duchesne, Rich, fence was built in 1992. Coyote Gulch Uinta, and Utah Counties (Albee et al. 1988, receives heavy recreational use, and all 3 pop- Welsh et al. 1993). The present location repre- ulations show signs of trampling by humans. sents a major range extension southward in the state and is at a lower elevation and in a Leersia oryzoides (L.) Swartz. ARIZONA: new to Coconino lower vegetation zone than is typical for the County. Colorado River, Horseshoe Bend, river mile species. It is common in beaver ponds in 8.8, left bank, rooted in mud along marshy edges of stream. Associated with Typha domingensis and Scir- Bowns Canyon. pus acutus. Elevation 950 m. UTM: 12E454100 N4081145. 23 September 1993. Spence 5223a. ROSACEAE

Rice cutgrass is known from western Col- Amelanchier alnifolia Nutt. UTAH: Kane County. Cow Can- orado and the Great Salt Lake and Utah Lake yon, Escalante Arm. At spring in shaded, north-facing areas of northern Utah. It has also been found alcove at base of Navajo Sandstone cliff, growing in mixed woodland of Acer grandidentatum, A. negundo, in southern Arizona, where it is possibly intro- Quercus gambelii, and Frangula betulifolia. Elevation duced (Kearney and Peebles 1960). The popu- 1200 m. UTM: 12E510400N4145350. 13 June 1994. lation was found in a natural, spring-fed marsh Spence and J.A.C. Zimmerman 5231; Wayne County. 2005] ADDITIONS TO THE FLORA OF GLEN CANYON 109

Millard Canyon, near Hans Flat, in alcove at upper yon Dam in 1963 and the subsequent filling of end of canyon under Douglas-fir, at base of wet detri- Lake Powell. Sporadic fieldwork occurred tus slope associated with Cornus sericea, Rosa woodsii, and Carex rossii. Elevation 1890 m. UTM: between the 1930s and 1950s, summarized in 12E574540N4233100. 3 August 1994 (not collected). Woodbury (1958). Clover and Jotter (1944) In both cases the species was represented sampled sites along the river corridor in 1938, by a single individual. This species generally including 2 in Glen Canyon, the mouth of For- occurs at much higher elevations on the Col- bidding Canyon, and the vicinity of Rainbow orado Plateau, generally in montane scrub and Bridge in Bridge Canyon. Gaines (1960) re- forests. ported some additional species from the Glen Canyon region. In the summers of 1957 and Rubus neomexicanus Gray. UTAH: new to Garfield County. 1958, two expeditions were launched to sur- Clearwater Canyon, left fork ca. 200 m above major vey the vegetation and collect the flora and fork in canyon, on shaded terrace along stream under Ostrya knowltonii. Elevation 1450 m. UTM: 12E572900 fauna of Glen Canyon, between Hite and Lees N4209100. 13 August 1992. Spence 5017. Ferry. Most of this work concentrated on the main river corridor, with occasional side can- Previously known from populations in yon visits including Aztec, Lake, Little Eden, Knowles, Ribbon, and Cataract Canyons, the North Wash, Red, and Trachyte Canyons (Wood- Clearwater Canyon location represents only the 4th record for Glen Canyon NRA and Utah. bury 1959). After the filling of Lake Powell, a major floristic study was conducted in Glen RUTACEAE Canyon NRA (Welsh 1984), which added numerous species to the known flora of the side Ptelea trifoliata L. ARIZONA: Coconino County. Glen Can- yon, on sandy benches along Colorado River 7.0 miles canyons. upstream from Lees Ferry, at base of Navajo Sand- Most of the new records for the Glen Can- stone cliff. Associated with Celtis reticulata, Forestiera yon region are associated with springs rather pubescens, Galium trifolium, and Quercus turbinella. than riparian or upland vegetation. The side Elevation 965 m. UTM: 12E451350N4080930 (not collected). canyons of Glen Canyon where many of these springs occur may have been inaccessible from Hoptree is a small tree found in the south- the river due to pour-offs and mass-wasting western deserts and mountains of North events in the narrow and deep lower portions America. Although common in the Grand of the canyons. Also, springs on the sides of Canyon, the species is not found along the the canyon walls along the river may have Colorado River in riparian vegetation. This been difficult to reach from river level. Hence new record is the 1st for Glen Canyon NRA some of the species reported here may have and is associated with remnant, pre-dam, old, been in sites difficult to access during pre-dam high-water-zone riparian vegetation. surveys. With the creation of Lake Powell, the lower portions of these canyons were drowned SCROPHULARIACEAE while many springs and upper portions of the Mimulus eastwoodiae Rydb. UTAH: Kane County. Millers side canyons became readily accessible by boat. Creek, off Halls Creek, south end of Waterpocket Fold. Seep in east-facing alcove, near Douglas-fir stand, Full pool elevation of the reservoir is 1130 m, associated with Calamagrostis scopulorum, Mahonia which is about 150–170 m above the original repens, Psuedotsuga menziesii, and Eriogonum corym- river. Of the 650 springs identified on the bosum var. orbiculatum. Elevation 1790 m. UTM: Glen Canyon NRA GIS theme, 240 (37%) 12E507100N4158200. 24 September 1992 (not collected). were drowned by the reservoir, including all the glens and springs described by J.W. Powell The population occurred at an elevation on his 1869 descent through Glen Canyon. It 415 m higher than previously reported for the is likely that many unusual and interesting species in Utah (Welsh et al. 1993). The plants plant communities and species existed at these were growing in an east-facing hanging gar- springs and alcoves that were not sampled den near the isolated stand of Douglas-fir in upper Millers Creek, Waterpocket Fold. during the 1957–1958 expeditions and were subsequently destroyed when Glen Canyon DISCUSSION Dam was built. At least 3 species reported from the river corridor are no longer extant in Glen Canyon was not thoroughly surveyed Glen Canyon NRA, Mamillaria tetrancistra, floristically before the completion of Glen Can- Montia perfoliata, and Prunus virginiana. Tw o 110 WESTERN NORTH AMERICAN NATURALIST [Volume 65 other species that were listed by Clover and favorable microsites since the Wisconsin period, Jotter (1944) and Woodbury (1958) as occur- other species are likely to have dispersed ring in Glen Canyon, Adiantum pedatum and more recently into the Glen Canyon region. Aquilegia chrysantha, may have been misiden- An interesting group of species common in the tified, as specimens do not exist and the lower Grand Canyon, and distributed primar- species are not known from the region. Other ily in the Sonoran and Mojave Deserts at species may have been locally extirpated from springs and in riparian vegetation, occurs in Glen Canyon as well. Glen Canyon: Cladium californicum, Symphyo- The surviving natural springs in Glen Can- phytum expansum, and Imperata brevifolia. yon support highly diverse plant communities These species may have expanded into the re- with many unusual species. These shaded, gion during the Holocene thermal maximum. cool, wet sites may have functioned as refugia They may also have been more widespread for species favoring microclimates that are prior to the creation of Lake Powell and could typically found at much higher elevations on have been missed during early surveys that the Colorado Plateau (Spence 2005). Numer- concentrated primarily at river level. Other ous boreal-temperate and montane species species, such as Cyperus squarrosus, Leersia have been found associated with springs in the oryzoides, Lycopus americanus, and Rorippa upper ends of drowned side canyons, includ- islandica, are currently found in marshes and ing Acer grandidentatum, Amelanchier alnifo- other new high-water-zone vegetation that has lia, Aralia racemosa, Betula occidentalis, Cala- developed downstream since the completion magrostis scopulorum, Carex rossii, Cornus of Glen Canyon Dam. The remaining 25 km of sericea, Cystopteris utahensis, Galium aparine, Glen Canyon below the dam supports abun- Glyceria striata, Mahonia repens, Prunus vir- dant wintering waterfowl populations that did giniana, Rhus glabra, Rosa woodsii, and Maian- not exist prior to the completion of the dam themum stellatum. A few of these were reported (Spence and Bobowski 2003). The rare and during the 1957–1958 trips in side canyons sporadic occurrence of wetland plant species and river level vegetation. These species are in this stretch of the river may thus be a result currently disjunct and isolated from higher- of long-distance dispersal by waterfowl that elevation mountain populations; they may rep- migrate south from areas where these plants resent remnants of Wisconsin-age glacial are common, such as northern Utah. The riparian woodlands that were widespread dur- Cyperus, Lycopus, and Rorippa were found in ing glacial climates. Some species associated a return-channel marsh that did not exist prior with these woodlands, such as Abies concolor to 1963 (Stevens et al. 1995). and Picea pungens, disappeared with the warming of the Holocene, while others could have ACKNOWLEDGMENTS persisted in cool, shaded alcoves where springs existed. Douglas-fir (Pseudotsuga menziesii), Thanks are extended to T. Ayers, R. Scott, another montane species, is a common macro- D. Atwood, and W. Rominger for identification fossil component in portions of the Escalante of difficult specimens. J.A. Crawford, T. Haberle, River drainage (Withers and Mead 1993) and and J. Muller helped in field surveys, for which still exists as small, isolated stands in the region I am grateful. (Spence 1995). Betancourt (1990) argued for a similar origin for the many rare and disjunct LITERATURE CITED boreal-temperate species found in Zion Canyon, which he suggested functioned as a “mega- ALBEE, B.J., L.M. SCHULTZ, AND S. GOODRICH. 1988. Atlas refugium” during the Holocene. Canyons off of the vascular plants of Utah. Utah Museum of Nat- the lower Escalante Arm of Lake Powell, in ural History, Salt Lake City. AYERS, T.J., R.W. SCOTT, L.E. STEVENS, K. WARREN, A. particular Cow and Fence Canyons, harbor PHILLIPS III, AND M.D. YARD. 1994. Additions to the many of the disjunct populations of boreal- flora of Grand Canyon National Park–1. Journal of temperate species in the Glen Canyon region the Arizona–Nevada Academy of Science 28:70–75. and on a smaller scale create microclimates BETANCOURT, J.L. 1990. Late Quaternary biogeography of similar to those of Zion Canyon. the Colorado Plateau. Pages 259–292 in J.L. Betan- court, R.R. Van Devender, and P.S. Martin, editors, Although the existence of boreal-temperate Packrat middens: the last 40,000 years of biotic species can be explained by persistence in change. University of Arizona Press, Tucson. 2005] ADDITIONS TO THE FLORA OF GLEN CANYON 111

BRIAN, N.J., AND E.E. SPAMER. 2000. Knowlton hop-horn- structure and environment. In: V. Meretsky and L.E. beam revisited (Ostrya knowltonii Cov.). Bartonia Stevens, editors, Springs of the American South- 60:49–56. west: ecology and management. University of Ari- CLOVER, E.U., AND L. JOTTER. 1944. Floristic studies in zona Press, Tucson. the Canyon of the Colorado and tributaries. Ameri- SPENCE, J.R., AND B.R. BOBOWSKI. 2003. 1994–1997 water can Midland Naturalist 32:591–642. bird surveys of Lake Powell, a large oligotrophic CRONQUIST, A. 1994. Intermountain flora. Volume five. reservoir on the Colorado River, Utah and Arizona. Asterales. New York Botanical Garden Press, The Western Birds 34:133–148. Bronx. SPENCE, J.R., AND J.A. ZIMMERMAN. 1996. Preliminary flora FLINT, S.D., AND I.G. PALMBLAD. 1978. Germination dimor- of Glen Canyon National Recreation Area. Unpub- phism and developmental flexibility in the ruderal lished report, National Park Service, Resource Man- weed Heterotheca grandiflora. Oecologia 36:33–43. agement Division, Glen Canyon National Recreation GAINES, X.M. 1960. An annotated catalogue of Glen Area. 27 pp. Canyon plants. Museum of Northern Arizona Tech- STEVENS, L.E., J.C. SCHMIDT, T.J. AYERS, AND B.T. BROWN. nical Series 4. 1995. Flow regulation, geomorphology, and Colo- HOLMGREN, A.H., AND N.H. HOLMGREN. 1988. Euphorbia rado River marsh development in the Grand Canyon, aaron-rossii (Euphorbiaceae), a new species from Arizona. Ecological Applications 5:1025–1039. Marble and Grand Canyons of the Colorado River, WELSH, S.L. 1984. Flora of Glen Canyon National Recre- Arizona. Brittonia 40:357–362. ation Area. Final report to the National Park Service, KEARNEY, T.H., AND R.H. PEEBLES. 1960. Arizona flora Glen Canyon NRA. with supplement. University of California Press, ______. 1989. On the distribution of Utah’s hanging gar- Berkeley. dens. Great Basin Naturalist 49:1–30. PHILLIPS, B.G., A.M. PHILLIPS III, AND M.A.S. BERNZOTT. WELSH, S.L., N.D. ATWOOD, S. GOODRICH AND L.C. HIG- 1987. Annotated checklist of vascular plants of GINS. 1993. A Utah flora. 2nd edition, revised. Print Grand Canyon National Park. Grand Canyon Nat- Services, Brigham Young University, Provo, UT. ural History Association Monograph 7. WELSH, S.L., N.D. ATWOOD, L.C. HIGGINS, AND S. GOOD- SCHULZ, L.M., E.E. NEELY, AND J.S. TUHY. 1987. Flora of RICH. 2003. A Utah flora. 3rd edition. Brigham Young the Orange Cliffs of Utah. Great Basin Naturalist 47: University Press, Provo, UT. 287–298. WITHERS, K., AND J.I. MEAD. 1993. Late Quaternary vege- SPENCE, J.R. 1995. Characterization and possible origins tation and climate in the Escalante River Basin on of isolated Douglas-fir stands on the Colorado Pla- the central Colorado Plateau. Great Basin Naturalist teau. Pages 71–82 in W. J. Waugh, editor, Climate 53:145–161. change in the Four Corners region. Proceedings of a WOODBURY, A.M., EDITOR.1958. Preliminary report on the symposium, Grand Junction, CO, 12–14 September biological resources of the Glen Canyon reservoir. 1994. Department of Energy NTIS CONF-9409325. University of Utah Anthropological Papers 31 (Glen ______. 1996. A survey and classification of the riparian Canyon Series 2). and spring vegetation in side canyons around Lake ______. 1959. Ecological studies of flora and fauna in Glen Powell, Glen Canyon National Recreation Area, Utah. Canyon. University of Utah Anthropological Papers Unpublished final report, National Park Service, 40 (Glen Canyon Series 7). Resource Management Division, Glen Canyon Na- tional Recreation Area. 90 pp. Received 10 February 2004 ______. 2005. Spring-supported vegetation along the Col- Accepted 3 August 2004 orado River, Colorado Plateau: floristics, vegetation Western North American Naturalist 65(1), © 2005, pp. 112–117

DISCOVERY OF THE MILLIPED SCYTONOTUS GRANULATUS (SAY, 1821) IN OKLAHOMA AND ALABAMA, WITH A REVIEW OF ITS DISTRIBUTION (POLYDESMIDA: POLYDESMIDAE)

Rowland M. Shelley1, Chris T. McAllister2, and Zachary D. Ramsey2

ABSTRACT.—Scytonotus granulatus (Say, 1821) (Polydesmida: Polydesmidae), the most widely ranging polydesmid species in North America and the 4th most widely distributed indigenous continental milliped, is recorded from LeFlore and Latimer Counties, Oklahoma, and Jackson County, Alabama, the 1st records from these states. It also occurs in Logan and Independence Counties, Arkansas. The Latimer County record corresponds approximately to the terminus of the eastern forested biome and extends the distributions of the species and the genus some 255 miles (408 km) westward; along with literature records from Shawnee County, Kansas, and Cass County, Nebraska, it constitutes the western range limits. The projected overall distribution extends around 1100 miles (1760 km) east–west and 985 miles (1576 km) north–south and encompasses parts of Ontario, Québec, and 19 states, including all of Pennsylvania, West Virginia, Ohio, Indiana, and Illinois. New localities are detailed as are those from Missouri and that from Dare County, on the Outer Banks of North Carolina.

Key words: Scytonotus granulatus, Scytonotus, Oklahoma, Arkansas, Missouri, Alabama.

A substantial number of the milliped gen- because fewer surveys have been conducted era and species that inhabit the Atlantic Coastal there and because some species are less abun- states range westward beyond the Mississippi dant and less frequently encountered in their River. Some distributions terminate in the for- range extremities. Consequently, distributions ested biome that ends primarily in eastern Texas west of this watercourse are typically poorly and Oklahoma and western Missouri, while documented, and the western termini are others extend into the prairie ecosystems of imprecise if not totally nebulous. With such the Central Plains. Two prominent examples minimal knowledge, a chance discovery of a are Narceus americanus (Beauvois) (Spirobol- single individual can represent new state ida: Spirobolidae)3, the most common eastern occurrences and significantly alter long-stand- chilognath milliped that Keeton (1960) recorded ing impressions of distributions, such that from Murray County in central Oklahoma, and published documentation is in order. This was Pleuroloma flavipes Rafinesque (Polydesmida: the situation with P. flavipes (Shelley et al. 2004), Xystodesmidae), which occurs from Connnecti- when the species and genus could be docu- cut to North Carolina and ranges westward to mented from Texas and the western limit ex- northeastern Texas, central Oklahoma and panded in Oklahoma and Kansas, and such is Kansas, southeastern Nebraska, and eastern now necessary for Scytonotus Koch and S. gran- North Dakota (Shelley 1980, Shelley et al. 2004). ulatus (Say, 1821) (Polydesmida: Polydesmidae). Most preserved samples of these westward- Occurring both east of the Central Plains and ranging millipeds are from eastern areas be- west of the Continental Divide, Scytonotus is cause of proximities to repositories in Atlantic 1 of only 2 indigenous polydesmidan genera Coastal states, as most past collectors were found on both sides of the North American located in the East and worked primarily there continent, the other being Ergodesmus Cham- and along the Gulf Coast. Fewer samples are berlin (Nearctodesmidae), which inhabits caves available from west of the Mississippi River in Illinois and epigean biotopes in the Pacific

1Research Lab, North Carolina State Museum of Natural Sciences, 4301 Reedy Creek Road, Raleigh, NC 27607. 2Biology Department, Texas A&M University–Texarkana, Texarkana, TX 75505. 3Opinions differ as to the relative statuses of N. americanus and N. annularis Rafinesque, which Keeton (1960) recognized as distinct species and recorded from west of the Mississippi. Shelley (1988) reduced the latter to a subspecies of the former, but Hoffman (1999) and Shelley (2001) did not recognize subspecies. Shelley (2003a) observed that Narceus is probably more complex than the current concept of 4 species and that a modern revision is needed to assess this possibility; substantial size differences among ostensibly conspecific individuals suggest that some names Keeton placed in synonymy represent valid species. For simplicity, we reference this milliped by the oldest name, N. americanus, until a definitive study can address the generic composition and taxonomic uncertainties in Narceus.

112 2005] DISTRIBUTION OF SCYTONOTUS GRANULATUS 113

Northwest and British Columbia (Hoffman laris Causey (Polydesmida: Eurymerodesmidae); 1962a, 1999, Shelley 1994, Whitney and Shel- Virgoiulus minutus (Brandt) (Julida: Blaniuli- ley 1995). It is 1 of 6 indigenous polydesmid dae); and Brachycybe lecontii Wood (Platy- genera in the West, where it comprises 6 spe- desmida: Andrognathidae). While sampling that cies and occupies 5 disjunct areas extending year on the western slope of Rich Mountain in from the Pacific Coast to western Wyoming the eastern fringe of LeFlore County, Oklahoma, and southeastern British Columbia (Shelley he collected a juvenile female of Scytonotus 1993, 2003b). In the East, however, it is 1 of only that constituted the 1st generic record from 2 family components, the other being Pseudo- this state and a dramatic range extension from polydesmus Attems, both of which occupy con- Jonesboro, the most proximate locality. The tinuous areas, extend northward into Ontario specific identity could not be determined from and Québec, and traverse the Mississippi this individual, and the range expansion was (Shelley 1988, 2002a, Hoffman 1999). Scytono- so significant that it might have been a new tus comprises 3 species in the East, 2 of which species, particularly because A. wilhelminae was are endemic to the Blue Ridge Mountains discovered simultaneously just 6 miles (9.6 km) from northern Virginia to north Georgia (Hoff- to the east in Polk County, Arkansas. We there- man 1962b, Shelley 1993). The rest of the fore deferred publication until a male could be eastern area is occupied solely by S. granula- taken in Oklahoma to positively determine the tus, a small-bodied species (adults ca. 8–9 mm species. Continued sampling at this site for 2 long) characterized by pinkish or grayish color, years failed to produce one, during which time dentate paranotal margins, and 4 to 5 rows of the senior author reexamined institutional hold- rounded, setose tubercles on the metatergites ings and discovered a sample of females from that impart a “fuzzy” or velveteen appearance Latimer County, the adjacent county to the to the dorsum. Present knowledge shows that west, and samples with a male of S. granulatus it ranges westward to the eastern plains in from Logan County, Arkansas, around 50 miles Kansas and riparian habitats, primarily along (80 km) to the northeast. This individual sug- the Missouri River, in southeastern Nebraska, gested that the Oklahoma species also was S. with the following additional occurrences west granulatus, and the 3rd author found a male in of the Mississippi: southeastern Minnesota, LeFlore County in December 2003. As the eastern and central Iowa and Missouri, and specific identity is now certain, we record be- northeastern Arkansas (Bollman 1893, Kenyon low the new Arkansas and Oklahoma localities 1893, Gunthorp 1913, Chamberlin 1928, 1942, along with another from northeastern Alabama, Chamberlin and Hoffman 1958, Shelley 1993, also constituting a new state record, that the Hoffman 1999). The westernmost records are 1st author discovered among unsorted samples in Shawnee County, Kansas, and Cass County, at the FSCA (see acronyms below). It contains Nebraska (Kenyon 1893, Gunthorp 1913), and a single female, which, though unidentifiable, the only Arkansas locality is Jonesboro, Craig- is surely S. granulatus that occurs in adjacent head County, in the northeast corner near the Franklin County, Tennnessee (Shelley 1993). “heel” of Missouri (Shelley 1993). We also detail localities from Missouri, which In 2001 the 2nd author began surveying were cited by county by Shelley (1993). Occur- myriapods in the “Ark-La-Tex” region to ele- rences west of the Mississippi are shown in vate knowledge of the fauna west of the Mis- Figure 1, and the projected overall distribution sissippi. One new species has been described, is depicted in Figure 2. Anatomical accounts Abacion wilhelminae Shelley, McAllister, and and pertinent illustrations are available in Hollis (Callipodida: Abacionidae), from Polk Hoffman (1962b) and Shelley (1978, 1988, 1993). County, Arkansas (Shelley et al. 2003), and sig- Repository acronyms are as follows: AMNH– nificant occurrences, including new state rec- American Museum of Natural History, New ords and western/southwestern limits, have York, New York; ANSP–Academy of Natural been published for 6 millipeds in addition to P. Sciences, Philadelphia, Pennsylvania; FSCA– flavipes (McAllister et al. 2002a, 2002b, 2003): Florida State Collection of Arthropods, Gaines- Thrinaxoria lampra (Chamberlin) (Polydesmida: ville; NCSM–North Carolina State Museum of Xystodesmidae); Eurymerodesmus mundus and Natural Sciences, Raleigh; UAAM–University E. dubius, both by Chamberlin, and E. angu- of Arkansas Arthropod Museum, Fayetteville; 114 WESTERN NORTH AMERICAN NATURALIST [Volume 65

UMO–Enns Entomological Museum, Ento- mology Department, University of Missouri, Columbia. ALABAMA: Jackson Co., ca. 17 mi (27.2 km) NW Scottsboro (0.5 mi [0.8 km] E Swaim), Swaim Cave (Cv.), , date and collector unknown (FSCA). New state record. ARKANSAS: Craighead Co., Jonesboro, , , 17 March 1962, N.B. Causey (FSCA) and , 30 October 1966, M. Hite (FSCA). Indepen- dence Co., 1.3 mi (2.1 km) W Cushman, vicinity of Blowing Cv., , 5 March 1973, R.M. Blaney (FSCA). Logan Co., Mt. Magazine, , 30 October 1986, J.A. Tedder (UAAM), Cam- eron Bluff, juv. , 25 July 1990, B. Leary (UAAM), and Signal Hill, 2 juvs., 21 July 1990, B. Leary (UAAM). MISSOURI: Boone Co., Columbia, 4, 4, 13 March 1966, W.W. Dowdy (FSCA) and , May 1968, G. Moe (UMO). Calloway Co., Jef- ferson City vicinity, 2, 28 July & 28 Decem- ber 1965, W.W. Dowdy (FSCA). Chariton Co., locality not specified, , , date unknown, C. Veatch (ANSP). Christian Co., Ozark, Smallin’s Cv., 2, 4, 8 December 1958, B. Ostting (FSCA). Cole Co., Jefferson City, 2 juvs., 29 August 1965, W.W. Dowdy (FSCA); along U.S. Hwy. 50E, juvs., 29 August 1965, W. W. Dowdy (FSCA), and Le Pages & Neil- horn Places, , , 26 February–8 March 1966, W.W. Dowdy (FSCA). Dent Co., 4 mi (6.4 km) SE Salem, 3, 12 October 1965, W. Ivie (AMNH). Phelps Co., 6 mi (9.6 km) S Rolla, , 11 October 1965, J. & W. Ivie (AMNH). OKLAHOMA: Latimer Co., locality not specified, 2, November 1990, K. Stephan (FSCA). LeFlore Co., ca. 22 mi (35.2 km) S Poteau, Choctaw Nation State Historic site along OK Hwy. 1 on Rich Mtn. a few yards inside OK state line (road becomes AR Hwy. 88 across the line), ca. 6 mi (9.6 km) W Queen Wilhel- mina St. Pk., Polk Co., AR, juv. , 24 Novem- ber 2001, C.T. & J.T. McAllister (NCSM), and under decaying oak log on Ouachita trail par- alleling highway, , juv. , 5 December 2003, Z. Ramsey (NCSM). New state record. Fig. 1. Occurrences of Scytonotus and S. granulatus west of the Mississippi River. Open symbols denote literature records believed to be valid. Of the indigenous North American milliped species whose distributions are reasonably well known, only 3 to our knowledge cover greater areas than S. granulatus; in decreasing order they are Oriulus venustus (Wood) (Julida: Parajulidae), N. americanus, and P. flavipes (Keeton 1960, Shelley 1980, 2002b, Shelley et 2005] DISTRIBUTION OF SCYTONOTUS GRANULATUS 115

Fig. 2. Overall distribution of S. granulatus; a smooth curve is drawn around range extremes in all directions. The dot denotes the approximate location of the new record from Alabama, and the square shows the record from Buxton, on the Outer Banks of North Carolina.

al. 2004). As S. granulatus occurs in Kansas, who overlooked the occurrence of A. v. reducta Nebraska, Minnesota, Wisconsin, Michigan, Chamberlin in Oklahoma, where it is known and north of Sault Ste. Marie, Ontario, its dis- from McCurtain County (Causey 1954, McAl- tribution appears to be greater than that of lister et al. 2002b). The exact site in Latimer Apheloria virginiensis (Drury) (Polydesmida: County is unknown, but we suspect the east- Xystodesmidae), based on range descriptions ern periphery because of the more optimal of the latter’s subspecies by Hoffman (1999), habitat and more substantial deciduous forests 116 WESTERN NORTH AMERICAN NATURALIST [Volume 65 there, as S. granulatus prefers moist, deciduous and western Arkansas, S. granulatus can also litter. This area of Latimer County and sites in be projected for northern Mississippi, whose Shawnee County, Kansas, and Cass County, diplopod fauna, though east of the river, is also Nebraska, are at roughly the same longitude poorly known. and constitute the western range limits for both S. granulatus and the generic distribution ACKNOWLEDGMENTS in eastern North America. Other range extremes are in Washington County, Minnesota; Algoma We thank the following professors, curators, County, Ontario; Nicolet County, Québec; and collection managers for providing access Windsor County, Vermont; Essex County, New to or loaning specimens to the 1st author: N.I. Jersey; Queene Anne’s County, Maryland; Dare Platnick (AMNH), D. Azuma (ANSP), G.B. County, North Carolina; and Orangeburg Edwards (FSCA), J.K. Barnes (UAAM), and County, South Carolina (Shelley 1988, 1993). R.W. Sites (UMO). The 1st author’s travel to As Shelley (1993) did not provide locality the FSCA in 2002, which resulted in discovery details for states in which S. granulatus occurs of the Jackson County, Alabama, and Latimer in more than 5 counties, none were given for County, Oklahoma, samples, was sponsored by a grant from the Center for Systematic Ento- North Carolina. He mentioned that the mil- mology. The 2nd author’s fieldwork at the liped occurs on the Outer Banks, as far east in LeFlore County, Oklahoma, site was supported the state as one can go, which is striking be- in part by TAMU-T Faculty Senate Research cause few native millipeds can survive in the Enhancement grant 200900. small wooded patches on these islands, which are essentially just long, narrow sand bars; LITERATURE CITED sample data are Dare Co., Hatteras Island, Buxton, 4, 19 November 1980, D.L. Stephan BOLLMAN, C.H. 1893. The Myriapoda of North America. (NCSM). United States National Museum Bulletin 46: 1–210. The new Oklahoma samples constitute a CAUSEY, N.B. 1954. Three new species and new records of southern millipeds. Tulane Studies in Zoology 2(4): westward range expansion of at least 255 miles 63–68. (408 km) from Jonesboro, Arkansas, and are CHAMBERLIN, R.V. 1928. Some chilopods and diplopods also well to the south of the latitude of this from Missouri. Entomological News 39:153–155. city, thereby forming the southern range limit ______. 1942. On a collection of myriopods from Iowa. along with Orangeburg, Orangeburg County, Canadian Entomologist 74:15–17. CHAMBERLIN, R.V., AND R.L. HOFFMAN. 1958. Checklist of South Carolina. The projected distribution the millipeds of North America. United States Nation- spans all 5 Great Lakes and such major rivers al Museum Bulletin 212:1–236. as the Arkansas, Mississippi, Missouri, Ohio, GUNTHORP, H. 1913. Annotated list of the Diplopoda and St. Lawrence, Hudson, Delaware, Potomac, Chilopoda, with a key to the Myriapoda of Kansas. Kansas University Science Bulletin 7:161–182. James, Roanoke, Cape Fear, Tennessee, and HOFFMAN, R.L. 1962a. A new genus and species in the Cumberland. It encompasses parts or all of 13 diplopod family Nearctodesmidae from Illinois (Poly- physiographic provinces and covers nearly the desmida). American Midland Naturalist 68:192–198. entire generic area east of the Central Plains, ______. 1962b. The milliped genus Scytonotus in eastern North America, with the description of two new excepting the aforementioned absence from the species. American Midland Naturalist 67:241–249. Blue Ridge Province, occupied by S. virginicus ______. 1999. Checklist of the millipeds of North and Loomis and S. australis Hoffman (Fig. 2). Shelley Middle America. Virginia Museum of Natural His- (1993) concluded that these species are young, tory Special Publication 8:1–584. KEETON, W.T. 1960. A taxonomic study of the milliped fam- derivative entities that have displaced S. gran- ily Spirobolidae (Diplopoda: Spiobolida). Memoirs ulatus from these mountains and the western of the American Entomological Society 17:1–146. periphery of the adjacent Piedmont Plateau in KENYON, F.C. 1893. A preliminary list of the Myriapoda of the Carolinas and Georgia. Otherwise, the pro- Nebraska, with descriptions of new species. Publica- jected range is cohesive, encompassing around tions of the Nebraska Academy of Science 3:14–18. MCALLISTER, C.T., C.S. HARRIS, R.M. SHELLEY, AND J.T. 1100 miles (1760 km) east–west and 985 miles MCALLISTER III. 2002a. Millipeds (Arthropoda: (1576 km) north–south and parts of 2 Cana- Diplopoda) of the Ark-La-Tex. I. New distributional dian provinces and 19 states, 6 of which— and state records for seven counties of the west Gulf Pennsylvania, West Virginia, Kentucky, Ohio, Coastal Plain of Arkansas. Journal of the Arkansas Academy of Science 56:91–94. Indiana, and Illinois—lie wholly within the MCALLISTER, C.T., R.M. SHELLEY, AND J.T. MCALLISTER range. With its discovery in Alabama, Oklahoma, III. 2002b. Millipeds (Arthropoda: Diplopoda) of 2005] DISTRIBUTION OF SCYTONOTUS GRANULATUS 117

Ark-La-Tex. II. Distributional records for some and diplopod faunistic studies. Canadian Journal of species of western and central Arkansas and eastern Zoology 80:1863–1875. and southeastern Oklahoma. Journal of the Arkansas ______. 2002b. The milliped genus Oriulus Chamberlin Academy of Science 56:95–98. (Julida: Parajulidae). Canadian Journal of Zoology MCALLISTER, C.T., R.M. SHELLEY, AND D.I. MOORE. 2003. 80:100–109. Noteworthy records of the millipeds, Eurymerodes- ______. 2003a (2002). Narceus woodruffi Causey, a forgot- mus angularis and E. mundus (Polydesmida: Eurymer- ten milliped species (Spirobolida: Spirobolidae). odesmidae), from northeastern and westcentral Texas. Insecta Mundi 16:25–29. Texas Journal of Science 56:73–77. ______. 2003b. A new polydesmid milliped genus and two SHELLEY, R.M. 1978. Millipeds of the eastern Piedmont new species from Oregon and Washington, U.S.A., region of North Carolina, U.S.A. (Diplopoda). Jour- with a review of Bidentogon Buckett and Gardner, nal of Natural History 12:37–79. 1968, and a summary of the family in western North ______. 1980. Revision of the milliped genus Pleuroloma America (Polydesmida: Polydesmidae). Zootaxa 296: (Polydesmida: Xystodesmidae). Canadian Journal of 1–12. Zoology 58:129–168. SHELLEY, R.M., C.T. MCALLISTER, AND J.L. HOLLIS. 2003. ______. 1988. The millipeds of eastern Canada (Arthro- A new milliped of the genus Abacion Rafinesque, poda: Diplopoda). Canadian Journal of Zoology 66: 1820, from Arkansas, U.S.A. (Callipodida: Abacion- 1638–1663. idae). Zootaxa 170:1–7. ______. 1993. Revision of the milliped genus Scytonotus SHELLEY, R.M., C.T. MCALLISTER, AND S.B. SMITH. 2004 Koch (Polydesmida: Polydesmidae). Brimleyana 19: (2003). Discovery of the milliped Pleuroloma flavipes 1–60. in Texas, and other records from west of the Missis- ______. 1994. The milliped family Nearctodesmidae in sippi River (Polydesmida: Xystodesmidae). Entomo- northwestern North America, with accounts of Sako- logical News 114:2–6. phallus and S. simplex Chamberlin (Polydesmida). WHITNEY, C.L., AND R.M. SHELLEY. 1995. Occurrence of Canadian Journal of Zoology 72:470–495. the milliped Ergodesmus compactus Chamberlin in ______. 2001 (2000). Annotated checklist of the millipeds Canada (Polydesmida: Nearctodesmidae). Insecta of Florida (Arthropoda: Diplopoda). Insecta Mundi Mundi 9:277–278. 14(4):241–251. ______. 2002a. The millipeds of central Canada (Arthro- Received 30 December 2003 poda: Diplopoda), with reviews of the Canadian fauna Accepted 20 July 2004 Western North American Naturalist 65(1), © 2005, pp. 118–122

WILLOW THICKETS PROTECT YOUNG ASPEN FROM ELK BROWSING AFTER WOLF REINTRODUCTION

William J. Ripple1,2 and Robert L. Beschta1

Key words: aspen, willow, elk, herbivory, wolves, predation risk, Yellowstone.

Aspen (Populus tremuloides) is the most (Ripple et al. 2001). Existing riparian areas are widely distributed deciduous tree in North dominated by sedge (Carex spp.) and grass- America (Bartos 2001). However, long-term dominated meadows with patches of willow assessments of its status indicate a decline in (Salix spp.). abundance throughout the western United Ripple and Larsen (2000) discovered that States during much of the 20th century (Kay the decline of aspen began with the extirpa- 1997, Bartos and Campbell 1998, Bartos 2001). tion of wolves (Canis lupus) from Yellowstone In Yellowstone National Park, aspen decline National Park in the 1920s. They hypothesized has been noted on the “northern range,” the that the failure of aspen to reach tree height wintering grounds for the park’s largest elk over the last half century may be due to changes (Cervus elaphus) herd. As one of the principal in northern range trophic structure involving deciduous woody species found in Yellowstone the gray wolf, elk, and elk herbivory on aspen. National Park, aspen contributes to ecological Following wolf extirpation from Yellowstone diversity by providing habitat for numerous National Park, aspen sprouts and other decid- vertebrate and invertebrate species, supports uous woody species were browsed more heav- a variety of plant associations, provides browse ily, changing their growth form dramatically for ungulates, and has aesthetic appeal for and sometimes increasing mortality (Singer et park visitors. al. 1998, Barmore 2003). In recent decades Concern about the loss of aspen on the neither willow, aspen, nor cottonwood (Popu- northern range began in the 1920s and has lus spp.) has been able to grow above the been the subject of debate and research ever browsing level of elk, resulting in low stature since (Warren 1926, National Research Coun- throughout most of the northern range (Bar- cil 2002). Most research has attributed the park’s more 2003, Beschta 2003, Larsen and Ripple aspen decline primarily to long-term browsing 2003). pressure by elk (see National Research Coun- The elk of Yellowstone National Park lived cil 2002 for review). Influences such as fluctu- in an environment free of wolves for approxi- ations in climate, altered fire regimes, and mately 7 decades, from the mid-1920s until conifer invasions have been identified as poten- 1995, when wolves were reintroduced. By the tially contributing to aspen decline (Romme end of 2001, nearly 80 wolves lived on the et al. 1995, Yellowstone National Park 1997, northern range of Yellowstone National Park Meagher and Houston 1998). (Smith et al. 2003). In recent years, following Vegetation on the northern range consists wolf reintroduction, elk have altered their of sagebrush (Artemisia spp.) steppe, primarily movements and foraging patterns to minimize big sagebrush (A. tridentata), and grassland their risk of being preyed upon by wolves (Rip- interspersed with small stands of trees, pri- ple et al. 2001, Ripple and Beschta 2003). Some marily Douglas-fir (Pseudotsuga menziesii) and young willow and cottonwood have been aspen (Despain 1990). Aspen covers approxi- growing taller along various stream reaches in mately 1% of the northern range landscape recent years in northern Yellowstone National

1College of Forestry, Oregon State University, Corvallis, OR 97331. 2Corresponding author.

118 2005] NOTES 119

Park (Ripple and Beschta 2003), though sup- intensity we determined whether each annual pressed aspen sprouts generally have not been leader segment had been browsed for each of able to escape browsing (Smith et al. 2003). the last 3 years. Field observations in late summer 2003 Confidence intervals (±95% CI) were plot- indicated that young aspens are growing taller ted to assess significant differences between within willow clumps along parts of Crystal height of aspen sprouts in the willow and Creek near its confluence with the Lamar height of adjacent aspen sprouts in the open. River. Linked mechanisms for why (1) willow Linear regression was used to test the null is growing taller and (2) aspen in willow is hypothesis of no relationship between height growing taller include the following: of willow and height of aspen sprouts growing in the same willow. 1. Increased predation risk following the re- Along the 500-m transect we found 16 young introduction of wolves caused less brows- aspen plants growing within clumps of 3 species ing by elk and the growth of taller willow of tall willow, including Bebb willow (S. beb- plants (Ripple and Beschta 2003). Taller biana), Booth willow (S. boothii), and Geyer and denser willow thickets contributed to willow (S. geyeriana). Heights measured dur- increased predation risk, thus further re- ing late summer 2003, after the summer grow- ducing browse intensity (Ripple and Larsen ing period and before winter browsing, indi- 2000, White et al. 2003). cated that aspen growing in willow clumps 2. Willow provides physical protection from were significantly taller (x– = 180 cm, 95% CI browsing for young aspen stems as well as ±19.7 cm) than aspen growing in the open (x– visual protection, making the stems often = 104 cm, 95% CI ±13.7 cm), but not signifi- indistinguishable from those of willow cantly different from their protecting willows (“safety in numbers”). (x– = 181 cm, 95% CI ±17.9 cm). We found a strong linear relationship between willow Our objective was to test the hypothesis height and height of aspen growing within the that taller willow represents a mechanism for same willow clumps (r2 = 0.71, n = 16, P < aiding aspen tree growth in an ungulate win- 0.01; Fig. 2). ter range by addressing 2 questions: (1) Are In addition, aspen stems growing inside aspen within willow clumps growing differ- willow clumps were significantly taller than ently from those growing in adjacent but rela- aspen growing in the open adjacent to willow tively open areas? (2) Is there a direct positive clumps in 2000, 2001, and 2002 (Fig. 3). We relationship between heights of aspen sprouts found no significant differences in willow growing in willow and heights of the willow? heights and corresponding aspen heights in In August 2003 we searched for aspen grow- willow clumps for 2001 and 2002 (Fig. 3). ing in willow clumps along a 500-m reach of Leaders browsed each year between 2000 and Crystal Creek. Whenever we encountered aspen 2002 ranged from 81% to 100% for willow, sprouts within willow clumps, we obtained plant 73% to 94% for aspen growing in willow, and height measurements on (1) the aspen plant 93% to 100% for aspen growing in the open. within the willow clump, (2) the tallest willow These results suggest that, on sites suitable leader in the same clump, and (3) the nearest for both willow and aspen regeneration, tall aspen growing outside the willow clump <5 m willows can play a supportive role in aspen away (Fig. 1). Plant measurements were used recruitment success by providing protection to evaluate the recent history of plant heights from ungulate browsing. These findings are and browsing levels (Keigley and Frisina 1998, similar to those of Ripple and Larsen (2001), Keigley et al. 2003). Since browsing usually who determined that aspen sprouts were removes the terminal bud, causing growth to escaping severe browsing and growing taller emerge from a lateral bud, the stems grow in a where dead conifer trees had recently fallen zigzag pattern, leaving behind stubs represent- and created protective “jackstrawed” barriers ing annual segments that can be measured to elk movement. The multi-stemmed configu- (Keigley et al. 2003). Thus, to assess plant growth ration of willows probably contributes to their history, we measured the height of annual ter- effectiveness in protecting aspen sprouts from minal bud scars (or annual segments) for the browsing. It also appears in at least some cases previous 3 years (2000–2002). For browsing that the annual growth rates of aspen leaders 120 WESTERN NORTH AMERICAN NATURALIST [Volume 65

Fig. 1. Photograph showing tall willow and aspen on Crystal Creek in August 2003. The maximum height of young aspen leaders here was 238 cm (above the subject’s right hand), the tallest willow stem adjacent to the aspen was 222 cm, and the aspen leader outside the willow clump measured 125 cm (in front and to the right of the subject). These willows and aspen appear to be growing much taller since wolf reintroductions of 1995/1996.

equal or exceed those of willow, increasing the et al. 2001). With wolves in the system, elk may likelihood that these aspen may grow into be spending less time browsing on individual large-stemmed trees and not be overtopped plants and removing a smaller proportion of by willows. the current year’s growth in areas with high We hypothesize that willows along Crystal levels of predation risk (Ripple and Beschta Creek are growing taller because the propor- 2003). This would account for the taller plant tion of the current year’s growth being con- growth even though browsing levels are rela- sumed by elk is less than it was before wolf tively high as shown by the percentage of reintroduction, due to increases in predation leaders browsed annually. risk and associated changes in elk foraging At the landscape scale, willow growing in behavior in the presence of wolves (Laundré valley bottoms may be browsed less since elk 2005] NOTES 121

300 175 Willow y = 0.92x + 13.05 Aspen IN 2 Aspen OUT 250 R = 0.71 150 125

200 100 75

150 50

Leader Height (cm) 25 2003 Aspen Height (cm)

100 0 100 150 200 250 300 2000 2001 2002

2003 Willow Height (cm) Year

Fig. 3. Heights (means ±95% confidence intervals) of Fig. 2. Relationship between late summer 2003 heights willow and aspen leaders after winter browsing in 2000, of willow and aspen growing within the same willow 2001, and 2002 based on plant measurements along Crys- clumps in Yellowstone National Park. This regression sug- tal Creek. Data include leader height of aspen in willow gests that willows on these sites are providing protection clumps, height of the tallest adjacent willow leaders, and to aspen from browsing by elk. height of the aspen outside the willow clumps. may be avoiding certain riparian areas and selecting for higher ground to lower their risk the wolf reintroductions of 1995 and 1996. Our of predation by wolves (Ripple and Beschta paired willow/aspen height measurements indi- 2003). In contrast to many riparian areas, up- cate that willow is able to withstand more lands may provide elk with a lower risk of pre- browsing than aspen. The spatial extent of this dation, better escape terrain, and fewer escape process whereby tall willows provide brows- impediments (Bibikov 1982, Kunkel and Plet- ing protection for aspen is unknown at this scher 2000, 2001). For example, Bergman (2003) time, but we do not expect large stands of found an inverse correlation between distance mature aspen developing from this process from streams and successful wolf kills in cen- since the occurrence of aspen in willow is not tral Yellowstone National Park. Likewise, Gula common in the study area. Furthermore, while (2004), while studying wolves and ungulates in the increased height growth of young aspen Poland, found that riparian terrain features may indicate that some recovery of aspen is appeared to be important for hunting strate- underway in the northern range, it is still too gies used by wolves. He discovered that wolves early to know if these aspen sprouts, averaging made most kills (74%) in ravines and creeks 180 cm tall in the late summer 2003, will be where ungulates (mostly elk) may be easier to able to overcome browsing by elk, will grow intercept as they slow down and change their fast enough to not be overtopped by willows, gait. Conversely, this same process of elk and will eventually grow to tree height. avoiding riparian areas could be causing high browsing pressure on upland aspen stands. LITERATURE CITED We know of no current aspen recruitment in the uplands on the northern range within Yel- BARMORE,W.J. 2003. Ecology of ungulates and their win- lowstone National Park. ter range in northern Yellowstone National Park: Willow typically grows as a shrub with many research and synthesis 1962–1970. Yellowstone Cen- leaders, while aspen tends to have a single ter- ter for Resources, Yellowstone National Park, WY. minal leader that is more susceptible to herbi- 528 pp. BARTOS,D.L. 2001. Landscape dynamics of aspen and vory by ungulates. This difference in growth conifer forests. Pages 5–14 in Sustaining aspen in form could account for the greater growth of western landscapes: symposium proceedings. USDA willow than aspen in recent years following Forest Service, RMRS-P-18, Fort Collins, CO. 122 WESTERN NORTH AMERICAN NATURALIST [Volume 65

BARTOS, D.L., AND R.B. CAMPBELL, JR. 1998. Decline of MEAGHER, M.M., AND D.B. HOUSTON. 1998. Yellowstone quaking aspen in the Interior West—examples from and the biology of time. Oklahoma State University Utah. Rangelands 20:17–24. Press, Norman. BERGMAN, E. 2003. Assessment of prey vulnerability NATIONAL RESEARCH COUNCIL. 2002. Ecological dynamics through analysis of wolf movements and kill sites. on Yellowstone’s northern range. National Academy Master’s thesis, Montana State University, Bozeman. Press, Washington, DC. 50 pp. RIPPLE, W.J., AND R.L. BESCHTA. 2003. Wolf reintroduction, BESCHTA, R.L. 2003. Cottonwoods, elk, and wolves in the predation risk, and cottonwood recovery in Yellow- Lamar Valley of Yellowstone National Park. Ecologi- stone National Park. Forest Ecology and Manage- cal Applications 13:1295–1309. ment 184:299–313. BIBIKOV, D.I. 1982. Wolf ecology and management in the RIPPLE, W.J., AND E.J. LARSEN. 2000. Historic aspen re- USSR. Pages 120–133 in F. J. Harrington and P.C. cruitment, elk, and wolves in northern Yellowstone Paquet, editors, Wolves of the world: perspectives of National Park, USA. Biological Conservation 95: behavior, ecology, and conservation. Noyes Publica- 361–370. tions, Park Ridge, NJ. ______. 2001. The role of post fire coarse woody debris in DESPAIN, D.G. 1990. Yellowstone vegetation: consequences aspen regeneration. Western Journal of Applied of environment and history in a natural setting. Forestry 16:61–64. Roberts Rinehart, Boulder, CO. RIPPLE, W.J., E.J. LARSEN, R.A. RENKIN, AND D.W. SMITH. GULA, R. 2004. Influence of snow cover on wolf Canis 2001. Trophic cascades among wolves, elk, and aspen lupus predation patterns in Bieszczady Mountains, on Yellowstone National Park’s northern range. Bio- Poland. Wildlife Biology 10:17–23. logical Conservation 102:227–234. KAY, C.E. 1997. Is aspen doomed? Journal of Forestry ROMME, W.H., M.G. TURNER, L.L. WALLACE, AND J.S. 95:4–11. WALKER. 1995. Aspen, elk, fire in northern Yellow- KEIGLEY, R.B., AND M.R. FRISINA. 1998. Browse evalua- stone National Park. Ecology 76:2097–2106. tion by analysis of growth form. Montana Fish, Wild- SINGER, F.J., L.C. ZEIGENFUSS, R.G. CATES, AND D.T. BAR- life, and Parks, Helena. NETT. 1998. Elk, multiple factors, and persistence of KEIGLEY, R.B., M.R. FRISINA, AND C. FAGER. 2003. A willows in national parks. Wildlife Society Bulletin method for determining the onset year of intense 26:419–428. browsing. Journal of Range Management 56:33–38. SMITH, D.W., R.O. PETERSON, AND D.B. HOUSTON. 2003. KUNKEL, K.E., AND D.H. PLETSCHER. 2000. Habitat fac- Yellowstone after wolves. Bioscience 53:330–340. tors affecting vulnerability of moose to predation by WARREN, E.R. 1926. A study of beaver in the Yancey wolves in southeastern British Columbia. Canadian region of Yellowstone National Park. Roosevelt Journal of Zoology 78:150–157. Wildlife Annual 1:1–191. ______. 2001. Winter hunting patterns of wolves in and WHITE, C.A., M.C. FELLER, AND S. BAYLEY. 2003. Preda- near Glacier National Park, Montana. Journal of tion risk and the functional response of elk-aspen Wildlife Management 65:520–530. herbivory. Forest Ecology and Management 181:77–97. LARSEN, E.J., AND W. J. R IPPLE. 2003. Aspen age structure YELLOWSTONE NATIONAL PARK. 1997. Yellowstone’s north- in the northern Yellowstone ecosystem: USA. Forest ern range: complexity and change in a wildland Ecology and Management 179:469–482. ecosystem. U.S. Department of the Interior, Yellow- LAUNDRÉ, J.W., L. HERNANDEZ, AND K.B. ALTENDORF. stone National Park, Mammoth Hot Springs, WY. 2001. Wolves, elk, and bison: reestablishing the “landscape of fear” in Yellowstone National Park, U.S.A. Received 12 January 2004 Canadian Journal of Zoology 79:1401–1409. Accepted 22 June 2004 Western North American Naturalist 65(1), © 2005, pp. 123–126

SPRUCE BEETLE (COLEOPTERA: SCOLYTIDAE) RESPONSE TO TRAPS BAITED WITH SELECTED SEMIOCHEMICALS IN UTAH

Darrell W. Ross1, Gary E. Daterman2, and A. Steven Munson3

Key words: Dendroctonus rufipennis, spruce beetle, pheromones, trapping, frontalin, seudenol, MCOL, ethanol, α-pinene, Scolytidae.

Spruce beetle, Dendroctonus rufipennis demonstrated that the addition of 1-methyl-2- (Kirby), populations periodically reach outbreak cyclohexen-1-ol (MCOL) to the 2-component densities throughout the range of spruce, Picea lure could significantly increase number of spp., in western North America. During out- spruce beetles captured in traps at some loca- breaks it may kill thousands to millions of tions (Werner 1994, Borden et al. 1996, Setter trees over vast areas, dramatically altering for- and Borden 1999). However, in 1 of these est structure, composition, and ecological pro- studies, spruce beetle response to racemic cesses, thus impacting a variety of resource MCOL and the 2 enantiomers was not consis- values (Schmid and Frye 1977, Veblen et al. tent at 5 different locations in Alaska, British 1991, Holsten et al. 1995, 1999, Ross et al. Columbia, and Alberta (Borden et al. 1996). 2001). Current options for reducing negative This latter study demonstrated the need to impacts on resource values caused by the understand the response of spruce beetles to spruce beetle include harvesting high-risk trees, semiochemicals throughout their range. sanitation/salvage logging, insecticide applica- Recently the spruce beetle has been at out- tions to high-value trees, felling and removal break densities in central and southern Utah of trap trees, lethal trap trees, burying infested (USDA Forest Service 2003). No published material, and burning or removing bark of reports exist that compare different semio- infested trees (Holsten et al. 1999). Resource chemicals as trap lures for spruce beetle in the managers have used mass trapping with semio- Southern Rocky Mountains. Resource managers chemical-baited traps to some extent, but fur- who want to use semiochemical-baited traps ther research is needed to make this treatment to reduce the impact of the spruce beetle in more effective. An attempt to mass trap spruce this region need the most attractive lure possi- beetle in Alaska reduced the number of subse- ble to have maximum effect on spruce beetle quent beetle-attacked trees compared with populations. The objective of this project was untreated control plots, but beetle catches to compare several different trap lures con- were considered low relative to beetle popula- taining frontalin and 1 or 2 of the following tions in the area (Werner and Holsten 1995). compounds: α-pinene, MCOL, 3-methyl-2-cy- In Utah semiochemical-baited traps were used clohexen-1-ol (seudenol), and ethanol. Specifi- in conjunction with felling and burning of cally, we wanted to determine whether the infested trees and trap trees to effectively sup- spruce beetle would respond to the addition of press an isolated spruce beetle population MCOL to the standard 2-component lure of (Bentz and Munson 2000). frontalin and α-pinene in the Southern Rocky Initially, the standard trap lure for spruce Mountains. Additionally, we wanted to com- beetle was composed of 1,5-dimethyl-6,8-dioxa- pare several novel 2- and 3-component lures biocyclo[3.2.1]octane (frontalin) and α-pinene, to the standard lures. Seudenol was included and this lure is still available commercially in the experiment because it is known to be (Phero Tech, Inc., Delta, BC, Canada, and IPM attractive to spruce beetles in other parts of Tech, Inc., Portland, OR). Subsequent studies their range (Furniss et al. 1976, Dyer and Lawko

1Department of Forest Science, Oregon State University, Corvallis, OR 97331. 2USDA Forest Service, Pacific Northwest Research Station, 3200 Jefferson Way, Corvallis, OR 97331. 3USDA Forest Service, Forest Health Protection, 4746 South 1900 East, Ogden, UT 84403.

123 124 WESTERN NORTH AMERICAN NATURALIST [Volume 64

1978, Dyer and Hall 1980), and ethanol was ported. All statistical analyses were performed included because it has been shown to induce with the SAS System for Windows Release colonization by spruce beetle when applied to 8.02 (SAS Institute, Inc., Cary, NC). host trees (Moeck 1981), and it is known to be Due to the low number of spruce beetles attractive to other scolytids (Ross and Dater- collected on 26 June and the large number of man 1995, Kelsey and Joseph 2001). missing samples on 16 July, data from those An experiment was installed on 23 June dates were excluded from the analyses. A total 1998 at the Cedar City Ranger District of the of 3858 spruce beetles were collected across Dixie National Forest in southwestern Utah all treatments and collection dates that were (37°39′N, 112°45′W). The study included 6 used in the analyses. The effect of trap lure replications of 8 treatments. The treatments composition on spruce beetle catches was high- were 16-unit, multiple-funnel traps (Lindgren ly significant (F7,35 = 3.48, P < 0.0062). The 2 1983) baited with frontalin combined with 1 lures that contained MCOL caught the largest or 2 of the other semiochemicals (Table 1). numbers of spruce beetles (Table 1). The lure α Frontalin, seudenol, and -pinene were for- composed of frontalin, α-pinene, and MCOL mulated in polyvinylchloride in our laboratory caught significantly more beetles than any other (Daterman 1974). MCOL and ethanol were lure except the one containing frontalin and formulated in bubble capsules and plastic MCOL. The lure containing frontalin and pouches, respectively, and were purchased from MCOL produced the 2nd highest catch, which Phero Tech, Inc. Chemicals with chiral cen- was significantly higher than all other lures ex- ters were tested as racemates. Chemical puri- cept the combinations of frontalin and seudenol; ties of semiochemicals were as follows: fronta- frontalin, α-pinene, and ethanol; and frontalin, lin, 95.0%; seudenol, 99.3%; MCOL, ≥98%; ≥ α ≥ ethanol, and seudenol. Only 12 T. undatulus ethanol, 98%; -pinene, 98%. Semiochemi- were collected. cal release rates determined by weight loss at We sorted 3241 spruce beetles according to 25°C over a 2-week period in mg ⋅ day–1 were gender and determined that the effect of trap as follows: frontalin, 3.3; seudenol, 4.2; MCOL, lure composition on percentage male beetles 4.0; ethanol, 88; α-pinene, 1.6. Within a repli- in the catches was not significant (F = 1.08, cate, traps were spaced 50–100 m apart and 7,35 P < 0.3972). Mean percentage male beetles for replicates were at least 100 m apart. Traps were placed in stand openings of various sizes all treatments ranged from 47% to 67 %. to keep them as far as possible from host trees A 2nd experiment was installed on 3 April to minimize potential “spillover” beetle attacks 2001 at the Heber Ranger District of the Uinta National Forest in central Utah (40°32′N, on nearby trees. All replicates were located ° ′ within a 16-km2 area that included numerous 111 01 W). The study included 10 pairs of 16- spruce beetle–infested trees. Traps were emp- unit, multiple-funnel traps spaced about 30 m tied on 26–29 June and 1, 3, and 16 July 1998. apart. Traps were placed in forested sites at Spruce beetles and an associated predator, least 10 m from potential host trees, with pairs Thanasimus undatulus Say (Coleoptera: Cleri- of traps at least 500 m apart. One trap in each α dae), were counted in all samples. A subsam- pair was baited with frontalin and -pinene and α ple of 100 spruce beetles or the entire sample, the other trap with frontalin, -pinene, and if it contained less than 100 spruce beetles, MCOL. Trap lures were purchased from Phero was separated according to gender based on Tech, Inc. All semiochemicals were ≥98% pure, characteristics of the elytral declivity (Wood and release rates in mg ⋅ day–1 were as follows: 1982); percentage male composition was de- frontalin, 2.6 at 23°C; α-pinene, 1.5 at 20°C; termined. Beetle catches for each trap were MCOL, 2.0 at 20°C. Eight pairs of traps were summed across all collection dates and trans- emptied on 5, 18, and 26 June, 5, 12, 18, and formed by ln(Y+1) to correct for hetero- 24 July, and 6 August 2001. The other 2 pairs scedasticity. Transformed beetle catch and per- of traps were less accessible and were emptied centage male composition data were subjected on 17 and 19 June, 2, 10, and 30 July 2001. We to analysis of variance (ANOVA) for a random- counted spruce beetles collected in each sam- ized complete block design. Means were com- ple and summed beetle catches for each trap pared and separated by Fisher’s protected LSD across all collection dates. We subjected total when P < 0.05. Nontransformed means are re- spruce beetle catches to ANOVA. 2004] NOTES 125

TABLE 1. Mean numbers (± s) of spruce beetles caught in District, Uinta National Forest, for providing multiple-funnel traps baited with selected semiochemicals technical assistance and access to the study in southern Utah. sites. We also thank Kimberly Wallin and John Lure composition Spruce beetles Borden for reviewing an earlier draft of the (number per trap)a manuscript. This article reports the results of Frontalin and α-pinene 27.3 ± 6.2 c research only. Mention of a proprietary prod- Frontalin and ethanol 29.5 ± 11.4 c uct does not constitute an endorsement or rec- Frontalin and seudenol 42.5 ± 20.6 bc ommendation for its use by USDA. The work Frontalin and MCOL 154.8 ± 95.3 ab Frontalin, α-pinene, and upon which this publication is based was funded ethanol 72.8 ± 37.9 bc in whole or in part through a grant awarded Frontalin, α-pinene, and by the Northeastern Area State and Private seudenol 22.8 ± 4.8 c Forestry, USDA Forest Service. Frontalin, α-pinene, and MCOL 236.0 ± 85.5 a ITERATURE ITED Frontalin, ethanol, and L C seudenol 57.2 ± 16.5 bc BENTZ, B.J., AND A.S. MUNSON. 2000. Spruce beetle popu- aMeans followed by the same letter are not significantly different (Fisher’s lation suppression in northern Utah. Western Jour- protected LSD, P < 0.05). nal of Applied Forestry 15:122–128. BORDEN, J.H., G. GRIES, L.J. CHONG, R.A. WERNER, E.H. HOLSTEN, H. WIESER, E.A. DIXON, AND H.F. CEREZKE. 1996. Regionally specific bioactivity of two new pher- A total of 9749 spruce beetles were col- omones for Dendroctonus rufipennis (Kirby) (Col., lected across both treatments and all collec- Scolytidae). Journal of Applied Entomology 120: 321–326. ± – tion dates. The mean ( sx) total number of DATERMAN, G.E. 1974. Synthetic sex pheromone for de- spruce beetles collected in traps baited with tection survey of European pine shoot moth. Research frontalin, α-pinene, and MCOL was 611 ± 82 Paper PNW-180, Pacific Northwest Forest and compared with 364 ± 91 in traps baited with Range Experiment Station, USDA Forest Service, just frontalin and α-pinene. The difference was Portland, OR. DYER, E.D.A., AND P. M . H ALL. 1980. Effect of the living highly significant (F1,9 = 10.42, P < 0.0103). host tree (Picea) on the response of Dendroctonus These results are consistent with those from rufipennis (Coleoptera: Scolytidae) and a predator Alaska, northwestern Alberta, and southeast- Thanasimus undatulus (Coleoptera: Cleridae) to fron- ern British Columbia, where racemic MCOL talin and seudenol. Canadian Entomologist 112: 167–171. increased catches of spruce beetles to varying DYER, E.D.A., AND C.M. LAWKO. 1978. Effect of seudenol degrees (Borden et al. 1996, Setter and Borden on spruce beetle and Douglas-fir beetle aggregation. 1999). The present study extends the range for Fisheries and Environment Canada Forestry Service which racemic MCOL appears to be attractive Bi-monthly Research Notes 34:30–32. FURNISS, M.M., B.H. BAKER, AND B.B. HOSTETLER. 1976. to the spruce beetle to include all sites that Aggregation of spruce beetles (Coleoptera) to seude- have been tested within and east of the Rocky nol and repression of attraction by methylcyclo- Mountains. These results indicate that in the hexenone in Alaska. Canadian Entomologist 108: Southern Rocky Mountains, addition of racemic 1297–1302. MCOL to spruce beetle lures containing fron- HOLSTEN, E.H., R.W. THIER, A.S. MUNSON, AND K.E. GIB- α SON. 1999. The spruce beetle. Forest Insect and Dis- talin and -pinene will significantly increase ease Leaflet 127, USDA Forest Service, Washington, catches. Furthermore, lures composed of fron- DC. talin and MCOL will likely be more effective HOLSTEN, E.H., R.A. WERNER, AND R.L. DEVELICE. 1995. in trapping spruce beetles than lures com- Effects of a spruce beetle (Coleoptera: Scolytidae) α outbreak and fire on Lutz spruce in Alaska. Environ- posed of frontalin and -pinene. However, mental Entomology 24: 1539–1547. other combinations of frontalin and MCOL KELSEY, R.G., AND G. JOSEPH. 2001. Attraction of Scolytus with seudenol or ethanol should be tested as unispinosus bark beetles to ethanol in water-stressed well as different ratios of the various semio- Douglas-fir branches. Forest Ecology and Manage- chemicals to determine if a more attractive ment 144:229–238. LINDGREN, B.S. 1983. A multiple funnel trap for scolytid lure can be developed. beetles (Coleoptera). Canadian Entomologist 115: 299–302. We thank Ron Wilson, Phil Eisenhauer, and MOECK, H.A. 1981. Ethanol induces attack on trees by Lucy Wilkins of the Cedar City Ranger Dis- spruce beetles, Dendroctonus rufipennis (Coleop- tera: Scolytidae). Canadian Entomologist 113:939–942. trict, Dixie National Forest, and Bob McPhie, ROSS, D.W., AND G.E. DATERMAN. 1995. Response of Den- Doug Page, and Lew Giles of the Heber Ranger droctonus pseudotsugae (Coleoptera: Scolytidae) and 126 WESTERN NORTH AMERICAN NATURALIST [Volume 64

Thanasimus undatulus (Coleoptera: Cleridae) to traps WERNER, R.A. 1994. Research on the use of semiochemi- with different semiochemicals. Journal of Economic cals to manage spruce beetles in Alaska. Pages 15–21 Entomology 88:106–111. in P. J. Shea, technical coordinator, Proceedings of ROSS, D.W., G.E. DATERMAN, J.L. BOUGHTON, AND T.M. the symposium on management of western bark bee- QUIGLEY. 2001. Forest health restoration in south- tles with pheromones: research and development. central Alaska: a problem analysis. General Techni- General Technical Report PSW-GTR-150, USDA cal Report PNW-GTR-523, USDA Forest Service, Forest Service, Pacific Southwest Research Station, Pacific Northwest Research Station, Portland, OR. Albany, CA. SCHMID, J.M., AND R.H. FRYE. 1977. Spruce beetle in the WERNER, R.A., AND R.H. HOLSTEN. 1995. Current status Rockies. General Technical Report RM-49, USDA of research with the spruce beetle, Dendroctonus Forest Service, Rocky Mountain Forest and Range rufipennis. Pages 23–29 in S.M. Salom and K.R. Experiment Station, Fort Collins, CO. Hobson, technical editors, Application of semio- SETTER, R.R., AND J.H. BORDEN. 1999. Bioactivity and effi- chemicals for management of bark beetle infesta- cacy of MCOL and seudenol as potential attractive tions—proceedings of an informal conference. Gen- bait components for Dendroctonus rufipennis (Coleop- eral Technical Report INT-GTR-318, USDA Forest tera: Scolytidae). Canadian Entomologist 131:251–257. Service, Intermountain Research Station, Ogden, UT. USDA FOREST SERVICE. 2003. Forest insect and disease WOOD, S.A. 1982. The bark and ambrosia beetles of North conditions in Utah—2001. Ogden Field Office Tech- and Central America (Coleoptera: Scolytidae), a tax- nical Report 03-01, USDA Forest Service, Inter- onomic monograph. Great Basin Naturalist Memoirs mountain Region, Forest Health Protection, Ogden, 6. Brigham Young University, Provo, UT. UT, and Utah Department of Natural Resources, Division of Forestry, Fire, and State Lands, Salt Lake Received 9 July 2003 City. Accepted 20 April 2004 VEBLEN, T.T., K.S. HADLEY, M.S. REID, AND A.J. REBERTUS. 1991. The response of subalpine forests to spruce beetle outbreak in Colorado. Ecology 72:213–231. Western North American Naturalist 65(1), © 2005, pp. 127–130

USE OF COVER AND RESPONSE TO COVER TYPE EDGES BY FEMALE SIERRA NEVADA RED FOXES IN WINTER

John F. Benson1,2, John D. Perrine3, Richard T. Golightly, Jr.1, and Reginald H. Barrett3

Key words: red fox, Vulpes vulpes necator, cover, cover type, tracks, Lassen, edge.

Red foxes (Vulpes vulpes) use a variety of threatened by the California Fish and Game habitats across their range, including semiarid Commission in 1980, and information regard- deserts, tundra, boreal forests, farmland, and ing the use of cover types by this subspecies urban areas (Larivière and Pasitschniak-Arts will be helpful for land managers when assess- 1996). Within these habitats there is much vari- ing the potential impact of habitat conversion ation in the use of different cover types among (e.g., from timber harvest) on the SNRF. We populations of red foxes (Jones and Theberge investigated use of cover types by 2 female 1982, Halpin and Bissonette 1988, Theberge foxes during winter by following tracks in snow and Wedeles 1989, St-Georges et al. 1995). In in the vicinity of Lassen Peak in the southern Maine, red foxes used coniferous stands and Cascade Range. We describe the use of cover open areas more than expected (Halpin and by these females and also their traveling be- Bissonette 1988). In British Columbia, red havior at cover type edges. foxes used shrub communities more than ex- The study was conducted in the Lassen pected and open areas less than expected National Forest, in a 30-km2 area immediately (Jones and Theberge 1982). In the Yukon Ter- adjacent to the southern entrance of Lassen ritory, red foxes used shrub habitats more than Volcanic National Park in northern California forests or open areas (Theberge and Wedeles where elevations range from 1750 m to 2015 m. 1989). Edge habitat, between forest and shrub The study area consists mainly of mixed con- stands, was important for red foxes in Quebec ifer forests, shrub communities, and meadows. (St-Georges et al. 1995). Variation in use of We followed 9 separate fox trails belonging cover by this species necessitates population- to 2 female foxes (F1 and F5) in snow for a specific studies of red fox habitat relations for total of 18.8 km during December 2000 and use in conservation and management. January 2001. Length of fox trails ranged from Sierra Nevada red foxes (Vulpes vulpes 0.92 km to 5.0 km. We used triangulation to necator; hereafter SNRF) are among the most locate radio-collared foxes and then snow- uncommon and least understood terrestrial shoed into the locations and looked for fresh mammals in California (Schempf and White tracks to backtrack and follow. 1977, Aubry 1997). The distribution of this We used ocular assessment to classify each subspecies is thought to be from 1370 m to cover type as forest, shrub, or open. Forest 3500 m elevation in the Sierra Nevada and cover was defined as having >40% canopy southern Cascade Ranges, and the greatest cover. Shrub cover had <40% canopy cover densities appear to occur near Lassen Peak from trees and ≥20% ground cover by vegeta- (Grinnell et al. 1937, Schempf and White tion with woody stems. Open cover had <40% 1977). Although many researchers have stud- canopy cover and <20% ground cover by veg- ied habitat use of other subspecies of red fox, etation with woody stems. This classification knowledge of the habitat relationships of the method was modified from Mayer and Lauden- SNRF is lacking. The SNRF was listed as slayer (1988).

1Department of Wildlife, Humboldt State University, Arcata, CA 95521. 2Present address: School of Renewable Natural Resources, Louisiana State University, Baton Rouge, LA 70803. 3Department of Environmental Science, Policy, and Management, University of California, Berkeley, CA 94720.

127 128 WESTERN NORTH AMERICAN NATURALIST [Volume 65

To determine cover availability, we followed suggest that the foxes were selecting forest random transects that were the same length as and avoiding open cover. This is consistent paired fox trails. We created these transects with several other studies that found red foxes using a random compass bearing originating used wooded areas more than expected (Cav- from the location where we began following allini and Lovari 1991, Adkins and Stott 1998) the paired fox trail. The distance traveled and avoided open habitats (Jones and Theberge through each cover type along fox trails and 1982, Theberge and Wedeles 1989, Adkins and random transects was paced and converted to Stott 1998). meters. All fox trails belonging to a single indi- Other studies have concluded that red foxes vidual and the associated random transects may use coniferous forests during winter be- were pooled. Thus, for both foxes we gener- cause snow is not as deep in these habitats ated a paired sample of tracks and random (Henry 1980, Halpin and Bissonette 1988). transects of identical distances. Although we did not measure snow depth, We examined traveling behavior at cover limited evidence suggests that snow depth may edges by classifying tracks that intersected edges have influenced traveling behavior. In 5 sepa- as either traveling straight across the edge into rate instances on 2 fox trails, foxes traveled in the adjacent cover type or altering direction cross-country ski tracks or snowshoe tracks for and traveling parallel to the edge. When tracks 748 m (range 19–430 m). The snow in these traveled parallel to the edge, we noted which tracks was probably harder than the surround- side of the edge the tracks traveled (i.e., forest ing snow, and foxes may have traveled in them side, shrub side, or open side). to avoid sinking into deep snow. Traveling in The 2 foxes traveled greater distances in coniferous forests during winter may be another forest and lesser distances in open cover than mechanism used by foxes to avoid deep snow. we measured along the paired random transects Jones and Theberge (1982) found that red (Table 1). Shrub cover was not used by F1, foxes in British Columbia avoided open areas and it constituted only 1% of the distance along and noted that these habitats may not meet the paired random transects. Use of shrub cover their intrinsic need for cover from weather and by F5 was similar to its availability on paired other predators. Competition between coyotes random transects (Table 1). (Canis latrans) and red foxes has been reported Foxes that came to forests from another often, and coyotes have been observed to chase cover type (n = 13) moved straight into forests and kill red foxes (Dekker 1983, Sargeant and (Table 2). Foxes that came to open cover from Allen 1989). In several studies red foxes were forests (n = 21) altered direction and followed found to avoid coyotes, possibly to reduce the edge (Table 2). These foxes followed the competition and harassment (Major and Sher- edge on the forest side (n = 17) more often burne 1987, Sargeant et al. 1987, Harrison et al. than on the open side (n = 4). In one case a 1989). Forests probably provide better cover fox came from shrub to open cover and moved than open habitats for hiding and escaping. straight into the opening. Thus, the patterns of cover use we observed Although statistical analyses were not per- may be a reflection of the need of red foxes for formed due to the small sample size, the 2 cover from larger predators such as coyotes. foxes in this study appeared to use forest more Because of the difficulty of finding and and open cover less than the availability of observing SNRF, we were able to obtain data these cover types. Behavior of the foxes at from only 2 foxes. However, our sample of 2 cover edges also suggested a preference for foxes represents the only information on habi- forest cover because in all cases they crossed tat use by SNRF to date and thus should be directly into the forest rather than following useful in developing further studies and con- the edge. When foxes traveling in forest came servation plans for this subspecies. Further to open cover, they showed a behavioral avoid- studies are needed with larger sample sizes ance of the opening by altering their traveling and statistical analyses to validate these results direction and following the edge. In addition, and investigate possible differences in cover after making turns to avoid openings, they use between genders. The snow-tracking tech- traveled on the forest side of the edges more nique could be combined with extensive vege- often than on the open side. All of these results tation sampling along fox trails and random 2005] NOTES 129

TABLE 1. Distance (m) and proportion of the distance measured along Sierra Nevada red fox tracks and paired random transects through forest, open, and shrub cover types in Lassen National Forest, CA, in winter 2000–2001. Forest Open Shrub Total Fox ID ______number m Proportion m Proportion m Proportion m F1 6714 0.98 133 0.02 0 0.00 6847 Random 5624 0.82 1143 0.17 80 0.01 6847 F5 10,994 0.92 218 0.02 748 0.06 11,960 Random 9124 0.76 1783 0.15 1053 0.09 11,960

TABLE 2. Reactions of red foxes upon coming to cover edges, characterized as either moving straight into the adjacent cover or turning and following the edge during winter 2000–2001. To open To shrub To forest Fox ID ______number Straight Turn Straight Turn Straight Turn F1 0 7 0 0 2 0 F5 1 14 7 7 11 0 Totals 1 21 7 7 13 0

transects to provide specific information about CAVALLINI, P., AND S. LOVARI. 1991. Environmental factors SNRF habitat requirements. If SNRF do select influencing the use of habitat in the red fox, Vulpes vulpes. Journal of Zoology (London) 223:323–339. forests and avoid open habitats, some silvicul- DEKKER, D. 1983. Denning and foraging habits of red tural practices (e.g., large-scale clear-cutting) foxes, Vulpes vulpes, and their interactions with coy- may negatively impact habitat for this sub- otes, Canis latrans, in central Alberta 1972–1981. species. Therefore, understanding the habitat Canadian Field-Naturalist 97:303–306. GRINNELL, J., J.S. DIXON, AND J.M. LINSDALE. 1937. Fur- features (canopy closure, basal area, etc.) of bearing mammals of California. University of Cali- forests selected by SNRF would be useful when fornia Press, Berkeley. potential forest management strategies are HALPIN, M.A., AND J.A. BISSONETTE. 1988. Influence of considered for the region. snow depth on prey availability and habitat use by red fox. Canadian Journal of Zoology 66:587–592. HARRISON, D.J., J.A. BISSONETTE, AND J.A. SHERBURNE. Lassen Volcanic National Park (NPS) and 1989. Spatial relationships between coyotes and red Lassen National Forest (USDA-FS) provided foxes in eastern Maine. Journal of Wildlife Manage- vehicles, office space, and other equipment for ment 53:181–185. HENRY, J.D. 1980. The urine marking behavior and move- this project. The California Department of ment patterns of red foxes during a breeding and post- Fish and Game issued permits authorizing breeding period. Pages 11–27 in D. Muller-Schwaze capture and telemetry on the state-threatened and R.M. Silverstein, editors, Chemical signals: ver- Sierra Nevada red fox. tebrates and aquatic invertebrates. Plenum Press, New York. JONES, D.B., AND J.B. THEBERGE. 1982. Summer home LITERATURE CITED range and habitat utilization of the red fox (Vulpes vulpes) in a tundra habitat, northwest British Colum- ADKINS, C.A., AND P. S TOTT. 1998. Home ranges, move- bia. Canadian Journal of Zoology 60:807–812. ments and habitat associations of red foxes Vulpes LARIVIÈRE, S.L., AND M. PASITCHNIAK-ARTS. 1996. Red fox vulpes in suburban Toronto, Ontario, Canada. Jour- (Vulpes vulpes). Mammalian Species 237:1–11. nal of Zoology (London) 244:335–336. MAJOR, J.T., AND J.A. SHERBURNE. 1987. Interspecific rela- AUBRY, K.A. 1997. The Sierra Nevada red fox (Vulpes tionships of coyotes, bobcats, and red foxes in western vulpes necator). Pages 47–53 in J.E. Harris and C.V. Maine. Journal of Wildlife Management 51:606–616. Ogan, editors, Mesocarnivores of northern Califor- MAYER, K.E., AND W. F. L AUDENSLAYER. 1988. A guide to nia: biology, management, and survey techniques. wildlife habitats of California. California Department Wildlife Society, North Coast Chapter, Arcata, CA. of Forestry and Fire Protection, Sacramento. 130 WESTERN NORTH AMERICAN NATURALIST [Volume 65

SARGEANT, A.B., AND S.H. ALLEN. 1989. Observed interac- hare, red foxes, and river otters in the boreal-tundra tions between coyotes and red foxes. Journal of Mam- transition zone of western Quebec. Canadian Jour- malogy 70:631–633. nal of Zoology 73:755–764. SARGEANT, A.B., S.H. ALLEN, AND J.O. HASTINGS. 1987. Spa- THEBERGE, J.B., AND C.H.R. WEDELES. 1989. Prey selection tial relations between sympatric coyotes and red foxes and habitat partitioning in sympatric coyote and red in North Dakota. Journal of Wildlife Management fox populations, southwest Yukon. Canadian Journal 51:285–293. of Zoology 67:1285–1290. SCHEMPF, P.F., AND M. WHITE. 1977. Status of six fur- bearer populations in the mountains of northern Cal- Received 15 December 2003 ifornia. USDA Forest Service, San Francisco, CA. Accepted 20 July 2004 ST-GEORGES, M., S. NADEAU, D. LAMBERT, AND R. DECARIE. 1995. Winter habitat use by ptarmigan, snowshoe Western North American Naturalist 65(1), © 2005, pp. 131–132

A MICROPTEROUS, CRENON-DWELLING POPULATION OF MEGARCYS SUBTRUNCATA HANSON (PLECOPTERA: PERLODIDAE)

Bill P. Stark1 and Richard W. Baumann2

Brachyptery and microptery, the condition includes 20 males, 13 females, and 23 nymphs. of having short wings, occur sporadically Forewing lengths for males ranged from 2.5 among perlodine stoneflies and are generally mm to 3.0 mm and 4.0 mm to 4.5 mm for more common among males in high-altitude females. Among male specimens the forewings and high-latitude populations. Indeed, brachyp- generally covered the base of the hindwings, tery is not uncommon in Megarcys species and the hindwings extended to the midpoint such as M. signata (Hagen) (e.g., Fig. 6.55 in of abdominal tergum 2. Females have slightly Stark et al. 1998) and M. watertoni (Ricker), longer hindwings that reach the posterior mar- but this, to our knowledge, involves only males gin of abdominal tergum 2. Figure 1 shows the and the wings usually cover at least through modified venation found on the female right abdominal tergum 7. It was, therefore, surpris- forewing. ing to discover micropterous male and female Nymphs agree in most respects with the specimens of M. subtruncata Hanson during generic description in Stewart and Stark (2002), recent fieldwork in the Pacific Northwest. but in this population the “dorsal fringe of Adults were collected by beating vegetation silky white hairs” is absent or reduced to a few around a large spring outflow in Quinn River obscure setae on the basal cercal segments Campground, Deschutes County, Oregon, and and is absent from the head, thorax, and ab- nymphs were clustered on large rocks and dominal terga as shown for M. signata. These woody debris near the spring source. Although fringes are present on M. subtruncata nymphs numerous specimens were present, our sample in other populations we have examined.

Fig. 1. Megarcys subtruncata, female right forewing, Oregon, Deschutes Co., Quinn River spring, Quinn River Camp- ground, 10-VI-2004, B.P. Stark and R.W. Baumann. Actual forewing length = 4.5 mm.

1Department of Biological Sciences, Mississippi College, Clinton, MS 39058. 2Department of Integrative Biology, M.L. Bean Life Science Museum, Brigham Young University, Provo, UT 84602.

131 132 WESTERN NORTH AMERICAN NATURALIST [Volume 65

The other oddity associated with this dis- subtruncata population can be expected in such covery is the habitat. We often find the perlid isolated populations. stonefly Hesperoperla pacifica (Banks) in large spring outflows such as the Head of the LITERATURE CITED Metolius River, and we have also found healthy populations of the perlodid Frisonia LILLEHAMMER, A. 1985. Studies of shortwingedness in stoneflies (Plecoptera). Fauna Norvegica Series B 32: picticeps (Hanson) in the Fall River spring. 58–61. However, we have not previously found a STARK, B.P., S.W. SZCZYTKO, AND C.R. NELSON. 1998. Amer- species of Megarcys as the dominant carnivo- ican stoneflies: a photographic guide to the Plecop- rous stonefly in a large crenon-influenced site. tera. Caddis Press, Columbus, OH. The Quinn River spring site is only a few miles STEWART, K.W., AND B.P. STARK. 2002. Nymphs of North American stonefly genera (Plecoptera). 2nd edition. from similar sites on the Fall River and Cultus Caddis Press, Columbus, OH. River in Deschutes County, Oregon, but it may be more isolated by mountainous topog- Received 20 July 2004 raphy. Lillehammer (1985) suggests “profound Accepted 27 September 2004 shortwingedness” of the type seen in this M. Western North American Naturalist 65(1), © 2005, pp. 133–135

PRONGHORN HYPERSENSITIVITY TO AVIAN SCAVENGERS FOLLOWING GOLDEN EAGLE PREDATION

Jon P. Beckmann1 and Joel Berger2

Key words: Golden Eagle, hypersensitivity, predation, pronghorn, ravens, scavengers.

In the high-elevation deserts of western snow. There were no other prints. Thus, we Wyoming, a facultative scavenger, the Common concluded that the cause of death was Golden Raven (Corvus corax), and an avian predator, Eagle predation. the Golden Eagle (Aquila chrysaetos), are sym- On 28 January 2003, USDI-BLM biologists patric with pronghorn (Antilocapra americana). observed a Golden Eagle predation event on a Here we report encounters between these 3 pronghorn within 8 km (5 miles) of the site of species during winter, when pronghorn re- our predation event (John Dahlke, wildlife sponded with extreme anti-predatory behavior. biologist subcontracted by BLM, Pinedale, WY, While conducting ecological surveys of personal communication). Similarly, other re- pronghorn on 23 January 2003 in the Green searchers have documented predation of prong- River basin of western Wyoming, we noticed a horn by Golden Eagles (Lehti 1947, Burns Golden Eagle atop a sagebrush (Artemisia tri- 1970, Goodwin 1976, Barrett 1978, Beale 1978, dentata) plant near an exceptionally large group Bodie 1978, Von Gunten 1978, Autenrieth 1980, of pronghorn (N = 775). Upon carefully Deblinger and Alldredge 1996). Most reported approaching the sagebrush plant, we discov- observations of Golden Eagle attacks on prong- ered a dead young female pronghorn (27 kg) horn involve newborn fawns in the spring and in snow 8–10 cm deep. We examined the car- summer, although Deblinger and Alldredge cass and found the body still warm and twitch- (1996) reported 6 incidents of eagle predation ing, with blood emanating from it. Our necropsy on pronghorn in winter (November–February). revealed fractured cervical vertebrae and deep In each of these reported attacks of eagles on puncture wounds from the talons of an eagle pronghorn in winter, group sizes ranged from on the back of the pronghorn’s neck. Because 120 to 350 pronghorn. The pronghorn carcass the fractured cervical vertebrae were under we investigated lay in the middle of a swath of the puncture wounds from the talons, and no fresh pronghorn tracks >100 m in width, sug- legs were broken or showed signs of injury, it gesting that the pronghorn had likely been is unlikely that the pronghorn fell while run- taken from a large group at the time of its ning, resulting in the fracture. Necropsy also death. Because the carcass we investigated was revealed wounds that had hemorrhaged be- still very warm and little feeding had taken neath the skin in the area of the punctures place, we estimated that the pronghorn had from the talons. All ribs were intact and no been dead for <20 minutes. meat was removed from them. Canid teeth The group of 775 pronghorn we observed marks were not observed on any of the bones, was within 200 m of the carcass. In nearly nor were there any canid tracks in the adja- 1000 hours of observing 968 groups of prong- cent region. No other visible marks were doc- horn in the upper Green River basin during umented on the carcass other than feeding 2002–2003, we never observed a group >250 marks made by the Golden Eagle on the pos- individuals. While we observed the hypervigi- terior end of the carcass. Investigation of the lant pronghorn, we documented 6 incidents scene surrounding the carcass revealed the within 5 minutes in which ravens arriving on presence of bloodstained eagle tracks in the the scene flew over the group of pronghorn at

1Wildlife Conservation Society, Eastern Idaho Field Office, 528 Marian, Rigby, ID 83442. 2Wildlife Conservation Society, Teton Field Office, Moose, WY 83012.

133 134 WESTERN NORTH AMERICAN NATURALIST [Volume 65 heights between 5 m and 25 m. In each case tween avian scavengers and predators, such as the pronghorn began to run. In 2 cases the Golden Eagles. Stockwell (1991) concluded that ravens vocalized above the group of prong- in order to distinguish between these 2 alter- horn. Following the 6th pass by a raven, in natives, additional data needed to be collected. which the raven vocalized, the pronghorn be- Why might pronghorn respond in a hyper- gan to run at top speed in a single-file line sensitive manner and display an extreme flight through the snow in a 5.6-km (3.5-mile) response in the presence of an avian scaven- straight-line sprint. At one point the single-file ger when fawns, which are vulnerable to preda- line of 775 pronghorn was over 1.6 km (1 mile) tion by scavengers, are not present in winter? in length. We documented the time of latency The data presented here seem to suggest that to 1st foraging bite (LTFB) as 52 minutes for pronghorn were unable to discriminate between the group of pronghorn following this sprint avian scavengers and predators and that the (LTFB was defined as 10% of the group begin- extreme grouping behavior and hypersensitive ning to browse). Following our sampling of the responses of pronghorn may be a maladaptive pronghorn, we returned to the site of the car- by-product of a generalized anti-predator re- cass; within 1 hour we documented 3 Golden sponse to large avian predators. Golden Eagles Eagles, 1 Bald Eagle (Haliaeetus leucocephalus), and ravens are often seen together in western 6 Black-billed Magpies (Pica pica), and 4 Com- Wyoming, and perhaps under duress, prong- mon Ravens on the carcass. horn may associate ravens with the real dan- Hypersensitive reactions by prey species to ger posed by eagles. Extreme responses to a the presence of predators following the death nonthreatening scavenger, such as the 5.6-km of a member of the group has been reported all-out sprint brought on by ravens described for a diverse array of species: e.g., Jackass Pen- here, could have serious energetic costs for guins (Spheniscus demersus; Randall and Ran- overwinter survival, especially in regions such as western Wyoming, where temperatures are dall 1990), ground squirrels (Spermophilus ele- ° gans; Pfeifer 1980), and crows (Corvus brachy- frequently –29 C or lower. It is possible that rhynchos hesperis; Hauser and Caffrey 1994), pronghorn hypersensitivity to the ravens was an indirect effect of the recently witnessed pre- but not for ungulates. Additionally, hypersen- dation event, and thus they would respond to sitive reactions to the presence of predators any stimulus in such a manner. However, given that actually represent a very marginal danger that the pronghorn did not respond to our to the prey species have been reported: e.g., vehicle or to us during the >50 minutes we ringtailed lemur (Lemur catta) and Verreaux’s observed the group discounts this idea. It seems sifaka (Propithecus verreauxi) in Madagascar more plausible that pronghorn were unable to (Goodman 1994). In these cases the adverse distinguish between avian scavengers and responses of these species to present-day hawks predators. This inability to discern true dan- and kites, which are not large enough to pose ger posed by an avian predator from the very as serious a threat as an extinct species of marginal danger of a scavenger would likely eagle did, seem maladaptive (Goodman 1994, have energetic repercussions for pronghorn. Csermely 1996). Responses of an ungulate species to avian scavengers have been previ- LITERATURE CITED ously noted for desert bighorn sheep (Ovis canadensis nelsoni; Stockwell 1991). However, AUTENRIETH, R.E. 1980. Vulnerability of pronghorn fawns mass hysteria and hypersensitivity by an un- to predation. Proceedings of the Pronghorn Antelope gulate species in response to the presence of a Workshop 9:77–79. BARRETT, M.W. 1978. Pronghorn fawn mortality in Alberta. scavenger species, such as ravens, have not Proceedings of the Pronghorn Antelope Workshop been reported previously in the literature. 8:429–444. Why might ungulates, and specifically prong- BEALE, D.M. 1978. Birth rate and fawn mortality among pronghorn antelope in western Utah. Proceedings of horn, respond to avian scavengers? Stockwell the Pronghorn Antelope Workshop 8:445–448. (1991) posed 2 possibilities: (1) avian scaven- BODIE, W.L. 1978. Pronghorn fawn mortality in the upper gers can be predators on neonates of several Pahsimeroi River drainage in central Idaho. Proceed- ungulate species (Gilbert 1919, Lehti 1947, ings of the Pronghorn Antelope Workshop 8:417–428. BURNS, E.H. 1970. Winter predation of Golden Eagles Larsen and Deitrich 1970, Schaller 1972), and and coyotes on pronghorn antelope. Canadian Field- (2) ungulates may be unable to distinguish be- Naturalist 84:301–304. 2005] NOTES 135

CSERMELY, D. 1996. Antipredator behavior in lemurs: evi- LEHTI, R.W. 1947. The Golden Eagle attacking antelope. dence of an extinct eagle on Madagascar or something Journal of Wildlife Management 11:348–349. else? International Journal of Primatology 17:349–354. PFEIFER, S. 1980. Aerial predation on Wyoming ground DEBLINGER, R.D., AND W. A LLDREDGE. 1996. Golden Eagle squirrels. Journal of Mammalogy 61:371–372. predation on pronghorns in Wyoming’s Great Divide RANDALL, R.M., AND B.M. RANDALL. 1990. Cetaceans as Basin. Journal of Raptor Research 30:157–159. predators of Jackass Penguins Spheniscus demersus: GILBERT, T.G. 1919. Turkey Vulture. Bird Lore 21: deductions based on behaviour. Marine Ornithology 319–322. 18:9–12. GOODMAN, S.M. 1994. The enigma of antipredator behav- SCHALLER, G. 1972. The Serengeti lion. University of ior in lemurs: evidence of a large extinct eagle on Chicago Press, Chicago. Madagascar. International Journal of Primatology 15: STOCKWELL, C.A. 1991. Behavioural reactions of desert big- 129–134. horn sheep to avian scavengers. Journal of Zoology GOODWIN,G.A. 1976. Golden Eagle predation on prong- 225:563–566. horn antelope. Auk 94:789–790. VON GUNTEN, B.L. 1978. Pronghorn fawn mortality on the HAUSER, M.D., AND C. CAFFREY. 1994. Anti-predator National Bison Range. Proceedings of the Pronghorn response to raptor calls in wild crows, Corvus bra- Antelope Workshop 8:394–416. chyrhynchos hesperis. Animal Behaviour 48:1469– 1471. Received 18 November 2003 LARSEN, K.H., AND J.H. DEITRICH. 1970. Reduction of a Accepted 22 March 2004 raven population on lambing grounds with DRC- 1339. Journal of Wildlife Management 34:200–204. Western North American Naturalist 65(1), © 2005, pp. 136–139

NATAL BURROWS AND NESTS OF FREE-RANGING PYGMY RABBITS (BRACHYLAGUS IDAHOENSIS)

Janet L. Rachlow1, Dana M. Sanchez1, and Wendy A. Estes-Zumpf1

Key words: pygmy rabbit, Brachylagus idahoensis, lagomorph, natal burrow, nest, reproduction, shrubsteppe.

Pygmy rabbits (Brachylagus idahoensis) are spp.) and other shrubs were provided, those a sagebrush obligate species of conservation materials were not incorporated into the nests concern. The Columbia Basin population in (Oregon Zoo 2001, Lamson and Shipley 2002). Washington is listed as federally endangered In captivity, females constructed nests during (Federal Register 2003), and in April 2003 the daylight hours, back-filled the entrances of U.S. Fish and Wildlife Service received a peti- natal burrows with soil, and camouflaged the tion for rangewide endangered status for the entrances so well that keepers often were un- species. Although studies investigating the able to identify the location of natal burrows ecology of the pygmy rabbit date back to the (L. Shipley, Washington State University, Pull- 1940s (Orr 1940, Janson 1946, Severaid 1950), man, personal communication). Juvenile pygmy many aspects of the species’ natural history rabbits remained in the nest for 13–21 days, remain poorly understood, especially with re- after which the females left the entrance to the spect to reproduction. Locations of parturition natal burrows open and the young vacated and nests have not been documented for this the nests (Oregon Zoo 2001, Lamson and Ship- species. Pygmy rabbits are 1 of 2 North Amer- ley 2002). ican leporids that dig extensive burrow systems We studied pygmy rabbits at 2 sites in the (Green and Flinders 1980). However, neither Lemhi Valley in south central Idaho. The study nesting materials nor neonates have been found areas consist of shrubsteppe habitat domi- in residential burrow systems that were either nated by big sagebrush (Artemisia tridentata) excavated or examined with a burrow camera and green rabbitbrush (Chrysothamnus viscidi- (Bradfield 1975, Rauscher 1997). Wilde (1978) florus) with a sparse understory of grasses and observed 2 isolated juveniles weighing 90 g forbs. Both study sites are located at the base under separate sagebrush plants. Based on of a ridge of low foothills dominated by earth- growth curves of captive pygmy rabbits, those mounded microtopography known as “mima juveniles would have been approximately 3 mounds.” While investigating reproductive pat- weeks of age (Lamson and Shipley 2002). From terns in pygmy rabbits during April–July 2003, this observation Wilde (1978) speculated that we located 7 natal burrows representing the females give birth aboveground and scatter 1st documentation of natal burrows for free- young to avoid loss of entire litters to predation. ranging pygmy rabbits. Our objective in this Observations during captive breeding have paper is to characterize natal burrows and nests provided the most detailed information on for this species in their natural habitat. reproductive behaviors of pygmy rabbits. Preg- We measured dimensions of the tunnel and nant females were observed digging single- nest chamber of the 7 vacated natal burrows. entrance natal burrows that terminated in a Using a compass, we recorded aspect of the en- nest chamber. Females excavated the natal bur- trance and angle at which the burrow tunnel rows from 7 to 10 days before parturition and declined. Additionally, we measured maximum lined the nest chamber with grass and hair; height of the 10 nearest shrubs to the natal although branches of sagebrush (Artemisia burrow entrance. We estimated that the natal

1Department of Fish and Wildlife Resources, University of Idaho, Moscow, ID 83844.

136 2005] NOTES 137

TABLE 1. Dimensions of 7 natal burrows and nest chambers of pygmy rabbits measured in the Lemhi Valley, Idaho, during May–July 2003.

______Natal burrow tunnel ______Nest chamber Total ______Entrance (cm) Angle Aspect ______Entrance (cm) ______Dimensions (cm) depth ID # Height Width (degrees) (degrees) Height Width Height Width Length (cm) 1 14.0 17.0 20 256 10.0 11.0 14.5 16.0 13.0 29.0 2 17.0 9.0 30 130 7.0 6.5 14.5 16.0 22.0 45.0 3 15.0 8.5 25 80 10.0 8.0 13.0 17.0 20.0 43.0 4 13.0 14.0 30 170 9.0 10.5 13.0 16.0 12.0 43.0 5a 7.5 10.0 25 25 10.5 9.5 18.0 19.0 21.0 74.0 6 10.0 8.5 25 108 9.5 10.5 14.0 12.0 16.0 45.0 7 10.0 10.5 35 186 5.0 15.0 12.0 16.0 11.0 30.0

Mean 12.4 11.1 27 10.1 10.1 14.1 16.0 16.4 44.1 s 3.3 3.2 4.9 2.4 2.7 1.9 2.1 4.6 14.9 aNest construction not completed.

burrows had been vacant for 1 to 3 weeks when appeared to be a natal burrow in which nest we conducted these measurements. Remain- construction was initiated but not completed. ing nesting materials were collected, air-dried, These 3 diggings were about 15 cm apart at weighed, and examined for content. Pygmy the base of sagebrush shrubs. Subsequent vis- rabbits use residential burrow systems through- its revealed limited evidence of further dig- out the year, and these systems are character- ging, and nest construction was not completed. ized by multiple entrances, substantial buildup Cottontail rabbits (Sylvilagus floridanus) have of excavated soil and rocks, and a surrounding been observed to initiate more than one nest carpet of fecal pellets. We evaluated spatial prior to reproduction (Beule and Studholme relationships between natal burrows and resi- 1942, Rongstad 1966). Our observations sug- dential burrow systems by searching the area gest that female pygmy rabbits may exhibit within a 50-m radius of each natal burrow for similar behaviors. currently active or recently active residential We located 5 additional natal burrows that burrow systems. had been vacated, and on one occasion we We observed 2 female pygmy rabbits dig- observed a juvenile pygmy rabbit run into an ging and subsequently back-filling burrows open natal burrow when disturbed. The 6 with soil during midmorning (1030 and 0815 completed natal burrows had similar construc- hours) on 6 May and 15 June 2003. Those be- tion. Each consisted of a tunnel with a single entrance ranging in length from 17 cm to 30 cm haviors were consistent with natal burrow con- that ended in a single, spherical chamber (Table struction observed in captive pygmy rabbits. 1). The uncompleted natal burrow was longer, Examination of the site dug in May revealed more curved, and declined at a steeper angle, an area of loose soil covered with plant debris perhaps to avoid roots of sagebrush. At each of at the base of rabbitbrush. On May 13 we the 6 completed natal burrows, we found nest- opened the burrow entrance and found a sin- ing material containing hair within the nest gle tunnel approximately 30 cm deep. How- chamber and occasionally outside the burrow ever, on May 26 we found a natal burrow that entrance. Examination indicated that the hair had been vacated by the juvenile rabbits morphology was consistent with pygmy rabbit approximately 20 cm away from the dead-end hair. Because cottontail rabbits were rare in tunnel at the base of a sagebrush. Despite our study areas and we observed pygmy rabbits repeated visits to the area during May, the real in association with 3 of the natal burrows, we natal burrow remained undetected until the assumed that the hair was from maternal pygmy nest was vacated and the tunnel entrance left rabbits. This conclusion is consistent with ob- open. At the site where we observed digging servations of nest construction by captive pygmy in June, we found 2 dead-end tunnels and what rabbits (Lamson and Shipley 2002). 138 WESTERN NORTH AMERICAN NATURALIST [Volume 65

Fine grasses, shredded bark from sagebrush row was located in the open between mounds, (Artemisia spp.), and hair were the primary but the burrow entrance was located under a components of nesting materials. We also weathered sagebrush stump. Based on these found fleas and mites in some nests; these are observations, rabbits appeared to place natal common parasites of both adult and juvenile burrows under relatively dense shrub cover. pygmy rabbits. The dry mass of nesting mate- Nests of rabbit species in North America rials recovered from each natal burrow ranged exhibit a graded variation in depth. Swamp from 30 g to 75 g (mean = 58.4, s = 16.1, n = rabbits (Sylvilagus aquaticus) construct nests of 6). These values likely underestimate the total woven vegetation at the surface of the ground mass of nesting materials because some may or in shallow depressions, which may be an have blown away or been removed. The uncom- adaptation to wet habitats (Holler et al. 1963, pleted nest contained 45 g of shredded bark. Sorensen et al. 1972). Mountain cottontails (S. Observations of captive pygmy rabbits indi- nuttallii) have been described as constructing cated that females began lining nests with similar cuplike nests (Chapman 1975). Several fresh grass from 8 to 3 days before parturition, other members of the genus Sylvilagus dig and that hair was pulled from the body and nest chambers below the ground surface and added to the nest shortly before birth (Lamson camouflage the opening with vegetation (Orr and Shipley 2002). Nests of the captive pygmy 1940, Ingles 1941, Skeels 1962, Casteel 1966). rabbits did not contain shredded bark; how- In contrast, pygmy rabbits in this study and in ever, this may have been influenced by a lack captivity (Oregon Zoo 2001) placed nest cham- of live shrubs within their enclosures. bers deeper into the ground at the base of Pygmy rabbits appeared to establish natal a tunnel and then back-filled the burrow en- burrows away from their residential burrow trances with soil. systems. Three of the natal burrows we exam- We revisited natal burrows 1 to 4 weeks ined had no active or recently active residen- after recording nest measurements. Most bur- tial burrow systems within 50 m. Mean dis- row entrances were partially collapsed and tances to active residential burrow systems for filled with debris or soil. If nesting materials the remaining nests were 34 m (n = 3; range were removed by wind or rodents after nests = 22–44), 47 m (n = 1), 28 m (n = 6; range = were vacated, it would be difficult to recog- 8–40), and 37 (n = 2; range = 35–38). Overall nize the holes as natal burrows. These condi- average distance to active residential burrow tions and the apparent isolation of natal bur- systems was >35 m. “Active” burrow systems rows from areas of general rabbit activity have were defined as having open entrances free of likely contributed to the lack of documenta- debris in association with fresh pellets, and tion of pygmy rabbit nests before this study. “recently active” burrow systems were associ- Given the inconspicuous nature of pygmy rab- ated with weathered pellets. Additionally, signs bit natal burrows, researchers may need to of rabbit activity (pellets, digging, and dust focus on observations of nest-building behav- baths) were not observed in the vicinity of iors (digging natal burrows, gathering vegeta- nests, suggesting that females excavate natal tion, and pulling hair) to locate nests of this burrows away from areas of general activity. species. This behavior may avoid attracting predators to the location of nests and neonates. ACKNOWLEDGMENTS Vegetation structure and composition around the natal burrows was typical for mima mounds This work was funded by the Idaho Depart- in shrubsteppe habitats. Shrubs tend to be ment of Fish and Game through the Wildlife taller and denser on the mounds, with little Conservation and Restoration Program (grant or lower shrub cover in the areas between R-1-6-0214). Additional support was provided mounds (Tullis 1995). Six of the natal burrows by the Bureau of Land Management, U.S. For- were associated with mima mounds, and the est Service, and University of Idaho. Our thanks burrow entrances were located at the base of to J. Witham, L. Shipley and her graduate stu- shrubs (5 under sagebrush and 1 under rabbit- dents, H. Roberts, B. Waterbury, V. Guyer, and brush). Average shrub height around the 6 the Leadore Ranger District. The manuscript natal burrows located on mima mounds was was improved by the comments of 2 anony- 59.7 cm (range = 41–86 cm). One natal bur- mous reviewers. 2005] NOTES 139

LITERATURE CITED LAMSON, R., AND L. SHIPLEY. 2002. Washington pygmy rabbit (Brachylagus idahoensis): captive breeding BEULE, J.D., AND A.T. STUDHOLME. 1942. Cottontail rabbit summary 2002. Unpublished report. nests and nestlings. Journal of Wildlife Management OREGON ZOO. 2001. Pygmy rabbit (Brachylagus idahoen- 6:133–140. sis) captive care and breeding. Unpublished report. 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. RAUSCHER, R.K. 1997. Status and distribution of the pygmy CASTEEL, D.A. 1966. Nest building, parturition, and copu- rabbit in Montana. Montana Fish, Wildlife, and Parks lation in the cottontail rabbit. American Midland Final Report. Naturalist 75:160–167. RONGSTAD, O.J. 1966. Biology of penned cottontail rabbits. CHAPMAN, J.A. 1975. Sylvilagus nuttallii. Mammalian Journal of Wildlife Management 30:312–319. Species 56:1–3. SEVERAID, J.H. 1950. The pygmy rabbit (Sylvilagus idaho- FEDERAL REGISTER, MARCH 5, 2003. Endangered and ensis) in Mono County, California. Journal of Mam- threatened wildlife and plants; final rule to list the malogy 31:1–4. Columbia Basin distinct population segment of the SKEELS, M.A. 1962. Nesting behavior of Sylvilagus audu- pygmy rabbit (Brachylagus idahoensis) as endangered. boni neomexicanus. Journal of Mammalogy 43:542– Federal Register 68:10388–10409. 544. GREEN, J.S., AND J.T. FLINDERS. 1980. Brachylagus idaho- SORENSEN, M.F., J.P. ROGERS, AND T.S. B ASKETT. 1972. ensis. Mammalian Species 125:1–4. Parental behavior in swamp rabbits. Journal of Mam- HOLLER, N.R., T.S. BASKETT, AND J.P. ROGERS. 1963. Repro- malogy 53:840–849. duction in confined swamp rabbits. Journal of Wildlife TULLIS, J.A. 1995. Characteristics and origin of earth- Management 27:179–183. mounds on the Eastern Snake River Plain, Idaho. INGLES, L.G. 1941. Natural history observations on the Master’s thesis, Idaho State University, Pocatello. Audubon cottontail. Journal of Mammalogy 22: WILDE, D.B. 1978. A population analysis of the pygmy 227–250. rabbit (Sylvilagus idahoensis) on the INEL site. Doc- JANSON, R.G. 1946. A survey of the native rabbits of Utah toral dissertation, Idaho State University, Pocatello. with reference to their classification, distribution, life histories and ecology. Master’s thesis, Utah State Received 15 December 2003 University, Logan. Accepted 18 May 2004 Western North American Naturalist 65(1), © 2005, p. 140

REVIEWERS NEEDED

The following books are available for review HARMON, R. 2002. Crater Lake National Park: A His- in the WESTERN NORTH AMERICAN NATURAL- tory. Corvallis, Organ State University Press. HOFRICHTER, R. 2002. Toxic Struggles. Salt Lake City, IST. If you would like to write a review for our readers, please contact Dr. C. Riley Nelson, University of Utah Press. HOLADAY, B. 2003. The Return of the Mexican Gray [email protected] Wolf: Back to the Blue. Tucson, University of Ari- zona Press. BAGNE, M., AND B. RICHARD. 2002. Yellowstone Coun- KEATOR, G. 2003. Introduction to Trees of the San try: The Photographs of Jack Richard. Lanham, Francisco Bay Region. Berkeley, Los Angeles, Lon- Roberts Rinehart Publishers. don, University of California Press. BAKER, M.B., JR., P.F. FOLLIOTT, ET AL. 2004. Riparian LESICA, P.2002. Flora of Glacier National Park. Corval- Areas of the Southwest. Boca Raton, London, New lis, Oregon State University Press. York, Washington D.C., Lewis Publishers. LOTT, D.F. 2002. American Bison: A Natural History. BARON, J.S. 2002. Rocky Mountain Futures. Washington, Berkeley, London, Los Angeles, University of Cal- Covelo; London, Island Press. ifornia Press. BUTCHER, R.D. 2003. America’s National Wildlife LOVE, M.S., M. YOKLAVICH, ET AL.2002. The Rockfishes Refuges: A Complete Guide. USA, Roberts Rine- of the Northeast Pacific. Berkeley, Los Angeles, hart. London, University of California Press. CANNINGS, R.A. 2002. Introducing the Dragonflies of MEE, W., J. BARNES, ET AL. 2003. Waterwise: Native British Columbia and the Yukon. Victoria, Royal Plants for Intermountain Landscapes. Logan, Utah British Columbia Museum. State University Press. DALY, H., V. ROBIN, ET AL. 2002. Sustainable Planet. MORRISON, M.L. 2002. Wildlife Restoration. Washing- Boston, Beacon Press. ton, Covelo; London, Island Press. DEBINSKI, D.M., AND J.A. PRITCHARD. 2002. A Field MOYLE, P.B. 2002. Inland Fishes of California. Berke- Guide to Butterflies of the Greater Yellowstone ley, Los Angeles and London, University of Califor- Ecosystem. USA, Roberts Rinehart. nia Press. DECOURTEN, F.L. 2003. Adventures in Great Basin PIANKA, E.R., AND L.J. VITT. 2003. Lizards: Windows Geology: The Broken Land. Salt Lake City, Univer- to the Evolution of Diversity. Berkeley, Los Ange- sity of Utah Press. les, London, University of California Press. DOMBECK, M.P., C.A. WOOD, ET AL. 2003. From Con- RINGHOLZ, R.C. 2002. Uranium Frenzy: Saga of the quest to Conservation: Our Public Lands Legacy. Nuclear West. Logan, Utah State University Press. Washington, Covelo; London, Island Press. THEODOROPOULOS, D.I. 2003. Invasion Biology: Cri- FERGUSON, G. 2003. Hawks Rest: A Season in the tique of a Pseudoscience. Blythe, California, Avvar Remote Heart of Yellowstone. Washington D.C., Books. National Geographic Adventure Press. TOPPING, G. 2002. Great Salt Lake: An Anthology. FESTA-BIANCHET, M., AND M. APOLLONIO. 2003. Animal Logan, Utah State University Press. Behavior and Wildlife Conservation. Washington, TUALATIN RIVERKEEPERS. 2002. Exploring the Tualatin Covelo; London, Island Press. River Basin. Corvallis, Oregon State University Press. GORDON, G. 2003. Landscape of Desire: Identity and VAN DEVENDER, T.R. 2002. The Sonoran Desert Tor- Nature in Utah’s Canyon Country. Logan, Utah toise. Tucson, University of Arizona Press and the State University Press. Arizona–Sonora Desert Museum. GRISMER, L.L. 2002. Amphibians and Reptiles of Baja WILLIAMS, R.L. 2003. “A Region of Astonishing Beauty”: California. Berkeley, Los Angeles, London, Univer- The Botanical Exploration of the Rocky Mountains. sity of California Press. USA, Roberts Rinehart. GROVES, C.R. 2003. Drafting a Conservation Blue- WOOD, D.L., T.W. KOERBER, ET AL. 2003. Pests of the print: A Practitioner’s Guide to Planning for Biodi- Native California Conifers. Berkeley, Los Angeles, versity. Washington, Covelo; London, Island Press. London, University of California Press.

BOOK REVIEW

Cannings, Robert A. 2002. The systematics of centrated his most detailed work on a section Lasiopogon (Diptera: Asilidae). Royal of the flies limited largely to the Nearctic region. British Columbia Museum, Victoria, Initially, Dr. Cannings assumed it to be a British Columbia, Canada. 353 pp. ISBN relatively straight-forward revision. He knew 0-7726-4636-8. that there were many described species and assumed that few new species would be en- The robber flies (Insecta: Diptera: Asilidae) countered in this relatively well-collected fam- are a diverse family of true flies, with the ily of flies. He clearly states he was wrong in number of described species in the world this assumption. He estimates that roughly approaching that of the number of bird species. half of the species he encountered were unde- Yet most people, familiar at least in passing scribed and new to science. He found these with robins, sparrows, and Kentucky Fried specimens in collections from 84 museums Chicken, are completely unaware of these fas- scattered throughout the northern hemisphere. cinating insects. He describes 14 new species in his opaculus Dr. Robert Cannings is familiar with robber section to add to the 15 previously named flies and has created a superb monograph of species. This is roughly a 50% new species rate part of one genus, Lasiopogon. This monograph in a group thought to be quite adequately is a sturdy, hardbound volume that resulted described for a part of the world thought to from years of study of this group of flies lim- have been well documented! He notes that ited to the Holarctic region. Dr. Cannings this section comprises approximately 25% of gives a wonderful summary of the biology and Lasiopogon worldwide. Assuming an equal rate natural history of the family as a whole. It will of new species to be described elsewhere serve as a jumping-off point for students of (actually a conservative estimate based on these flies for years to come because of its bibli- sparse collecting in central Asia and eastern ographic thoroughness. He tackles the plethora Europe), one can see that the described fauna of morphological terms used in past descrip- of 51 species is probably closer to 100. In fact, tions and settles on standard terminology for as Dr. Cannings sorted through and meticu- obvious and obscure body parts that can serve lously dissected the taxonomically diagnostic as a basis for future detailed studies of robber terminalia of many of these flies, he set aside fly morphology. 49 as new species awaiting description! Oh, The study of this apparently monophyletic such a wonderfully diverse world is that of genus Lasiopogon, resting as it does in a phylo- insect taxonomy! One might have hoped for a genetic morasse of interesting genera, serves complete, detailed revision and description of as a model for elucidating and clarifying rela- all species in the genus, but alas, life is short, tionships among genera and problematic sub- and artificial deadlines by seen and unseen families. The approach of this monograph is to masters impose their need to publish at mile- consider the broad taxonomic categories to stones. Nevertheless, an appropriate mile- which Lasiopogon belongs and methodically stone has been reached. Dr. Cannings codifies dissect these categories into less inclusive the morphological terms that can be used to groups using the standard phylogenetic tools advantage in years to come and outlines a of outgroup comparison and parsimony. As the clear methodology for preparing and dissect- author immersed himself in this monumental ing specimens. work, he clearly realized that a complete and The volume is full of clear line drawings to easy end was not to be reached. Instead, he con- be used in identification. For these aspects he

141 142 WESTERN NORTH AMERICAN NATURALIST [Volume 65 is to be commended. These sections of the book Parsimony, or some other biogeographic analy- will have value for years to come and will set a sis techniques. Dr. Cannings was clearly tenta- standard not only for future work in the genus, tive in pushing the phylogenies in this fashion but also for all detailed taxonomic studies of because he knew that so many species are yet flies in general. He continues with a well-pre- to be described and so many areas are yet to sented, yet complex hypothesis of phylogenetic be collected thoroughly! But the trail is blazed relationships among genera, sections, groups, and the gauntlet is dropped for those who fol- and species groups. He is apologetic for the low. This is a great scientific treatment. lack of resolution this analysis gives (no apology needed; simply a few nodes have trichoto- C. Riley Nelson mies!). He considers these phylogenetic rela- Department of Integrative Biology tionships in a somewhat rambling biogeogra- Brigham Young University phic section that might have benefited from Provo, UT 84602 additional analysis using nested-clade, Brooks [email protected] ERRATA NOTICE

It has been brought to the attention of the financial support.” The sentence should read: WESTERN NORTH AMERICAN NATURALIST that “We thank our 2 reviewers, J. Mead and S. several errors were introduced by the editorial Lucas, for their comments and suggestions, staff and subsequently published in volume 64, and the University of Texas (UT) Geology issue 4, of the journal in the article, “First Foundation for financial support.” Pleistocene jumping mouse (Zapus, Zapodinae, Also, the running heads in this article should Rodentia) from Utah,” by Dennis R. Ruez, Jr., have read WESTERN NORTH AMERICAN NATU- and Christopher J. Bell. RALIST on the left-hand pages and FIRST PLEIS- On page 443, under Acknowledgments, the TOCENE ZAPUS FROM UTAH on the right-hand copy reads: “We thank our 2 reviewers, J. Mead pages. and S. Lucas, for their comments and sugges- We apologize for any misunderstanding tions, and the Utah Geology Foundation for these errors may have caused.

CONTENTS

(Continued from back cover)

Notes (continued) Use of cover and response to cover type edges by female Sierra Nevada red foxes in winter . . . . John F. Benson, John D. Perrine, Richard T. Golightly, Jr., and Reginald H. Barrett 127 A micropterous, crenon-dwelling population of Megarcys subtruncata Hanson (Plecoptera: Perlodidae)...... Bill P. Stark and Richard W. Baumann 131 Pronghorn hypersensitivity to avian scavengers following Golden Eagle predation ...... Jon P. Beckmann and Joel Berger 133 Natal burrows and nests of free-ranging pygmy rabbits (Brachylagus idahoensis) ...... Janet L. Rachlow, Dana M. Sanchez, and Wendy A. Estes-Zumpf 136

Book Review The Systematics of Lasiopogon (Diptera: Asilidae) by Robert A. Cannings ...... C. Riley Nelson 141

143