response to two years of restoration techniques in an

The Analysis of Management Strategies toRestore and Enhance Nelson’s Checker-mallow

(Sidalcea nelsoniana) Habitat at William L.Finley National Wildlife Refuge

(Order Nos. 1448-13420-97-M303, 1448-13420-98-M279,101819-M584)

Response to two years ofrestoration techniques in anexisting Sidalcea nelsoniana

habitat: Final report

Submitted to

U. S. Fish and Wildlife Service

Submitted by

Mark V. Wilson

Department of Botany and Plant PathologyOregon State University

7 May 2004

Final report: Sidalcea nelsoniana field experiment

Response to two years of restoration techniques in anexisting Sidalcea nelsoniana habitat

SummaryBecause of extreme habitat loss and encroachment by woody and weedy pest plants,

Sidalcea nelsoniana is threatened with extinction. This study tested the effectiveness of twoyears of prescribed burning and mowing as restoration techniques for existing S. nelsonianapopulations and habitat. Vegetation and four measures of S. nelsoniana vigor were measuredbefore (1998) and after (2000) experimental applications of burning and mowing in the BurnedSwale population within the W. L. Finley National Wildlife Refuge. Treatments had theintended effect of reducing competitive vegetation only for canopy cover and tall woody cover. Burning significantly stimulated the growth of herbaceous cover, which would increasecompetition with S. nelsoniana. Some treatment effects on vegetation varied with hydrologicconditions.

After two years of treatments, there was no evidence of increased vigor of S. nelsoniana. In fact, S. nelsoniana cover and flowering declined with burning. Longer-term control of woodyplants might eventually lead to increased vigor of this perennial plant, of course.

Based on the results from this study, burning in similar habitats with S. nelsoniana shouldbe approached with caution. Mowing, the alternative examined in this study, can be asbeneficial in controlling woody vegetation without harming S. nelsoniana plants.

IntroductionSidalcea nelsoniana (Nelson’s checker-mallow) is a native Willamette Valley plant listed

under the federal Endangered Species Act as threatened with extinction (US Fish and WildlifeService, 1993). One of the largest populations of this species occurs at W. L. Finley NationalWildlife Refuge. The Endangered Species Act calls for the protection and recovery ofpopulation viability of listed species and habitat restoration plays a key role in this task (Falk1990, Soulé 1991, Wilson et al. 1992, Sinclair et al. 1995). Yet effective techniques formanaging S. nelsoniana and its habitat remain unavailable. The goal of the study described hereis to develop the basic information and test the applied tools needed by land managers to protectand restore S. nelsoniana. This study is part of a series of studies on S. nelsoniana undertakenby the prairie research group at Oregon State University in cooperation with the WillametteValley National Wildlife Refuge Complex of the US Fish and Wildlife Service (Bartels 2000,Bartels and Wilson 2001, Bartels and Wilson 2003, Wilson in prep.)

Sidalcea nelsoniana is known from wetland prairies and streamsides of the WillametteValley and from some adjacent areas of the Oregon Coast Range and Cowlitz County,Washington (U.S. Fish and Wildlife Service 1993). Two major factors threaten the survival ofthe species (U.S. Fish and Wildlife Service 1993). First, habitat loss, primarily through


conversion to agricultural use, has been some of the most severe in the nation (Noss et al. 1995),limiting the current distribution of S. nelsoniana to only 48 sites in 5 population centers (U.S.Fish and Wildlife Service 1993). Second, habitat degradation from the invasion of shrubs andtrees continues in remaining habitat.

Two unresolved questions have key importance to the conservation and recovery of S.nelsoniana. The prairie and streamside habitats of S. nelsoniana are changing because of theinvasion and growth of woody plants or aggressive herbaceous pest plants like reed canarygrass. These changes are likely causes for some apparently precipitous declines in S. nelsoniananumbers at Finley National Wildlife Refuge (M. B. Naughton, personal observation). Inparticular, it is unclear how well S. nelsoniana adults and seedlings can tolerate shade and othercompetitive pressures (U.S. Fish and Wildlife Service 1993, CH2M Hill 1994, Glad et al. 1994,Gisler and Meinke 1995). Prudent management will probably require active manipulation of S.nelsoniana habitat. The impact of management strategies to improve habitat quality is bestunderstood by knowing both the tolerance of S. nelsoniana to existing woody and herbaceousplant cover and the response of S. nelsoniana to experimental manipulations.

The objective of this study was to evaluate the effectiveness of two years of prescribedburning and mowing as restoration techniques for existing S. nelsoniana habitat. Specifically,the study addressed three questions:

1. Do burning or mowing reduce the abundance of vegetation that competes with S.nelsoniana? The hope is that maintaining an open habitat will promote long-term S.nelsoniana growth and reproduction.

2. Do burning or mowing harm S. nelsoniana in the short term? The hope is that vegetationmanagement measures do not have adverse, non-target effects on S. nelsoniana.

3. If burning or mowing do not harm but in fact benefit S. nelsoniana, is the benefit related tothe control of competing vegetation?

First-year results are reported by Bartels (2000) and Bartels and Wilson (2003).

MethodsStudy species

Sidalcea nelsoniana is a gynodioecious species that can propagate from either rhizomes orseeds. Reproductive individuals have 30-cm to 100-cm tall flowering stalks terminating inspikelike inflorescences of pinkish-lavender to pinkish-purple flowers (U.S. Fish and WildlifeService 1993). Peak flowering occurs mid-June through mid-July in the Willamette Valley. Thefruits are several-seeded schizocarps and mature in late July and early August. S. nelsoniana ismost often found in wetland prairies, ash swales, streamsides, and roadside ditches. Associated

1All nomenclature follows Hitchcock and Cronquist (1973).


vegetation includes graminoids such as Festuca arundinacea1, Phalaris arundinacea, Agrostisspp., and Carex spp.; weedy forbs such as Heracleum lanatum and Vicia spp.; and woodyspecies such as Rosa spp., Rubus spp., and Fraxinus latifolia (U.S. Fish and Wildlife Service1998, Bartels 2000, Wilson in prep.). Adjacent vegetation typically senesces between lateAugust and mid-September (Bartels pers. obs.).

Study site

One of the largest populations of S. nelsoniana is found at William L. Finley NationalWildlife Refuge (NWR), 16 km south of Corvallis in the Willamette Valley, Oregon, USA. Thestudy population of Sidalcea nelsoniana is located within the refuge in a riparian hedgerowbetween two agricultural fields planted with perennial rye grass (Field 5—Burned Swale, Figure1). The hedgerow runs east-west and is approximately 750 m long by 50 m wide. The elevationdifference between the highest and lowest measurement quadrats is approximately 7.1 m. Theeast end of the site floods to approximately 20 cm above the soil surface from November throughApril, while the west end of the site remains unflooded throughout the year (Bartels 2000).

The dominant vegetation of this site includes Festuca arundinacea, Vicia spp., Rubusdiscolor at the driest end with Phalaris arundinacea, Phalaris aquatica and Carex spp.dominating in the wettest portion of the site. Fraxinus latifolia is found throughout. Sidalceanelsoniana individuals are scattered throughout the length of the site, sometimes growing underfull sun and sometimes under partial to heavy shade. The soil at the site is Waldo silty clayloam, while soils in the surrounding agricultural fields are classified as Coburg silty clay loamand Amity silt loam (Soil Conservation Service 1975).

The region experiences a modified maritime climate with cool, wet winters and warm, drysummers. Average annual precipitation as recorded in Corvallis is 108.5 cm, with 93.0 cmoccurring October through April. Average minimum and maximum temperatures are 0.6ºC and7.5ºC in January and 10.6ºC and 26.8ºC in July (Oregon Climate Service 1990).

Experimental design

Because of the spatial heterogeneity of the site and logistical constraints in applyingtreatments, typical experimental designs were impractical. Rather, the study was designed toinvestigate the vigor of S. nelsoniana under both experimentally manipulated conditions(prescribed burning and mowing) and a range of pre-existing conditions of woody plant coverand hydrology, using General Linear Modeling (described below, McNeil et al. 1996) as the toolfor statistical analysis.

During the spring of 1998, the site was surveyed and S. nelsoniana individuals weremarked and categorized into strata according to pre-manipulation density of woody plant cover


and general hydrologic conditions (Table 1). Flooding and woody plant cover were describedcategorically only for the purposes of stratifying the manipulations. Analysis of the effects ofthese variables on S. nelsoniana used actual measurements of flooding and woody plant cover(as explained below). A total of 347 S. nelsoniana plants were marked throughout the site. Ofthese, 25 were vegetative and 322 were reproductive, as indicated by the presence of floweringstalks. Only individuals that were reproductive in 1998 were included in the study.

Of the reproductive individuals, 30 plants were randomly selected from each of the pre-manipulation strata, for a total of 120 plants. A 0.5-m2 square quadrat was centered on eachselected individual. These plant-centered quadrats were the observational units in the study. Treatment areas were delineated to be of practical size for safe and effective manipulations andto allow for buffers between treatments. Therefore, treatment areas varied in size andencompassed varying numbers of quadrats from one or more strata. Treatments were randomlyassigned to each area, with slight adjustment to balance the number of quadrats in each of the 12treatment-by-condition strata (Table 1).

Because of the extent of the site, three “pseudoblocks” were assigned along the length ofthe site. The boundaries were assigned to balance both the number of replicate treatments andthe number of quadrats.

Vegetation measurements

Pre-manipulation conditions were recorded within each quadrat between 26 June 1998 and7 July 1998. The same characteristics were measured for post-manipulation conditions between5 July 2000 and 11 July 2000. Measurements of S. nelsoniana vigor included growth (cover)and flowering intensity (number of flowering stalks, number of inflorescences, and height of thetallest flowering stalk). Measurements of the surrounding vegetation included canopy canopycover (above 1.5 m), tall woody cover (above S. nelsoniana mid-height of 40 cm but below 1.5m) short woody cover (below 40 cm), tall herbaceous cover (above 40 cm) and short herbaceouscover (below 40 cm). Litter depth was also recorded. Canopy cover was estimated from thenorth side of the quadrat using a spherical densiometer, so that measurements emphasizedcanopy cover from the south. Cover of S. nelsoniana and surrounding herbaceous and woodyvegetation was estimated by consensus of two investigators using calibration templates.

Hydrologic measurements

Because the hydrologic impact on wetland plants is largely determined by the timing andduration of flooding, hydrologic impacts on S. nelsoniana were characterized through the use ofelevation surveys and site observations throughout the fall, winter and spring as described inBartels (2000). Elevation is often correlated with soil moisture (Nelson and Anderson 1983) andwas used to quantify the relative hydrologic impact on each S. nelsoniana measurement plot.


Manipulations

Prescribed burning and mowing manipulations were conducted 10-11 September 1998 byU. S. Fish and Wildlife Service personnel. As discussed above, treatments were applied togroups of quadrats and ranged between 10 and 30 m in width by 25 to 50 m in length, with abuffer of at least 1 m between the treatment edge and the nearest measurement quadrat. Therewere at least four quadrats within each manipulated and unmanipulated area within eachpseudoblock (median = 10 quadrats).

Convection burns were ignited at the southwest corner of each burn treatment area. Flames burned for 5 to 20 minutes after ignition, with average flame heights between 0.5 and 2m. In areas with thick vegetation, flames often shot into the ash canopy. In similar prairies,experimental-scale fires have been shown to closely approximate the behavior of large-scalefires (Maret 1996).

Within each mowing treatment area, herbaceous vegetation and shrubs were mowed with a4.5 m-wide tractor mower to approximately 15 cm above the soil surface. Large trees were cutwith a chain saw and all woody brush was removed from the site.

Burning and mowing manipulations were repeated in mid-September, 1999. The reductionin fuels after one year of burning reduced fire intensity in 1999, yet fire still carried over the bulkof the treatment plots.

Statistical analysis

Vegetation

The effect of burning and mowing on vegetation was tested by analysis of covariance. Theresponse variable was the change in vegetation cover or litter depth from 1998 (beforemanipulations) to 2000 (after two years of burning, mowing, or no manipulation). The covariatewas the relative elevation of each plot, a good indicator of the hydrologic condition of the plot. Variables were transformed as necessary to meet statistical assumptions. Alpha was set to 0.05as a compromise between the exploratory nature of the analysis (suggesting a more lenientvalue) and multiple analyses (requiring a more stringent value). The F-ratio was calculated bycomparing the treatment mean square to the mean square of the interaction of treatment andrelative elevation, if that interaction was statistically significant (Underwood 1997, Newman etal. 1997). If interaction was significant, interactions details were examined before interpretingthe main effects. Least square means were then compared using Tukey’s test (Day and Quinn1989).

Sidalcea nelsoniana

The analysis approach was to test for a treatment (burning or mowing) effect on S.nelsoniana only after accounting for differences in the pre-existing conditions of shading,


crowding, hydrology, and blocking (Table 2, Appendix 1). S. nelsoniana vigor was measured ascover, number of stalks, number of inflorescences, and height. The response variable was thechange from 1998 (before manipulations) to 2000 (after two years of burning, mowing, or nomanipulation). The data variables were transformed as necessary to meet statistical assumptions. Using a backwards-and-forwards stepwise procedure, a best full model was developed startingwith treatment and its interaction with all other explanatory variables. The restricted model wasthe same model with all the terms containing Treatment eliminated. The difference inexplanatory power of the two models measured both the direct and indirect treatment effects onS. nelsoniana. The statistical significance of these effects were evaluated by the generalized F-test (McNeil et al. 1996).

Results and discussionPre-manipulation vegetation conditions

Although quadrats were centered on S. nelsoniana plants, pre-manipulation cover of S.nelsoniana (July 1998) within each quadrat averaged only 8.6% (Table 3). Thus, this rare plant,even in a relatively large and important population, was not dominant in the vegetation.

Pre-manipulation herbaceous cover below S. nelsoniana mid-height (40 cm) was oftendense, averaging 45.6%. Herbaceous cover above S. nelsoniana mid-height averaged 15.3%. Woody cover below S. nelsoniana mid-height was relatively sparse, averaging 3.2%. Above S.nelsoniana mid-height (and below 1.5 m), woody cover averaged 8.7%. Average canopy coverwas 23.3% and litter depth averaged 5.4 cm (Table 3).

1. Did treatments reduce competing vegetation?

Canopy cover and tall woody cover were dramatically and significantly less in burned andmowed plots in 2000, after two years of treatments (Tables 4 and 5). In contrast, canopy coverand tall woody cover increased between 1998 and 2000 in unmanipulated plots, reflecting thesuccessional trend of these sites in the absence of disturbance. Mowing – by definition – iseffective at removing these plants. Although one or two fall mowing treatments do not often killplants directly, mowed plants did not regrow fully by the next seasons. Burning consumed muchof the smaller woody fuels, including branches, leaving the larger stems intact. Nevertheless,these intact stems either died or were unable to produce enough new growth to compensate forthe foliage consumed.

Short woody cover had no significant variation across treatments (Tables 4 and 5). Although these plants were largely different species from those in the tall woody cover andcanopy cover groups, they were removed by burning and mowing at rates comparable to those ofthe taller plants. Short woody cover exhibited faster regrowth, however, eliminating anytreatment effect.


In still another pattern, both tall and short herbaceous cover averaged significantly higherin burn treatments as compared to control plots (Tables 4 and 5). Herbaceous plants are usuallynot directly harmed by fall burning or mowing because these species are then dormant and evenshallow burial protects regenerative structures. Burning seemed particularly to stimulate growthduring the next season of tall herbaceous cover, much of which was Vicia spp. scrambling overother plants.

Relative elevation had a strong impact on the regrowth of short herbaceous plants (Table5). The wettest areas gained, on average, 3 times as much short herbaceous cover as did thedriest areas, regardless of treatment. If expansion of short herbaceous plants causes competitivepressure on S. nelsoniana, it should be felt more severely in wetter microhabitats.

A significant interaction of treatment and relative elevation masks their main effects onlitter depth. There was a large and significant reduction in litter depth with burning, as expected(Tables 4 and 5), and this reduction occurred throughout the elevation gradient (Figure 2). Inboth mowed plots and controls, lower, hence wetter, elevations tended to lose litter depth, withthe rate of loss slightly greater with mowing (Figure 2). Interpreting the significance of thesechanges in litter depth on S. nelsoniana vigor must await more detailed studies.

In summary, treatments had the intended effect of reducing competitive vegetation only forcanopy cover and tall woody cover. These two categories of vegetation are the major sources ofshading pressure on S. nelsoniana, so in this sense the treatments were successful. On the otherhand, the stimulus of herbaceous cover by burning could potentially increase competitivepressure on S. nelsoniana for both nutrients and light.

2. Did burning or mowing treatments cause short-term harm to Sidalceanelsoniana?

All four measures of S. nelsoniana vigor tended to decline between 1998 and 2000 in no-manipulation controls (Table 6), highlighting the urgency of management to protect thisthreatened species. The question for this study is whether burning and mowing differed fromcontrols on their effect of S. nelsoniana. The treatment factor had statistically significant effectson all four measures of S. nelsoniana vigor (Table 7), although for all four measures theinfluence of treatment was felt at least in part through interactions with other variables (Table 8).

S. nelsoniana cover

Treatments had their main influence on S. nelsoniana cover through interactions with othervariables, in particular short woody cover (Table 8). In fact, the only place where S. nelsonianaconsistently increased in cover from 1998 to 2000 was in no-manipulation plots where shortwoody cover was present in 1998 (Figure 3). Overall, burning caused a greater and statisticallysignificant decline in S. nelsoniana cover compared with the decline in no-manipulation plots,with a loss of over a third of cover with burning (Table 6).


Number of flowering stalks

Treatments had their influence on the number of flowering stalks exclusively throughinteractions with other variables, predominately with short woody cover; direct treatment effectwas not statistically significant (Table 8). Where short woody cover was absent, the number offlowering stalks declined with burning and no manipulation (median proportional changes of-50% and -31%) but remained almost unchanged with mowing. Where short woody cover waspresent, mowing caused a sharp decline (-56%), burning caused a decline (-37%), but stalksincreased in number (+33%) in no-manipulation plots. The mechanism by which short woodycover influences the effects of treatments on flowering stalks remains unknown. Overall,burning caused the most decline in number of flowering stalks (Table 6).

Relative elevation had a statistically significant influence on changes in numbers offlowering stalks (Table 8), with increases slightly more likely at higher, drier elevations. Therewere no significant interactions between relative elevation and treatments.

Number of inflorescences

Treatments had both direct effects on the changes in the numbers of inflorescences andindirect effects through interactions with other variables (Tables 6-8). The main interaction waswith pseudoblock (Table 8). The number of S. nelsoniana inflorescences tended to declinethroughout the site, except in mowed plots in the dry end (pseudoblock A), where floweringsharply increased, with an average of 21 more inflorescences per plant in 2000 than in 1998. The mechanism by which mowing and hydrology interact to influence S. nelsoniana floweringremains unknown.

Overall, burning caused the largest overall decline in number of inflorescences, about half(Table 6), although the strength of this effect was not quite at the " = 0.05 level (P = 0.053).

Height

Treatments had both direct effects on the changes in S. nelsoniana height and indirecteffects through interactions with other variables (Tables 6-8). Height tended to decline in alltreatments, most strongly in mowed plots (Table 6). An important interaction was with tallwoody cover (Table 8). The more tall woody cover present in 1998 in no-manipulation plots,the greater the decline in height (Figure 3). This result is consistent with the premise that theseplants can suppress S. nelsoniana by overtopping and crowding. In contrast, the stronglyadverse effect of mowing was ameliorated by the presence of tall woody cover: the more tallwoody cover there was in 1998, the more chance there was for the release of S. nelsoniana fromsuppression to balance the adverse effect of mowing. Neither pattern of suppression or releasewas present in burned plots (Figure 4).


Another important interaction of treatments was with pseudoblock (Table 8), which isclosely related to hydrology and relative elevation. S. nelsoniana tended to maintain or increaseheight only in the wetter end of the habitat (Figure 5).

Summary of short-term effects on S. nelsoniana

In general, the rate of S. nelsoniana decline was higher with burning. Hydrology and woodycover also influenced S. nelsoniana. Drier microhabitats within the population tended to bebetter for flowering, but wetter microhabitats tended to be better for height growth.

3. If burning or mowing do not harm but in fact benefit Sidalcea nelsoniana, isthe benefit related to the control of competing vegetation?

These results offer little evidence that burning or mowing provide short-term benefit to S.nelsoniana. Only in the combination of treatment and narrow microhabitats were S. nelsonianaplants benefitted: cover increased in mowed plots with tall woody cover between 15% and 30%and flowering increased in mowed plots in pseudoblock A. Otherwise, either S. nelsoniana wasnot released from competition by encroaching vegetation or any benefits from release wereoutweighed by direct, adverse affects of burning or mowing. Of course, control of competingvegetation could still lead to long-term benefits to this threatened species.

Comparison with results from one year of treatments

One year of treatment by burning or mowing significantly changed aspects of thesurrounding vegetation but had little influence on S. nelsoniana (Bartels 2000, Bartels andWilson 2001). The patterns in vegetation change seen after one year of treatment continued andwere amplified after two years of treatment (Table 9). This consistency suggest that treatmentseffects are neither strictly temporary nor a result of a single year’s weather.

Burning and mowing had no significant effect on S. nelsoniana cover and flowering after asingle year of treatment. After two years of treatment, however, S. nelsoniana cover declinedsignificantly in burned plots (Tables 7 and 9). Burning also caused the biggest declines in thenumber of S. nelsoniana flowering stalks and inflorescences (Table 6), although these patternswere not statistically significant.

Summary of results and their implications for managing S. nelsoniana

All four measure of S. nelsoniana vigor declined from 1998 to 2000. The intensive,quantitative results from unmanipulated plots in this study match past surveys and observationsof S. nelsoniana. At the measured rate of decline in cover (about -20%), S. nelsoniana coverwithout intervention would be half its 1998 level by 2004, with similar losses of reproductive


output. These results reinforce the urgency of some management intervention to protect andconserve the threatened S. nelsoniana.

Prescribed burning and mowing are two available options for managing S. nelsonianahabitats. In this study, burning and mowing caused some similar results but also differed in theireffects on S. nelsoniana and the surrounding vegetation. Moreover, their effects were not alwaysbeneficial.

• Burning and mowing reduced the cover only of the canopy and tall woody plants. Thesetwo categories of vegetation are the major sources of shading pressure on S. nelsoniana,so in this sense the treatments were successful.

• On the other hand, the stimulus of short and tall herbaceous cover by burning couldpotentially increase competitive pressure on S. nelsoniana for both nutrients and light. Wetter microhabitats particularly gained short herbaceous cover, so burning in wetterareas might promote competition from herbs.

• Only burning showed evidence of significant harm to S. nelsoniana. Two years ofburning caused a 40% decline in cover compared to the 20% decline in cover in mowedand unmanipulated plots.

These experimental results of vegetation and S. nelsoniana after two years oftreatment suggest caution when considering burning in similar habitats with S. nelsoniana. Mowing, an alternative, can be as beneficial as burning, through controlling woodyvegetation, and is less likely to cause either direct or indirect harm to S. nelsoniana.

Location along the hydrologic gradient influenced these experimental results. Carefulmanagement would adjust accordingly for hydrologic conditions. Wetter areas tended to showrapid increases in short herbaceous cover. Because neither burning nor mowing was effective inreducing short herbaceous cover, manual removal might be necessary in the wettermicrohabitats. S. nelsoniana in wetter microhabitats also tended to decline in the number offlowering stalks and inflorescences. These adverse changes are consistent with a species at theedge of its range. Bartels and Wilson (2003), using a mesocosm experiment, drew similarconclusions that S. nelsoniana is ill-suited to inundation greater than its current distributionwithin the Burned Swale site.

Acknowledgments

I thank Marilynn Bartels and Jennifer Goodridge for their very capable effort in leading thefield work and for supervising the field crews. Special thanks go to Marilynn Bartels for longdiscussions concerning all phases of this work.


References cited

Bartels, M. B. 2000. Conservation of Sidalcea nelsoniana through habitat management: Effectsof burning, mowing, and altered flooding regime on a rare Willamette Valley perennial. Thesis,Oregon State University.

Bartels, M. R., and M. V. Wilson. 2001. Fire and mowing as management tools for conservinga threatened perennial and its habitat in the Willamette Valley, Oregon. Pages 59-65 inProceedings of the 17th North American Prairie Conference: Seeds of the Future, Roots of thePast, N. P. Bernstein and L. J. Ostander (eds.).

Bartels, M. B., and M. V. Wilson. 2003. Flood tolerance of the threatened Sidalcea nelsoniana(Malvaceae). Madroño 50:265-270.

CH2M Hill. 1994. Sidalcea nelsoniana monitoring--1994. Technical memorandum preparedfor McMinnville Water and Light.

Day, R. W., and G. P. Quinn. 1989. Comparisons of treatments after an analysis of variance inecology. Ecological Monographs 59:433-463.

Falk, D. A. 1990. Restoration of endangered species: a strategy for conservation. Pages 328-334 in J. J. Berger, editor. Environmental Restoration: Science and Strategies for Restoring theEarth. Island Press, Washington, D C.

Gisler S, and R. Meinke. 1995. Sidalcea nelsoniana: Examination of previous MWL-sponsoredresearch. Report of Oregon Department of Agriculture research results (1994).

Glad, J.B., R. R. Halse, and R. Mishaga. 1994. Observations on distribution, abundance, andhabitats of Sidalcea nelsoniana Piper (Malvaceae) in Oregon. Phytologia 76:307-323.

Hitchcock, C. L., and A. Cronquist. 1973. Flora of the Pacific Northwest: An IllustratedManual. University of Washington Press, Seattle.

Maret, M. P. 1996. Effects of fire on seedling establishment in upland prairies in the WillametteValley, Oregon. M.S. Thesis. Oregon State University, Corvallis.

McNeil, K. A., I. Newman, and F. J. Kelly. 1996. Testing research hypotheses with the generallinear model. Southern Illinois University Press, Carbondale.

Nelson, D. C., and R. C. Anderson. 1983. Factors related to the distribution of prairie plantsalong a moisture gradient. American Midland Naturalist 109:367-375.


Newman, J. A., J. Bergelson, and A. Grafen. 1997. Blocking factors and hypothesis tests inecology: Is your statistics text wrong? Ecology 78:1312-1320.

Noss, R. F., E. T. LaRoe III, and J. M. Scott. 1995. Endangered ecosystems of the UnitedStates: A preliminary assessment of loss and degradation. National Biological Service,Biological Report 28.

Oregon Climate Service. 1990. Climatological summaries (1961-1990): Monthly means andextremes, Corvallis, Oregon (http://www.ocs.orst.edu/pub_ftp/climate_data/mme/mme1862.html).

Sinclair, A. R. E., D. S. Hik, O. J. Schmitz, G. G. E. Scudder, D. H. Turpin, and N. C. Larter. 1995. Biodiversity and the need for habitat renewal. Ecological Applications 5:579-587.

Soil Conservation Service. 1975. Soil survey of Benton County Area, Oregon. U.S. Departmentof Agriculture.

Soulé, M. E. 1991. Conservation: Tactics for a constant crisis. Science 253:744-750.

U.S. Fish and Wildlife Service. 1993. Final rule: Nelson’s checker-mallow. Federal Register,February 12, 1993.

U.S. Fish and Wildlife Service. 1998. Recovery plan for the threatened Nelson’s checker-mallow (Sidalcea nelsoniana). Portland, Oregon.

Underwood, A. J. 1997. Experiments in ecology. Cambridge University Press, Cambridge.

Wilson, M. V., P. C. Hammond, and J. B. Kauffman. 1992. The value of restoration andmanagement in protecting biodiversity. Northwest Environmental Journal 8:201-202.


Table 1. Stratification scheme for applicationof experimental manipulations.

Extensivelyflooded

Notextensively

flooded

Woody plants dense

Burned Burned

Mowed Mowed

Control Control

Woody plantsnot dense

Mowed Mowed

Burned Burned

Control Control


Table 2. Response and explanatory variables used in the analysis of changes in vegetationand Sidaclea nelsoniana from 1998 (before treatments) to 2000 (after two years of treatmentor no manipulations). See Methods for details.

Vegetation

Response variables Explanatory variables

Canopy cover (>1.5 m) (%)Tall woody cover (between 40 cm and 1.5 m) (%)Short woody cover (below 40 cm) (%)Tall herbaceous cover (between 40 cm and 1.5 m) (%)Short herbaceous cover (below 40 cm) (%)Litter depth (cm)

Treatment (burning, mowing, or nomanipulation)

Relative elevation (m)

Sidalcea nelsoniana

Response variables Potential explanatory variables

Vegetative cover (%)Number of flowering stalksNumber of inflorescencesHeight of flowering stalks (cm)

Treatment (burning, mowing, or nomanipulation)

Relative elevation (m)PseudoblockCanopy cover in 1998 (%)Tall woody cover in 1998 (%)Short woody cover in 1998 (%)Tall herbaceous cover in 1998 (%)Short herbaceous cover in 1998 (%)Litter depth in 1998 (cm)


Table 3. Summary of vegetation and Sidalcea nelsoniana in 1998, before treatments, within112 S. nelsoniana-centered quadrats.

Variable Mean Standarderror Minimum Maximum

S. nelsoniana cover (%) 8.6 0.5 1 25

Flowering stalks (number) 5.7 0.5 1 34

Inflorescences (number) 27.3 2.5 0 135

Height of tallest flowering stalk (cm) 114.7 2.3 52 180

Canopy cover (%) 23.3 2.7 0 97

Tall woody cover (%) 8.7 1.4 0 75

Short woody cover (%) 3.2 0.5 0 25

Tall herbaceous cover (%) 15.3 1.0 2 60

Short herbaceous cover (%) 45.6 2.3 4 98

Litter depth (cm) 5.4 0.3 0 21


Table 4. Average ±standard error of the cover of vegetation components (%) and depth of litter (cm)in 1998 and 2000, after two years of treatment, and the proportional change of the averages.

OverallTreatments

Burning Mowing No manipulation

Canopy cover,1998

22.2 ±2.7 23.6 ±4.6 21.5 ±4.9 21.5 ±4.5

Canopy cover,2000

11.3 ±2.0 5.0 ±1.3 1.1 ±0.6 28.2 ±5.0

Prop. change -49.1% -78.8% -94.9% +31.2%

Tall woody cover,1998

8.5 ±1.4 11.8 ±2.8 4.4 ±1.3 9.3 ±2.8

Tall woody cover,2000

8.4 ±1.5 5.4 ±1.4 2.7 ±1.0 17.3 ±3.7

Prop. change -1.2% -54.2% -38.6% +86.0%

Short woody cover,1998

3.3 ±0.5 3.5 ±0.8 1.8 ±0.7 4.5 ±1.1

Short woody cover,2000

3.9 ±0.6 3.9 ±0.9 2.7 ±1.0 5.0 ±1.2

Prop. change +18.2% +11.4% +50.0% +11.1%

Tall herbaceouscover, 1998

15.5 ±1.0 15.1 ±1.7 17.6 ±1.8 13.7 ±1.4

Tall herbaceouscover, 2000

31.4 ±1.7 38.3 ±3.1 28.5 ±2.7 27.5 ±2.5

Prop. change +102.6% +153.6% +61.9% +100.7%

Short herbaceouscover, 1998

46.1 ±2.3 42.1 ±3.8 50.1 ±4.4 46.1 ±3.6

Short herbaceouscover, 2000

70.4 ±1.7 73.1 ±2.6 77.4 ±2.8 60.4 ±2.9

Prop. change +52.7% +73.6% +54.5% +31.0%

Litter depth (cm),1998

5.5 ±0.3 4.9 ±0.4 5.6 ±0.5 6.0 ±0.4

Litter depth (cm),2000

3.6 ±0.3 0.8 ±0.2 3.7 ±0.4 6.4 ±0.7

Prop. change -34.5% -83.7% -33.9% +6.7%


Table 5. Analysis of variance for the changes in vegetation from 1998 (before treatments) to2000 (after two years of treatment or no manipulation). The headings show the form of theresponse variable, including any transformation used. P-values in bold are <0.05. Significantdifferences from the no-manipulation controls were tested with Dunnett’s test, " = 0.05; B:burning, M: moving, C: no-manipulation controls.

df SS MS F P Significant differences

Canopy cover (%): (cover 2000)½ - (cover 1998)½; R2 = 34.0%

Treatment 2 71.1 35.5 6.64 <0.01 M, B < C

Relative elevation 1 13.8 13.8 2.57 0.11

Treatment × Relative elevation 2 5.3 2.7 0.5 0.61

Residuals 107 573.0 5.4

Tall woody cover(%): (cover 2000)½ - (cover 1998)½; R2 = 22.3%

Treatment 2 53.6 26.8 9.78 <0.01 B, M < C



Residuals 107 293.1 2.7

Short woody cover (%): cover 2000 - cover 1998; R2 = 1.9%

Treatment 2 25.0 12.5 0.46 0.63



Residuals 107 2903.3 27.1

Tall herbaceous cover (%): cover 2000 - cover 1998; R2 = 12.2%

Treatment 2 1634.8 817.4 3.01 0.05 B > C



Residuals 107 29104.6 272.0

Table 5. Analysis of variance for the changes in vegetation from 1998 (before treatments) to2000 (after two years of treatment or no manipulation). The headings show the form of theresponse variable, including any transformation used. P-values in bold are <0.05. Significantdifferences from the no-manipulation controls were tested with Dunnett’s test, " = 0.05; B:burning, M: moving, C: no-manipulation controls.

df SS MS F P Significant differences


Short herbaceous cover (%): cover 2000 - cover 1998; R2 = 20.6%

Treatment 2 3154.1 1577.0 2.92 0.06 B > C

Relative elevation 1 7352.0 7352.0 13.61 <0.01


Residuals 107 57806.6 540.3

Litter depth (cm): (depth 2000)½ - (depth 1998)½; R2 = 34.8%

Treatment† 2 28.8 14.4 5.12 0.16 B < C

Relative elevation† 1 6.0 6.0 2.13 0.28


Residuals 107 79.5 0.7

†Because of the significant interaction term, Treatment and Relative elevation main effects were tested against theinteraction mean square.


Table 6. Average ±standard error abundance and vigor of Sidalcea nelsoniana in 1998 and2000, after two years of treatment, and the proportional changes of the averages.

OverallTreatments

Burning Mowing Nomanipulation

Cover, 1998 8.5 ±0.5 7.4 ±0.7 8.8 ±0.9 9.4 ±0.8

Cover, 2000 6.4 ±0.5 4.6 ±0.5 7.1 ±1.0 7.5 ±0.8

Prop. change -24.7% -37.8% -19.3% -20.2%

Stalks, 1998 5.7 ±0.5 5.4 ±0.7 6.5 ±1.0 5.2 ±0.6

Stalks, 2000 4.2 ±0.4 3.3 ±0.5 4.8 ±0.7 4.6 ±0.7

Prop. change -26.3% -38.9% -26.2% -11.5%

Infls, 1998 27.5 ±2.6 24.6 ±4.3 29.9 ±4.7 28.0 ±4.3

Infls, 2000 20.3 ±2.8 11.8 ±2.5 28.7 ±7.2 20.4 ±3.4

Prop. change -26.2% -52.0% -4.0% -27.1%

Height, 1998 114.6 ±2.2 112.1 ±3.5 116.4 ±4.6 115.4 ±3.1

Height, 2000 88.1 ±3.7 85.4 ±6.6 79.3 ±6.1 99.9 ±6.2

Prop. change -23.1% -23.8% -31.9% -13.4%


Table 7. Generalized F-test for the changes in Sidalcea nelsoniana from 1998 (beforetreatments) to 2000 (after two years of treatment or no manipulation). The models withtreatment were derived from stepwise procedures. (See Methods for details.) The F-statisticis based on the change in explanatory power when treatment is removed from the model. heading show the form of the response variable, including any transformation used. P-valuesin bold are <0.05. Significant differences from the no-manipulation controls were tested withDunnett’s test, " = 0.05; B: burning, M: moving, C: no-manipulation controls.

Variable and transformation

R2 of modelwith

treatment

Test of treatment effect

Generalized F(degrees

of freedom)

P Significantdifferences

Cover (%); rank of proportionalchange

23.1% 2.3 (12,95) 0.01 B < C

Number of stalks; rank of proportionalchange

24.4% 2.2 (6,100) 0.05

Number of inflorescences; differenceof ranks

24.4% 2.1 (10,97) 0.03 (B < C)

Height (cm); difference of ranks 31.9% 30.1 (10,98) <0.01


Table 8. Models derived from stepwise procedures with treatment included. Responsevariables (with transformation) are the changes from 1998 to 2000. Explanatory vegetationvariables are pre-treatment conditions (1998). P-values for individual terms in bold are <0.05.

Variable Df SS MS F P

Cover (%) (rank of proportional change)Treatment 2 5135.1 2567.6 2.65 0.08Pseudoblock 2 202.5 101.3 0.10 0.90Short woody cover 1 1716.4 1716.4 1.77 0.19Tall woody cover 1 822.9 822.9 0.85 0.36Litter 1 25.3 25.3 0.03 0.87Treatment × Pseudoblock 4 8347.6 2086.9 2.16 0.08Treatment × Short woody cover 2 13170.7 6585.3 6.80 0.02Treatment × Tall woody cover 2 4842.5 2421.2 2.50 0.09Treatment × Litter 2 6327.6 3163.8 3.27 0.04Residuals 95 91942.7 967.8

Number of flowering stalks (rank of proportional change)Pseudoblock 2 5409.2 2704.6 2.99 0.06Relative elevation 1 5666.0 5666.0 6.26 0.01Litter 1 2269.5 2269.5 2.51 0.12Treatment 2 424.9 212.4 0.23 0.79Short woody cover 1 2647.2 2647.2 2.92 0.09Tall woody cover 1 772.1 772.1 0.85 0.36Treatment × Short woody cover 2 11905.3 5952.7 6.57 <0.01Treatment × Tall woody cover 2 5340.1 2670.1 2.95 0.06Residuals 100 90580.1 905.8

Number of inflorescences (difference of ranks)Short herbaceous cover 1 4171.5 4171.5 2.68 0.10Treatment 2 9724.6 4862.3 3.12 0.05Pseudoblock 2 18646.1 9323.1 5.98 <0.01Relative elevation 1 1432.0 1432.0 0.92 0.34Litter 1 472.4 472.4 0.30 0.58Treatment × Pseudoblock 4 30827.8 7706.9 4.95 <0.01Treatment × Relative elevation 2 7658.1 3829.0 2.46 0.09Treatment × Litter 2 8946.2 4473.1 2.87 0.06Residuals 97 151103.0 1557.8

Height (cm) (difference of ranks)Treatment 2 10166.5 5083.2 4.24 0.02Pseudoblock 2 11108.1 5554.1 4.63 0.01Short woody cover 1 2881.4 2881.4 2.40 0.12Tall woody cover 1 2050.4 2050.4 1.71 0.19Treatment × Pseudoblock 4 10968.7 2742.2 2.29 0.07Treatment × Short woody cover 2 6756.0 3378.0 2.82 0.07Treatment × Tall woody cover 2 8630.3 4315.1 3.60 0.03Residuals 98 117553.3 1199.5


Table 9. Comparison of significant (" = 0.05) treatment effects on vegetation and Sidalceanelsoniana in 1999, after one year of treatment, and in 2000, after two years of treatments.

VariableSignificant treatment effects

One year of treatments Two years of treatments

Vegetation

Canopy cover (%) Mow < Burn < Control Mow, Burn < Control

Tall woody cover (%) — Burn, Mow < Control

Short woody cover (%) — —

Tall herbaceous cover (%) Burn, Mow > Control Burn > Control

Short herbaceous cover (%) Burn > Control Burn > Control

Litter depth (cm) — Burn < Control

Sidalcea nelsoniana

Cover (%) — Burn < Control

Flowering stalks (number) — —

Inflorescences (number) — —

Height of tallest flowering stalk (cm) — —


Figure 1. Aerial photograph of the Burned Swale study area within W. L. FinleyNational Wildlife Refuge. Image courtesy of the U.S. Geological Survey andTerraServer USA, taken May 7, 1994.


-8 -6 -4 -2Relative elevation (m)

-3-2-1012

-3-2-1012

-3-2-1012

sqrt

(Litt

er 2

000)

- sq

rt (L

itter

199

8)

Burning

Mowing

No manipulation

Wet Dry

Figure 2. Litter depth declined in wetter microhabitats and with burning.


0 5 10 15 20 25Short woody cover (%)

-2

0

2

4

6-2

0

2

4

6-2

0

2

4

6

Prop

ortio

nal c

hang

e in

S. n

elso

nian

a co

ver

Burning

Mowing

No manipulation

Figure 3. Sidalcea nelsoniana cover tended to increase slightly in no-manipulation plots whenin the presence of short woody cover. S. nelsoniana cover tended to decrease in burned plotswith short woody cover.


10 30 50 70Tall woody cover (%)

-200

-100

0

100

-200

-100

0

100

-200

-100

0

100

rank

(Hei

ght 2

000)

- ra

nk (H

eigh

t 199

8)

Burning

Mowing

No manipulation

Figure 4. Tall woody cover suppressed height growth of S. nelsoniana in the absence ofmanipulation. Mowing where tall woody cover was present released S. nelsoniana fromsuppression.


Dry end Middle Wet endPseudoblock

-100

-50

0

50

100

rank

(Hei

ght 2

000)

- ra

nk (H

eigh

t 199

8)

Figure 5. S. nelsoniana grew taller in the wetter end of the study area.


Appendix 1: Hypothetical example illustrating the General LinearModeling approach to hypothesis testing

Symbols

S seed production (seeds per plant)Ti treatment (1=burning, 2=mowing, 3=no manipulation)H hydrologic regime (continuous variable)L litter depth (cm)Ch1 Cover of herbaceous plants (low stratum)Ch2 Cover of herbaceous plants (high stratum)Cw1 Cover of woody plants (low stratum)Cw2 Cover of woody plants (high stratum)

Hypotheses

H0: hydrologic regime is unimportant to seed production

H1: hydrologic regime is needed to understand seed production

Models

Model 0 (corresponding to H0 true):

Model 1 (corresponding to H1 true):

Generalized F test statistic

where and are the coefficients of determination from models 0 and 1, respectively. If

F > Fcrit, then H0 is rejected: hydrologic regime is needed to understand seed production.

response to two years of restoration techniques in an

Documents