Journal of Vegetation Science 14: 25-34,2003: @ 1A VS; Opulus Press Uppsala,

25i

i

The importance of environment vs. disturbancein the vegetation mosaic of Central Arizona

Huebner, Cynthia D.I,2* & Vankat, John L.I

rJ Department of Botany, Miami University, Oxford, OR 45056, USA; 2Current address: Northeastern Research Station,

USDAFS, Morgantown, WV 26505, USA; .Author for correspondence; Fax + 13042851505; E-mail chuebner@fsJed.us

Abstract. The vegetation of centr~l Arizona is a mosaic of Introductionfour vegetation types: chaparral, chaparral grassland, wood-

, land, and woodland grassland, We analysed ten environmental Vegetation science has a long history of studies on! variables, three disturbance variables, and five disturbance the influence of environmental variables and distur-

indicators to answer the question: What is the relative impor- b tatJ"atte (Pickett & White 1985).f ' d d. b ' I ' , th ance on vege on p rntance 0 envuonment an Istur ance In exp aIling e veg- M d. h h . th 1 tJ.. .any stu Ies ave soug t to eXaInlne e re a ve Im-

etatlon pattern of our study area? We found that chaparral, ...

i chaparral grassland, and woodland are differentiated prima- ~o~ce of envIronment. ~d di~turbance, foCUSIng o~

.I rily by environmental factors and have high stability in the IndIVIdual plant commumtIes (0 Connor & Roux 1995,I. landscape. In contrast, woodland grassland is differentiated Miller & Halpern 1998; Leach & Givnish 1999; Whitef primarily by disturbance and is likely an early-successional et al, 2001) or a mosaic of plant communities across a

stage of woodlands, Although other researchers have indi- landscape (Naveh 1967; Harmon et al. 1983; Brosofskei cated that semi-arid vegetation is generally unstable, the veg- et al. 2000; Woods 2000). We compare environment vs.: et,ation of central Arizona is co~~osed of t~o syst~ms: .those disturbance in the semi-arid landscape of central Ari-: WI~ a more stable landscape pOSItI~n deterrmned pnmanly by zona, which is a mosaic of chaparral shrubland, wood-: envuonmental factors and those WIth a less stable landscape 1 d d I d Th d ' b d. d; ., d t ' d ' . 1 b d ' t b " t an an grass an. e IStur ances we stu Ie arei pOSItIon e errmne pnman y y IS ur ance lac ors.

common to semi-arid vegetation throughout the world,[' (e.g. Noy-Meir et al. 1989; Pickup & Stafford Smith

Keywords: Canonical Correspondence Analysis; Chaparral; 1993; Dodd 1994; Milton et al. 1994).I Conv~rsion; Fire; Grassl~d; Grazing;. Multi-Response Per- The e~vironmental determinants of the vegetation of

mutatIon Procedure; StabIlIty; SuccessIon; Woodland. central Arizona are not well known. In general, chaparral.I

[c' is bordered by woodlands at higher elevations and

I. grasslands at lower elevations (Lowe 1977; Carmichaelf Nomenclature: Kearney & Peebles (1960) and McDougall et al. 1978), but these vegetation types commonly occurI (1973), .d th .. al I . d . il' outsl e elf typIC e evatIon zones an on vanous so; types, slope aspects, and slope inclinations (Lowe 1977;I

Abbreviations: A = Grazing allotment size and chance cor- Saunier & Wagle 1967; Bedell 1987). Water and nitrogenrected within-agreement; C = Number of permitted domestic have been considered the most important limiting re-grazing animals; CCA = Canonical Correspondence Analysis; sources for each of these vegetation types (WoodmanseeDCA = Detrended Correspondence Analysis; MRPP = Multi- 1979; Klopatek & Klopatek 1987; Ellis & KummerowResponse Pennutation Procedure; NMS = Non-Metric Multi- 1989; Vankat 1989; Stephens & Whitford 1993; Weber etdimensional. ~caling; S = Fire size class; T = Grazin.g period; al. 1999) and may help explain the patchy mosaic. WaterW =k Prol babIlIty that rue burned a stand or a converSIon event needs are 230-500 mm.yrl for grasslands, 300-500too p ace. I

mm.yrl for woodlands, and 330-630 mm.yr forchaparral (Bolander 1981). Total soil nitrogen (10 cmdepth) has been estimated to be 6600 kg.ha-1 in wood-lands (Neary et al. 1996), 1313 kg.ha-1 in chaparral(Wienhold & Klemmedson 1992), and likely intermedi-ate in grassland (Emmerich 1999).

The disturbance factors likely to have affected thecurrent distribution of vegetation in central Arizona arelivestock grazing, fIre, and conversion (removal of woodyplants to produce grassland). Livestock grazing becameimportant in Arizona about 200 yr BP and coincided

~

!

i: 26 Huebner, C.D. & Vankat, J.L.

with increases of woodlands (Johnson 1962; O. Davis or J. monosperma, with some Pinus edulis. Both grass-1987) that may be ongoing (Bahre 1991). Little is known land types are extensive, with chaparral grassland domi-about the past role of fire in central Arizona; however, nated by Hilaria belangeri and woodland grasslandLeopold (1924) speculated that fire had maintained dominated by an exotic annual, Bromus tectorumArizona grasslands and restricted woodlands to areas of (Huebner 1996).rocky soils and rough topography. Fire prevention andsuppression were common in the 20th century until at Sampling designleast 1980, when prescribed burning became common(Wright & Bailey 1982; Bahre 1985; Miller 1999). We focused on grassland patches within chaparralConversion occurred since at least 1960 (Aro 1971; and woodland vegetation. Therefore, we sampled 30Sheridan 1995) and usually involved changing wood- stands of chaparral paired with 30 stands of grassland,

! land to grassland, with tree removal followed by re- and 30 of woodland paired with 30 of grassland (total of! peated prescribed burning (Johnson J999). Conversion 120 paired stands; Huebner 1996). These stands wereI of chaparral to grassland has been less common selected from recent aerial photographs, followed by a

(Longstreth & Patten 1975; E. Davis 1987). site visit, such that (1) the paired stands were adjacent toWe address the question: What is the relative impor- one another and (2) each stand was ~ 0.4 ha (although

tance of environment and disturbance in explaining the variable in shape). We sampled nearly all stands thatvegetation pattern of our study area? If the types occur met these two criteria.in similar environments, they may represent alternative Within each stand, we determined the cover of woodystable states or different successional stages of a devel- species using the line intercept method (Bonham 1989)opmental sequence (Clements 1936; Holling 1973; Suth- along a 50 m line transect placed parallel to the topo-erland 1990). Regardless of their stability, they may be graphic contour and at least 20 m from stand edges. Onebest explained by past disturbance (Connell & Sousa end of the transect was permanently marked with a 0.5-1983; Johnson & Mayeux 1992; Sullivan 1996). m metal stake. We visually estimated the cover of each

herbaceous and suffrutescent species (to ca. 5% accu-racy) in 10 quadrats (0.5 m x 0.5 m) randomly located

Material and Methods along each transect.

i Study area Environmental variablesii We conducted our study within ca. 1500 km2 of the We recorded slope aspect, inclination, and elevationr Chi?o V alley ~d V erde ~anger Dis~cts of Prescott at ~e midpoint of each transect (values were gen~rallyf NatIonal Forest m YavapaI County, Arizona (Huebner unIform along transects). The north-south coordmatesj et al. 1999). This area is located in the Tonto Transition (herein referred to as latitude) were determined in me-I Section of the Colorado Plateau Semi-Desert Province ters using the Universal Transverse Mercator projec-, in the Dry Domain (Bailey et al. 1994). The topography tion. Angular data for aspect (O) were linearized accord-i is variable, with slopes ranging from flat plains to steep ing to the following equation: SIN(0-135) + 1 (modified

hillsides and elevations ranging from approximately from Huebner et al. 1995), where larger values represent1100 to 2500 m (Huebner 1996). Mean annual precipi- drier slopes (west and south-facing) and smaller valuestation is 250-650 mm and is concentrated in two sea- represent more mesic slopes (east and north-facing).sons, late-summer and winter (Cross et al. 1960; Bailey We collected a soil sample of 3 cm diameter and 10et al. 1994). The soil is generally low in organic matter, cm depth from the interspace between woody plants that

: which along with relatively high volatilization rates was nearest each quadrat (sampling in interspaces re-i results in low levels of nitrogen (Cross et al. 1960). duces sample variability caused by removal of nutrients

Central Arizona is characterized by four vegetation by woody plants; Klopatek & Klopatek 1987). All 10types: chaparral, woodland, chaparral grassland (grass- samples from each stand were combined, mixed, passedland patch within a chaparral matrix), and woodland through a 2 mm mesh sieve, and dried immediately in agrassland (grassland patch within a woodland matrix). forced-air oven at 40 °Covernight.Chaparral is a broad-sclerophyll shrubland usually domi- Subsamples of each soil sample were passed throughnated by Quercus turbinella and often including Rhus .a 0.5-mm sieve. Those used for nitrogen and carbontrilobata, Arctostaphylos pungens, and Cercocarpus analyses were kept frozen (to reduce nitrification andmontanus (Huebner 1996). Woodlands in this region are denitrification) until immediately before analysis. Totalmore widespread (Spang 1987; Evans 1988) and tend to nitrogen (organic and inorganic) and carbon were deter-be dominated by small trees of Juniperus osteosperma mined using a LECO CNS-2000 Carbon, Nitrogen and

---I

I

i-The importance of environment vs. disturbance in the vegetation mosaic of Central Arizona -27!

Sulfur Analyzer, which measures percent nitrogen and stand being within the fire boundaries: S(S x W), wherecarbon (converted to N2 and CO2, respectively) for a S is the size class of the fire and W approximates thegiven mass of soil via thermal conductivity of the gases. probability that the fire burned the stand. We includedAnalyses were conducted by IAS Laboratories of Phoe- fIre size in our index because size is often correlatednix, AZ. We determined soil texture by the hydrometer with fIre intensity in our study area. Our estimates of Wtechnique (Bouyoucos 1927) and estimated soil water- were 0.3 for fires from 1966-1971 when fire locationsholding capacity as mass loss from a sample weighed had to be estimated from fire names, 0.5 for fires whosewhen saturated and again when dried at 110 DC for 24 h centers were recorded within 1.6 km of the stand, and(Stock & Lewis 1986). Although soil structure is impor- 1.0 for fires mapped as having covered the stand. Wetant to water-holding capacity, we used sieved soils to averaged the annual fire index for 1946-1993.avoid variation in soil structure introduced by our sam- We examined conversion using records from bothpIing and transporting of soils. Ranger Districts. Record keeping began in 1946, and the

.records included only the date and location of conver-Disturbance factors sions; size was recorded inconsistently. For each stand,

we calculated an annual conversion index (rate) byWe considered livestock grazing between 1930 (when summing the conversion attempts (usually 0 or 1). We

records were first kept for our entire study area) and averaged the annual conversion index for 1946-1993,1993 using records of the two Ranger Districts. Grazing but included in this calculation additional, undated,records included the number of livestock allowed for unrecorded conversion events. Evidence for these addi-each of the 16 grazing allotments (sectors). The number tional conversion events was our personal observation

: of permitted livestock was likely slightly higher than the of stumps or slash in stands with no written record ofJ actual number, but actual numbers were available for conversions. Because the presence of stumps and slashj most years after 1950. In addition, the grazing records could reflect only partial clearing of woody vegetation! noted the length of the grazing period and the area of and not complete conversion and because some of thesei each grazing allotment. Records on type of livestock conversions may have occurred before 1946, we

I and season of grazing were inconsistent, and no records downweighted these unrecorded conversions by esti-1 were kept on native grazing animals such as deer, elk, mating the probability of the stand being subject to; and other herbivores. complete conversion: W. Our estimates ofW were 0.3 ifI

j For each grazing allotment, we calculated an annual the stand had tree stumps but lacked slash and 0.5 if thej ~azing index in animal months p.er ha using the fol~ow- stand had stumps and slash.1 mg fo~ula: (C. x T)/~, where C!S number o~ permlt~ed..domestIc grazIng anImals, T IS the grazIng penod Disturbance indicators'I (months), and A is the size (ha) of the grazing allotment.

The annual grazing index was averaged for 1930-1993. Various limitations in the records of grazing, fire,Grazing allotments were much larger than a single stand, and conversion restricted assessment of disturbance.so we were forced to treat all stands within a given Therefore, we also computed four disturbance indica-allotment as having the same grazing pressure. There- tors: vegetation stability, number of nearby vegetationfore, differences in grazing between paired stands (i.e., a types present, number of patches present, and fractalchaparral stand and its adjacent grassland stand) usually dimension.could not be estimated. For vegetation stability, we assumed an inverse rela-

We obtained fire records from both Ranger Districts tionship with disturbance because the half-century pe-and the Prescott Fire Center (at the Henry Y.H. Kim riod (for which aerial photographs of our landscape areAviation Facility). Consistent record keeping began in available) is likely too short to include significant cli-1946. Fire locations were estimated from fIre maps for matic change and resultant vegetation shifts. To calcu-1946-1965, from written descriptions for 1965 and 1972- late vegetation stability, we constructed transition ma-1993, and from fIre names (based on geographic features) trices for 1940-1989 based on aerial photograph over-for 1966-1971. Fire size was recorded by class: 1 if < 0.1 lays (cf. Huebner et al. 1999) for circular 1 km2 areasha, 2 if 0.1-4.0 ha,3 if 4.1-40 ha, and 4 if 41-100 ha (no surrounding the center point between each pair of stands.larger fires were recorded for our stands). Records also We defined stability of the dominant vegetation (hereaf-separated wild fires (lightning- and human-caused) from ter 'dominant stability') as the probability of the vegeta-prescribed fIres. Ratings of fire intensity were available tion of a stand remaining unchanged and stability of allafter 1980; however, rating scales were inconsistent. vegetation (hereafter called 'overall stability') as the

For each stand, we calculated a unitless annual fire mean of the stabilities of all vegetation types within theindex that reflected fire size and the probability of the 1 km2 area. Therefore, paired stands can be differenti-

28 Huebner, C.D. & Vankat, J.L.

ated from each other by dominant stability, but not by Resultsoverall stability.

The relationship of number of vegetation types, A total of 169 taxa were sampled in the 120 standsnumber of patches, and fractal dimension to disturbance (Huebner 1996). A multi-response permutation proce-is likely landscape-specific (Turner 1989; Sughara & dure (MRPP), with the plots grouped by vegetation type,May 1990). For central Arizona, we assumed an inverse documented that the four vegetation types were signifi-relationship between these three variables and distur- cantly different in species composition (p = 0.000; the

bance because aerial photographs of our region indi- chance corrected within-agreement (A) is 0.356).cated that areas known to have been disturbed had fewer Among the four vegetation types, woodland had thepatches and more linear vegetation boundaries than highest mean richness and diversity but the lowest meanareas with less disturbance (Huebner 1996; Huebner et plant cover (Table 1). Chaparral had the lowest meanal. 1999). richness and diversity and the highest total cover. The

Using a 1 km2 circular area cent~red on each pair of two types of grasslands were generally intermediatestands, we summed the number of vegetation types, (Table 1).summed the number of patches, and calculated fractal Bonferroni (-test analysis of the means of the envi-dimension by regressing summed patch area against ronmental variables showed that the four vegetationsummed patch perimeter (measured with IDRISI 4, a types had similar slope aspect, elevation, % clay, waterraster-base GIS package) and multiplying by 2 (O'Neill holding capacity, nitrogen, and carbon (Table 2). Inet al. 1988). For each of these three disturbance indica- contrast, chaparral and chaparral grassland were gener-tors, we averaged values for the 1940-1989 time period. ally associated with steeper slopes, southern latitudes,Unfortunately, these indicators do not differentiate be- higher % sand, and lower % silt than woodland andtween paired stands. woodland grassland; however, there were no statisti-

cally significant differences between grasslands andtheir associated woody vegetation type.

Data analysis Analysis of the means of the disturbance variablesshowed that the four types of vegetation had similar graz-

We expressed species composition using absolute ing and fire indices (Table 2). The woodland grassland,i cover values and calculated species richness as the; number of species present and diversity as an inverted ...t ., Table 1. A. SpecIes WIth ~ 2% mean cover In chaparral,I SImpson s Index (the InVerSIOn allows expressIon of h al I d dl d d dl d I d Tt c aparr grass an ,woo an, an woo an grass an. 0-;. dIversity m te~s of spe~les e~ulvalents; ~~et 1974). tal cover averaged 76 %, 66 %, 57 % and 63 %, respectively.i We evaluated differences m species composItion among Dominant species in bold. B. Diversity and cover parameters.

vegetation types using multi-response permutation pro- c W dl d..., A. haparral 00 ancedure (MRPP; Blondim et al. 1985; Zimmerman et al. Species Chaparral Grassland Woodland Grassland

1985;MJMSoftwarePC-ORDv.4;Anon.1999b)' Q tub . lla 356uercus r IRe .We compared environmental and disturbance vari- Gutierreziasarothrae 5.2 6.8 2.4 5.2

abIes among vegetation types using two complementary Rhus trilobata 4.5.Bouteloua curtipendula 3.8 4.0 2.7 2.5

analyses, Bonferrom (-tests (Anon. 1990; Anon, 1999a) Hilaria belangeri 3.2 11.1and MRPP. We compared the importance of environ- Eriogonumwrightii 2.6 7.9

ment vs, disturbance in determining vegetation types MB~utelouab.gra~llis 22.52 3.3 2.6 5.5Imosa lunc!lera .

using canonical correspondence analysis coupled with Arctostaphylos puRgeRS 2.2correlation analysis (CCA; ter Braak 1988; Anon. 1999b; Cea?othus greggii 2.2

.JunIperus osteosperma 17.5Bonferroni (-tests were used to compare mean ordina- Cordylanthus laxiflorus 2.3tion scores of vegetation types), Other ordination meth- Aristida spec. 2.2 2.2d . 1 d . D d d C d Anal .Bromus tectorumJ 10.6

0 s, mc u mg etren e orrespon ence YSIS Hilaria mutica 3.7 3.9

(DCA), indirect CCA (final scores derived from species Viguiera annua 5.9scores instead of a linear combination of environmental Bromus rubensJ 5.2

..,.., Muhlenbergia repens 2.9variables), and Non-metric Multi-dimensional ScalIng Plantagopurshii 3.3(NMS), produced similar results, so only CCA is pre- Lotus humistratus 2.1

..Festuca octo flora 2.0sented here, We chose CCA because It gave easIlyrepeatable results, did not appear to be affected by noisy ~ean richness2 16.1 14.2 14.8 15.8data, and included a combined effect of all variables on Diversity3 5.4 3.9 4.9 5.4the plot and axes ordination scores (Minchin 1987; Totalcover(%) 76 66 57 630k1and 1996; McCune 1997). JExotic annual; 2Number of species; 3Species equivalents

I

I -The importance of environment vs. disturbance in the vegetation mosaic of Central Arizona -29!

Table 2. Means for environmental and disturbance variables. An MRPP indicated that, in comparison to wood-

Different letters indicate statistically significant differences land and woodland grassland, chaparral and chaparral

: among vegetation types by Bonferroni (-tests (p-value < 0.05). grassland had significantly steeper slopes, higher eleva-, Slope aspect was calculated using Sin (0 -135) +1. tion lower % silt lower latitude lower conversion

, , , ,Chaparral Woodland higher dominant and overall stability, and higher fractal

Chaparral Grassland Woodland Grassland d.. (p 0 01 T bl 3) I ddi .: h I h dlIDenSlon ~. ; a e .n a uon, c aparra aEnvironmental variables: higher percentages and and higher dominant stabilitySlope aspect 0.84 A 0.97 A 0.84 A 0.53 A .Slope inclination (%) 17.7 A 12.5 AB 9.0 BC 2.9 C than chaparral grassland, woodland grassland had higherElevation (m) 1451 A 1438 A 1432 A 1409 A conversion than woodland and woodland had higher %Latitude (m) 3827 B 3827 B 3854 A 3857 A .'% Sand 52 A 44 AB 41 B 37 B carbon than eIther chaparral or chaparral grassland.

i % Silt 30 BC 25 C 37 A 36 AB A CCA of our 120 stands using environmental vari-% Clay 23 A 28 A 23 A 24 A bl b d . b . bl I d ... Water Holding Capacity (g) 31 A 34 A 34 A 34 A a es ut no Istur ance vana es resu te m a statlstl-

% Total Nitrogen 0.145 A 0.138 A 0.127 A 0.117 A cally significant separation along axis 1 of chaparral and% Carbon 1.87 A 1.57 A 2.38 A 2.19 A chaparral grassland stands (right side of the ordination

Disturbance variables:fi ) fr dl d d dl d I d dGrazing index 0.17A 0.17A 0.17A 0.16A Igure om woo an an woo an grass an stan S

Fire index 0.023 A 0.028 A 0.026 A 0.035 A (left side), but did not separate either woody vegetationConversion index 0.0002 B 0.0007 B 0.0022 B 0.018 A from its associated grassland along any of axes 1-3 (Fig.

Dominant stability (%) 0.82 A 0.77 AB 0.59 B 0.59 B

Overall stability (%) 0.62 A 0.68 A 0.46 B 0.49 B 1, Table 4). Axis 1 is most strongly correlated withFractal dimension 1.067 A 1.056A 1.049A 1.035 A latitude (r=-0.883), % silt(-0.659),andslopeinclina-

Number of vegetation types 4.5 A 4.3 A 4.2 A 3.9 A . (0 620 T bl 5) A . 2 h. h didNumber of patches 18A 17A 13A 12A tion .; a e .XIS ,w lC not separate any

of the vegetation types and therefore is related more to

differences within vegetation types rather than among

however, had higher rates of conversion than the other types, is most strongly correlated with % clay (0.782),

vegetation types. water holding capacity (0.764), % sand (- 0.631) and %

Analysis of the means of disturbance indicators carbon (- 0.594). cshowed that the four types of vegetation had similar A second CCA using disturbance variables but no I

fractal dimensions, number of nearby vegetation types, environmental variables also resulted in a statistically

and number of patches (Table 2). In contrast, chaparral significant separation of chaparral and chaparral grass-

had significantly higher stability (dominant and overall) land stands from woodland and woodland grassland

than both woodland and woodland grassland, but had stands along the first axis (Fig. 2, Table 4). In addition.

similar stability as its associated grassland. Chaparral this axis also separated woodlands from woodland

grassland had higher overall stability than both woodland grasslands. Axis 1 is most strongly correlated with

and woodland grassland. overall stability (- 0.776), conversion (0.760), and domi-

nant stability (- 0.631; Table 5). Axis 2, which also

Table 3. Multi-Response Permutation Procedure (MRPP) or- differentiates some of the vegetation types, is most

ganized by vegetation type and variable. Vegetation types strongly correlated with dominant stability (0.373) andwith different letters are significantly different from each conversion (0.566).

other (p < 0.01). Because the T (observed-expected/standard A third CCA using environmental and disturbance

: deviation of expected) stat~stic and A (chance c~rrected wi~hin variables also resulted in a statistically significant sepa-i group agreement) values dIffer for each vegetatIon type paired

comparison, the presentation of these data in a single, simple

table is impossible. Only variables showing significant differ- Table 4. Means axis scores of the vegetation types. Different

ences in one or more vegetation types are listed. letters indicate statistically significant differences among vegeta-

Iw tion types as determined by Bonferroni (-tests (p < 0.05). OnlyChaparral oodland h .. gnifi diffi beI Chaparral Grassland Woodland Grassland axe~ s owrng SI cant erences tw~n ~ne or more veg-

1c"! ..etatlon types are shown; other axes are not sIgnificant..I EnvIronmental varIables: Slope inclination A A B B Chaparral Woodland; Elevation A A B B Ordination Chaparral Grassland Woodland Grassland

Latitude B B A A .% Sand A B B B Environment% Silt B B A A Axis I 0.5897A 0.4789A -0.5716B -0.5173B% Carbon C C AB AC Disturbance

Disturbance variables Axis I -0.3804C -0.3462C 0.1392B 0.6l90AConversion index C C B A Axis 2 0.046lBA -0.0027BC -0.26l6C 0.3120ADominant stability ABC C Environment and disturbanceOverall stability A A B B Axis I 0.5984A 0.4797A -0.5694B -0.5l97BFractal dimension A A B B Axis 3 -0.02533B -0.03385B -0.37008C 0.52056A

30 Huebner, C.D. & Vankat, J.L.

0 04

0 0

0

4e 04 0

4 '" 4 0 0i .44 4 .: ". 4£ 40 £ £ 4

.0 0 £WHCAP "'CLAY ~ £ 0~ £ '" 0 00. t 0 ..£~ 4. 0 0

..44 £ £ 4 0 0! N ~. "'SILT 4 i'£ ';;::-4 DOMSTAB 0 0 0'c ~ .0 .4 4 ~ 'x " --~::~~.://~O ONVERSION 0:< x LAmUDE INCLI~ 4 < OVE~. .('": < 0 4 4 r 4 £ STAS£: . 00 .44 4 £Oe 0 4 £4 ~ 4

.~ ..~ £ % SAND '" £ 4 4l £ ~ ~ ~ ~ £ £000.0 "CARBON £ £ £ 4 0 £.. ~" 44 . ' 00.0 0 £ 4"££ 0" 4. 4 .

0 .0 .£ £ .4 4 4": O.0 j £..4. 4 00 .

£ .4ce ... ..£...

Axis 1 Ai<is 1

Fig. 1. CCA ordination on environmental variables only. Fig. 2. CCA using only disturbance variables. Eigenvalues forEigenvalues for axes 1-3 are 0.492, 0.297 and 0.171. Total axes 1-3 are 0.343,0.185 and 0.164, respectively. Total vari-variance in the species data is 10.4 and 9.2 % of this is explained ance is 10.4 and 6.6% of this is explained by the disturbanceby the environmental variables. INCLIN = slope inclination; variables. DOM STAB = dominant stabilitY; OVERALL STABWHCAP = water holding capacitY. Woodland = 8; Woodland = overall stability. Woodland = 8; Woodland Grassland = 0;Grassland = 0; Chaparral = A; Chaparral Grassland = L::,.. Chaparral =.A.; Chaparral Grassland = L::,.. [

: ration of chaparral and chaparral grassland stands from 1989). Therefore, degree of bedrock fracturing may bewoodland and woodland grassland stands along the first important in separating chaparral from chaparral grass-axis (Fig. 3, Table 4). Axis I is most strongly correlated land. Also, although nitrogen is likely the most importantwith latitude (- 0.871), % silt (- 0.649), overall stabil- limiting nutrient in Arizona, phosphorus and calcium,ity (0.609), and slope inclination (0.604; Table 5). which were not measured, have been noted as important

I Axis 2 did not separate any of the vegetation types (Fig. for some species and communities (Ellis & KummerowI 3, Table 4) and therefore is related more to differences 1989; Kramer 1999; Quideau 1999). However, if no

.\ within rather than among vegetation types. It is most important variables have been neglected, chaparral and} strongly correlated with environmental variables: % clay chaparral grasslands as well as woodland and woodland

J\, (0.779), water holding capacity (0. 755), ~ sand (- 0.604), 'tJfasslands may re~esent multiple stable states, depend-, 11 ' and % carbon (- 0.578; Table 5). AXIS 3 produced a mg on the role of disturbance." statistically significant separation of woodland from Grazing and fire indices did not account for vegeta-

! woodland grassland (Fig. 4; Table 4) and is strongly tion differences; however, our characterization of distur-! correlated only with conversion (0.728; Table 5). bance, especially grazing and fire, was limited by the fact

that the records were incomplete and imprecise. Forexample, data on seasonal variation in grazing were lack-

Discussion ing, yet this variation could be significant, especially forgrassland vegetation (Valone 1999). Also, the scale of

Most analyses showed that environmental variables records was generally too broad to explain differences indid not separate either chaparral or woodland from its vegetation of adjacent paired stands. For example, theassociated grassland; however, it is possible that other, grazing records are for areas several times larger than theunmeasured environmental variables could provide this areas of paired stands. Therefore, if grazers were concen-separation. For example, chaparral shrubs such as Quercus trated in grassland patches, this preference would notturbinella have deep roots that can penetrate rock frac- have been detected with our estimate of grazing.tures to ground water (Saunier & Wagle 1967; E. Davis & Despite statistical similarity in our fire index, therePase 1977; E. Davis 1989), a potentially critical factor is evidence that fire, at least when restricted to onebecause periodic droughts characterize this region (Vankat vegetation type in paired stands, is more frequent in

~

,i

-The importance of environment vs. disturbance in the vegetation mosaic of Central Arizona -31

0 0

00

6 0.0. i ..

0 60 66 6 0

6. 0, ..0; .0..0 0 WHCAP :iCLAY. .., 6

';;: .00 ~ ..21 00 ~O '0 A .-6 6 ~ 0

~ ..0 ~6 0 o. ~0 OVERAli"ST';/ ~ C~ERSION

.0 % SILT 6 .66 0 ."., .". O' .! ..'CONVERb1C>N INCUN 6"~ 6 0 ~ ~~SlLT '.. """s~At. "6

0 .S~ '& ..LAm *""" A(J' ~

.." OVERAlL~AB0 LATITUDE CARBON'k SAN D ...". 6 ,( ." . . ..0 .... 0 O .." .." .CU ." .

...0 .6I 0 0 0 6. 0 ..0. " ..6., O. 6... 6~ 6.-" 6

.O'0 0 ..Ico O. ,. . ./...! .I .Axis 1

; Axis 1 Fig. 4. CCA using all variables (axis 1 and 3). See Fig. 3

: Fig. 3. CCA using all variables (axes 1 and 2). Eigenvalues for caption.: axes 1-3 are 0.5Q(), 0.315 and 0.243. Total variance is 10.4 and

10.2% of this is explained by all variables. WHCAP = water

holding capacity; INCLIN. = slope inclination; DaM STAB = , dominant stability; OVERALL STAB = overall stability. Wood- chaparral IS tYPIcally less dense than CalIfornIa chaparral

: land = e; Woodland Grassland = 0; Chaparral = .A.; Chaparral (Bolander 1981).

: Grassland = ~ Our finding that woodland grassland had a higher

, conversion index than the other vegetation types was not

surprising. Conversion has been used primarily to at-

grassland, with 11 in woodland grassland, eight in tempt to form grassland from woodland (cf. Arnold 1964).

chaparral grassland, six in woodland, and one in The two stability measures and fractal dimension

chaparral. The reason Arizona chaparral is so stable and may be more sensitive indicators of disturbance in our

1; does not appear to require fire for self perpetuation, study area than number of vegetation types and patches. .

unlike some California chaparral stands (V ogle 1981; Also, our finding of differences in disturbance indicator

Hilbert & Larigauderie 1990), may be that Arizona variables where there were no differences in grazing,

.Table 5. Pearson correlations (r) of environmental and disturbance variables with ordination axes for CCAs with environmental,

, disturbance, and environmental+disturbance variables. The sign in front of each r-value indicates direction of correlation. In CCA,

p-values are not provided for each variable, but a Monte Carlo test was used 10 determine the significance of each axis. The

, correlations for the Environmental and Environmental + Disturbance CCA are significant at the 0.01 level for all three axes.

, However, the correlations for the Disturbance CCA hadp-values of 0.01 for axis 1,0.06 for axis 2 and 0.13 for axis 3. Only variables

f with r-values of 0.5 or greater (i.e. the most important variables) for at least one axis of any of the three CCAs are shown.

I; Environmental or Disturbance CCA Environmental+Disturbance CCAII Axis] Axis 2 Axis 3 Axis ] Axis 2 Axis 3 .

Environmental variablesSlope inclination 0.620 -0.]95 0.]87 0.604 -0.]29 -0.326

! Latitude -0.883 -0.357 0.239 -0.87] -0.359 0.063% Sand 0.40] -0.63] -0.318 0.404 -0.604 -0.037% Silt -0.659 0.07] 0.020 -0.649 0.035 0.207%C]ay 0.076 0.782 0.409 0.062 0.779 -0.143Water holding capacity -0.117 0.764 0.278 -0.132 0.755 -0.176% Carbon -0.207 -0.594 -0.542 -0203 -0.578 -0.132

Disturbance variablesFire index 0.015 0.266 0.643 0.000 0232 0.]45Conversion index 0.760 0.566 -0.102 -0.487 -0.]74 0.728Dominant stability -0.63] 0.373 -0.340 0.503 -0.173 0.015Overall stability -0.776 0.248 0.023 0.609 0.001 -0.133

32 Huebner, C.D. & Vankat, J.L.

fire, and conversion reaffirms the shortcomings in the position in the landscape. At least in this sense, we

disturbance records or indicates that other disturbances conclude that the semi-arid vegetation of central Arizona

may be important. Fine-scale disturbances are impor- is composed of two systems: those whose landscape

tant in some arid and semi-arid areas, but were not position is more stable and determined primarily by envi-

examined in this study. For example, small mammals ronmental variables and those whose landscape position

and ants affect the distribution of species by moving is less stable and determined primarily by disturbance.

seeds, promoting seed germination with soil disturbance,

and enhancing plant growth with increased nutrientsaround nests (Huntly 1991; MacMahon 1997). Where Acknowledgements. This research would have been impossible

these plant species provide favorable forage for the ani- without support and funding from the ~taff of the Verde and

ma1s, positive feedback among populations may occur Chino R~g~r Di~trict.s ~fPrescott ~ational Forest and fu~~ing

(Wiens 1985) thereby reinforcing vegetation differences from MIamI Uruversity s Acadelll1c Challenge and WillIam: ' .Sherman Turrell Herbarium Fund grant programs. We particu-

In general, envlfonment IS more lmportant than dis- 1 I thank D M Ph D .. R f th V d D . ., ar y oug ac ee, IStnCt anger 0 e er e IS-

turbance m the central Arizona landscape (both among trict.

and within vegetation types). For example, environment

separates chaparral from the other vegetation types,

with chaparral occurring on lower latitudes, coarser References

soils, and steeper slopes than woodland and woodlandgrassland and on coarser soils than chaparral grassland. Anon. 1.990. SAS/STAT User's Guide, Version 6, 4th ed. SAS

The high stability of chaparral also indicates that distur- InstItute Inc., Cary, NC.

bance is relatively unimportant. Chaparral's stability Anon. 1999cah'. SYSTILAT v. 9. SPSS Science Marketing Depart-

1 f h . f hrub h Q ment. Icago, .resu ts rom t e resproutlng 0 s s suc as uercus A 1999b PC ORD M 1 .. t l . if 1 . 1..non.. -: u tlvana e ana ysls 0 eco oglca

turbmella and Arctostaphylos pungens, somethIng rec- data, v. 4. MjM Software Design, Gleneden Beach, OR.

ognized by land managers who only infrequently at- Arnold, I.E. 1964. The Pinyon-Juniper type of Arizona: Effects

tempt to convert chaparral to grassland (Longstreth & of grazing, fire. and tree control. USDA For. Serv. Produc-

Patten 1975; E. Davis 1987). tion Research Report No. 84, Washington, DC.

Only in the case of woodland grassland is distur- Aro, R.S. 1971. Evaluation of pinyon-juniper conversion to

bance more important than environment. Here conver- grassland. J. Range Manage. 24: 188-197.

sion is the key factor and instability also evidences the Bahre, C.I. 1985. Wildfire in southeastern Arizona between

importance of disturbance. There are many examples of 1859 and 1890. Desert Plants 7: 190-194.

..Bahre, C.I. 1991. Legacy of change: Historic human impact ondIsturbance-dependent grasslands, some of WhICh rep- .. h A . b d 1 nd U .. f Ari..vegetation In t e rlzona 07' er a s. mversity 0 -

resent successIonal stages of assocIated woody vegeta- zona Press Tucson AZ.

tion (cf. Weaver & Fitzpatrick 1934; Cusick 1981;Bock Bailey, R.G., Avers, F.E., King, T. & McNab, W.H. 1994.

& Bock 1992). Formerly, naturally-occurring woodland Ecoregions and subregions of the United States (map).

grassland in central Arizona was likely maintained by USDA For. Serv., Washington, DC.

fires which curtailed tree invasion. However, since the Bedell, T.E. 1987. Range management concerns on juniper

beginnings of overgrazing in the late 1800s (Cable woodlands. In: Everett, R.L. (ed.) Proceedings Pinyon-

1975) and fire suppression in the 20th century, the rate Juniper Conference, pp. 43~. USDA For. Servo Gen.

of tree invasion has increased (Bahre 1991) and is .T~~. Rep. INT-215, Ft. Collms, CO.

t d .. 1 b . I tr t h al Biondini,M.E.,Bonham,C.D.&Redente,E.F.1985.Secondarycoun ere pnmaTl y y converSIon. n con as c aparr ., ...

.' successIonal patterns m a sagebrush (Artemisia tndentate)grassland does not appear ~o be ~ successl0~~ stage of community as they relate to soil disturbance and soil biologi-

chaparral, because where mvaslon occurs It IS not by cal activity. Vegetatio 60: 25-36.

chaparral shrubs but rather by Juniperus osteosperma, Bock, I.H. & Bock, C.E. 1992. Vegetation responses to wildfire

J. monosperma, Prosopis juliflora and Acacia greggii in native versus exotic Arizona grassland J. Veg. Sci. 3: 439-

(Huebner 1996; Huebner et al. 1999). 446.Our finding that most vegetation types are defined Bolander, D.H. 1981. Chaparral in Arizona. In: Conrad, C.E. &

primarily by environment and are stable conflicts with the Oeche1, W.C. (eds.) Proceedings of the Symposium on Dy-

assumption that semi-arid systems are generally unstable namics and Management of Mediterranean-Type Ecosys-

(Johnson & Mayeux 1992; Sullivan 1996; Walker & tems,~p.60-63. USDA For. Serv. Gen. Tech. Rep.PSW-58,

W ' l 2002) H did al b ' li ' San Diego, CA.1 son .owever, we not ev uate sta 1 ty m Bonh C D 1989 M . fi t t . 1 .' am, ...easuremen,s or erres na vege,atzon.

the sense of recovery of species composition to equilib- Wiley, New York, NY.

rium after disturbance (Holling 1973; DeAngelis & Bouyoucos, G.I. 1927. The hydrometer as a method for the

Waterhouse 1987). Instead we examined stability in the mechanical analysis of soils. Soil Sci 23: 343-353.

sense of persistence ofa vegetation type in its geographic Brosofske, KD., Chen, I. & Crow, T.R. 2001. Understory

i-The importance of environment vs. disturbance in the vegetation mosaic of Central Arizona -33

vegetation and site factors: implications for a managed mental factors affecting understory diversity in second-Wisconsin landscape. For. Ecol. Manage. 146: 75-87. growth deciduous forests. Am. Midl. Nat. 134: 155-165.

Cable, D.R. 1975. Range management in the chaparral type and Huebner, C.D. 1996. Effects of disturbance and environment onits ecological basis: The status of our knowledge. USDA the chaparral-woodland-grassland landscape mosaic ofFor. Servo Gen. Tech. Rep. RM-155, Ft. Collins, CO. Central Arizona. Dissertation, Miami University, Oxford,

Carmichael, R.S., Knipe, O.D., Pase,C.P. & Brady, W.W.1978. OH.Arizona chaparral: Plant associations and eco./ogy. USDA Huebner, C.D., Vankat, J .L. & Renwick, W.H. 1999. Change inFor. Servo Gen. Tech. Rep. RM-202, Ft. Collins, CO. the vegetation mosaic of central Arizona USA between 1940

Clements,F.E. 1936. Nature and structure of climax. J. Ecol. 24: and 1989. Plant Ecol. 144: 83-91.252-284.. Huntly, N. 1991. Herbivores and the dynamics of communities

Connell, J.H. & Sousa, W.P. 1983. On the evidence needed to and ecosystems. Annu. Rev. Ecol. Syst. 22: 477-503.judge ecological stability or persistence. Am. Nat. 121: 789- Johnson, H.B. & Mayeux, H.S. 1992. Viewpoint: A view on824. species additions and deletions and the balance of nature. J.

Cross, J.L., Shaw, E.H. & Scheifele, K. (eds.) 1960. Arizona: Its Range Manage. 45: 322-333.people and resources. The University of Arizona Press, Johnson, P. 1999. Economics of redberry juniper control in theTucson, AZ. Texas Rolling Plains. J. Range Manage. 52: 569-574.

Cusick, A. W. 1981. The prairies of Adams County, Ohio: 50 Johnson, T.N. 1962. One:-seed juniper invasion of northernyears after the studies of Lucy Braun. In: Stuckey, R.L. & Arizona grasslands. Ecol. Monogr. 32: 187-207.Reese, K.J. (eds.) The prairie peninsula -in the 'shadow' of Kearney, T.H. & Peebles, R.H. 1960. Arizona Flora. UniversityTranseau: Proc. North American Prairie Conference, pp. of Califomia Press, Berkeley, CA.56-60. Ohio BioI. Surv. BioI. Notes 15. Klopatek, C.C. & Klopatek, J.M. 1987. Mycorrhizae, microbes

Davis, E.A. 1987. Chaparral conversion to increase streamflow and nutrient cycling processes in pinyon-juniper systems.in Arizona: Sequential treatments extend duration of nitrate In: Everett, R.L. (ed). Proc. Pinyon-Juniper Conference, pp.loss to stream water. Forest Sci. 33: 89-103. 360-364. USDA For. Servo Gen. Tech. Rep. 1NT-215, Ft.

Davis, E.A. 1989. Patterns in mixed chaparral stands differential Collins, Co.water status and seedling survival during summer drought. Kramer, S. 1999. Phosphorus pools in tree and intercanopyIn: Keeley, S.C. (ed) The California chaparral: Paradigms rnicrosites of ajuniper-grass ecosystem. Soil Sci. Soc. Am. J.reexamined. pp. 97-105. No. 34 Science Series, Natural 63: 1901-1905.History Museum of L.A. County, Los Angeles, CA. Leach,M.K. & Givnish, T.J.1999. Gradients in the composition,

Davis, E.A. & rase, C.P. 1977. Root system of shrub live oak: structure, and diversity of remnant oak savannas in southernimplications for water yield in Arizona chaparral. J. Soil Wisconsin. EcoL Monogr. 69: 353-374.Water Conserv. July-August 1977: 174-179. Leopold, A. 1924. Grass, brush, timber, and fire in southern

Davis, O.K. 1987. Palynological evidence for historic juniper Arizona. J. Forest. 22: 1-10.! invasion in central Arizona: A late-Quaternary perspective. Longstreth, D.J. & Patten, D.T. 1975. Conversion of chaparral toj In: Everett, R.L. (ed). Proc. Pinyon-Juniper Conference, pp. grass in central Arizona: Effects on selected ions in water-! 12-124. USDA For. Servo Gen.Tech. Rep.INT-215, Ft. shed runoff. Am. Midl. Nat. 93:25-34.: Collins, CO. Lowe, C.H. 1977. Arizona's natural environment. The Univer-,; DeAngelis, D.L. & Waterhouse, J.C. 1987. Equilibrium and sity of Arizona Press, Tucson, AZ.; nonequilibriumconcepts in ecological models. Ecol. Monogr. MacMahon, J.A. 1997. Ecological restoration. In: Meffe, G.K.

57: 1-21. & Carroll, C.R. (eds.) Principles of conservation biology,Dodd, J.L. 1994. Desertification and degradation in Sub-Saha- 2nd ed., pp. 479-511. Sinauer, Sunderland, MA.

ran Africa: the role of livestock. BioScience 44: 28-34. McCune, B. 1997. Influence of noisy environmental data on, ' .Ellis, B.A. & Kumrnerow, J. 1989. The importance ofN2 fixa- canonical correspondence analysis. Ecology 78: 2617-2623.! tion in Ceanothus seedlings in early postflre chaparral. In: McDougall, W.B. 1973. Seed plants of Northern Arizona. The

Keeley, S.C. (ed) The California chaparral: Paradigms Museum of Northern Arizona, Flagstaff, AZ.reexamined, pp. 115-117. No. 34 Science Series, Natural Miller, E.A. & Halpern, C.B. 1998. Effects of environment andHistory Museum of L.A. County, Los Angeles, CA. grazing disturbance on tree establishment in meadows of the

Emmerich,W.E.1999.Nutrientdynamicsofrangelandburnsin central Cascade Range, Oregon, USA. J. Veg. Sci. 9: 265-Southeastern Ariwna. J. Range Manage. 52: 606-614. 282.

Evans, R.A. 1988. Management of pinyon-juniper woodlands. Miller, R.F. 1999. Fire history and western juniper encroach-: USDA For. Servo Gen. Tech. Rep. 1NT-249,Ft. Collins, CO. ment in sagebrush steppe. J. Range Manage. 52: 550-559.! Harmon, M.E., Bratton, S.P. & White, P.S. 1983. Disturbance Milton, S.J., Dean, W.R.J., du P1essis, M.A. & Siegfried, W.R.

and vegetation response in relation to environmental gradi- 1994. A conceptual model of arid rangeland degradation: theents in the Great Smoky Mountains. Vegetatio 55: 129-139. escalating cost of declining productivity. BioScience 44: 70-

Hilbert, D.W. & Larigauderie, A. 1990. The concept of stand 76.senescence in chaparral and other Mediterranean type eco- Minchin, P.R. 1987. An evaluation of the relative robustness ofsystems. Acta DecoL 11: 181-190. techniques for ecological ordination. Vegetati071: 145-156.

Holling, S. 1973. Resilience and stability of ecological systems. Naveh, Z. 1967. Mediterranean ecosystems and vegetation typesAnnu. Rev. Ecol.Syst. 4: 1-23. in Califomia and Israel. &ology 48: 445-459.

Huebner, C.D., Randolph, J.C. & Parker, G.R. 1995. Environ- Neary, D.G., Overby, S.A., Gottfried, G.J. & Perry, H.M. 1996.

! 34 Huebner, C.D. & Vankat, J.L.

Nutrients in fIre-dominated ecoystems. In: Ffolliott, P.F., Turner, M. G. 1989. Landscape ecology: the effect of pattern onDeBano, L.F.,Baker Jr., M.B., Gottfried, G.J., Solis-Garza, process. Annu. Rev. Ecol. Syst.20: 171-197.G., Edminster, C.B., Neary, D.G., Allen, L.S. & Hamre, Valone, T.J. 1999. Fire and grazing in a shrub-invaded aridR.H. (eds.) Effects offire on Madrean Province ecosystems, grassland community: independent or interactive ecologicalpp. 207-117. USDA For. Servo Gen. Tech. Rep. RM-289, Ft. effects? J. Arid Environ. 42: 15-28.Collins, CO. Vankat, J.L. 1989. Water stress in chaparral shrubs in summer- .'

Noy-Meir, I., Gutman, M. & Kaplan, Y. 1989. Responses of rain versus summer-droughtc1irnates:WhithertheMediter-Mediterranean grassland plants to grazing and protection. J. ranean-type climate paradigm? In: Keeley, S.C., (ed) TheEcol. 77: 290-310. Calijornia chaparral: Paradigmsreexamined,pp. 117-124.

O'Connor, T.G.&Roux,P.W.1995. Vegetation changes (1949- No. 34 Science Series, Natural History Museum of L.A.71) in a semi-arid, grassy dwarf shrubland in the Karoo, County, Los Angeles, CA.South Africa: influence of rainfall variability and grazing by Vog1, R. 1981. Chaparral succession. In: Conrad, C.E. & Oechel,sheep. J. Appl. Ecol. 32: 612-626. W.C. (eds) Proceedings of the Symposium on Dynamics and

0k1and, R.H. 1996. Are ordination and constrained ordination Management of Mediterranean-Type Ecosystems, pp. 81-alternative or complementary strategies in general ecologi- 85. USDA For. Servo Gen. Tech. Rep. PSW-58, San Diego,cal studies? J. Veg. Sci. 7: 289-292. CA.

O'Neill, R.V., Krummel, J.R., Gardner, R.H., Sugihara, G., Walker, S. & Wilson, J.B. 2002. Tests for nonequilibrium,Jackson, B., DeAngelis, D.L., Milne, B. T., Turner, M.G., instability, and stabilizing processes in semiarid plant com-Zygmunt, B., Christensen, S.W., Dale, V.H. & Graham, munities. Ecology 83: 809-822.R.L. 1988. Indices of landscape pattern. Landscape Ecol. 1: Weaver,J.E.&Fitzpatrick, T.J.1934. The prairie. Ecol. Monogr.153-162. 4: 109-295.

, Peet, R.K. 1974. The measurement of species diversity. Annu. Weber, D.J., Bunderson, E.D., Davis, J.N, Nelson, D.L. &i Rev. Ecol. Syst. 5: 285-307. Hreha, A. 1999. Diseases and Environmental factors of thel Picke~, S.T.A. & White, P.S. (eds).1985. Thee~ologyofnatur~l pinyon-junipe~communities.In:Monsen,S.B.&Stev~nsR., disturbance and patch dynamics. AcademIc Press, San Di- (eds) Proceedings: Ecology and Management of Pmyon-: ego, CA. Juniper Communities within the Interior West, pp. 118-120..I ;~ Pickup, G. & Stafford Smith, ?M. 1993. ~ob~~ms, prospects .USDA For: Servo Gen. Tech. Rep. RMRS-P-9, O.gd~n, l!T.

.'" and procedures for assessmg the sustainabIlity of pastoral White, P.S., WIlds, S.P. & Stratton, D.A. 2001. The dIstrIbutIon\ land management in arid Australia. J. Biogeogr. 20: 471- of heath balds in the Great Smokey Mountains, North Caro-I 487. lina and Tennessee. J. Veg. Sci. 12: 453-466.

I Quideau, S.A. 1999. Biogeochemical cycling of calcium and Wienhold, B.J. & Kle~edson, J.O. 1992. Effect of prescribedi magnesium by Ceanothus and charnise. Soil Sci. Soc. Am. J. fIre on nitrogen and phosphorus in Arizona chaparral soil-.1c 63: 1880-1888. plant systems. Arid Soil Res. Rehabil. 6: 285-296.

;h;~ Sa~er, R.E. & ~agle, R.F. 1967. Facto~ affecting the distribu- Wiens, J..A. 1?85. Verte~ra!eresponses to env~onmental patchi-ti tIon of shrub live oak (Quercus turbmella Greene). Ecology ness m arId and semIarId ecosystems. In: Pickett, S.T.A. and

48: 35-41. White, P.S. (eds.) The ecology of natural disturbance andSheridan, T.E. 1995. Arizona: A history. The University of patch dynamics, pp.180-189. Academic Press, San Diego,

Arizona Press, Tucson, AZ. CA...Spang, E.F. 1987. Juniper from a Bureau of Land Management Woodmansee, R.G. 1979. Factors influencing input and output1. perspective. In: Everett, R.L. (ed.) Proceedings of the Pinyon- of nitrogen in grasslands. In: French, N. (ed) Perspectives in

fcc Juniper Conference, pp.489-493. USDAFor.Serv. Gen.Tech. grasslandecology,pp.117-129.Springer-Verlag,NewYork,,. .

CO NY1 Rep. INT-215, Ft. Collins, ..j Stephens, G. & Whitford, W.G. 1993. Responses of Bouteloua Woods, K.D. 2000. Long-term change and spatial pattern in a

eriopoda to irrigation and nitrogen fertilization in a late-successionalhernlock-northernhardwoodforestJ.EcoLChihuahuan desert grassland. J. Arid Environ. 24: 415-421. 88: 267-282.

Stock, W.D. &Lewis,0.A.M.1986. Soil nitrogen and the role of Wright, H.A. & Bailey, A.W. 1982. Fire ecology: United States; fire as a mineralization agent in a South African coastal and Southern Canada. Wiley, New York, NY.

fynbos ecosystem. J. EcoL 74: 317-328. Zimmennan, G.M., Goetz, H. & Mielke, Jr., P.W. 1985. Use ofSughara, G. & May, R.M. 1990. Applications of fractals in an improved statistical method for group comparison to fIre

ecology. Trends EcoL EvoL5: 79-86. effects of prairie fire. Ecology 66: 606-611.Sullivan, S. 1996. Towards a non-equilibrium ecology: perspec-

tives from an arid land. J. Biogeogr. 23: 1-5.Sutherland, J.P. 1990. Perturbations, resistance, and alternative Received 15 December 2000;

views of the existence of multiple stable points in nature. Revision received 28 July 2002;Am. Nat. 136: 270-275. Accepted 11 September 2002.

ter Braak, C.J.F. 1988. CANOCO -a FORTRAN program for Coordinating Editor: P.S. White.

canonical community ordination by partial de trended ca-nonical correspondence analysis, principal componentsanalysis and redundancy analysis (version 2.1). Center forBiometry, Wageningen, NL.

~