Printer Friendly

FACILITATION OF CONSPECIFIC SEEDLING RECRUITMENT AND SHIFTS IN TEMPERATE SAVANNA ECOTONES.

Abstract. In contrast to documented increases in woody plant dominance of savannas and grasslands of North America, oak (Quercus L.) savannas that form lower tree lines in the southwestern United States and northwestern Mexico have been stable over the last several centuries. We sought to identify potential biotic and abiotic constraints on seedling recruitment of Quercus emoryi within the context of potential shifts in lower tree line. We used field surveys to describe seedling distribution at and below lower tree line, and to determine the potential for acorn dispersal from lower tree line into adjacent grassland. Field and greenhouse experiments were used to test explanatory hypotheses generated by descriptive surveys.

Q. emoryi seedlings were located almost exclusively beneath mature, conspecific tree canopies within the woodland and savanna and were absent from adjacent semidesert grassland in 1993 and 1995. Seed bank surveys indicated that acorns were concentrated beneath tree canopies and were dispersed into adjacent grassland in low numbers. Although soil N, C, and P were about two times greater beneath trees than in adjacent grassland, experimental nutrient amendments to subcanopy and grassland soils indicated that soil nutrients did not limit Q. emoryi growth. Reciprocal transfers of subcanopy and grassland soil to subcanopy and grassland microsites indicated that microsite was more important than soil source for seedling growth. Overstory shade was important at all stages of seedling development investigated: the provision of artificial or natural shade increased rates of seedling emergence and subsequent survival as much as 19-fold and increased recruitment rates between 30and 60-fold.

We conclude that rates of Q. emoryi recruitment within grasslands below tree line are relatively low and are constrained by low rates of seed dispersal coupled with a low probability of seedling emergence. In contrast, large numbers of acorns are dispersed directly beneath Q. emoryi trees, where they have a higher probability of emergence than in adjacent grassland. Survival rates of emerged seedlings were low, regardless of landscape position. Thus, observed patterns of seedling distribution on the landscape resulted from interactions between seed dispersal and habitat-specific response of seedlings to environmental variation.

Results of this and complementary research suggest that the lower tree line in southern Arizona is stabilized by self-enhancing feedback mechanisms of overstory shade, seed dispersal, and seedling establishment, coupled with strong abiotic constraints beyond the current ecotone. These processes stabilize the woodland-grassland ecotone both spatially and temporally, consistent with Wilson and Agnew's one-sided positive feedback switch. Although this switch would not produce an indefinitely stable vegetation mosaic, upslope or downslope shifts in lower tree line are apparently resistant to decadal or even century-scale climate perturbation. The observed shift in tree line in the last millennium was less likely the result of slow, spatial progression of autogenic safe sites than the result of episodic and infrequent allogenic processes that simulated or negated the importance of conspecific, biogenic safe sites.

Key words: acorn; ecotone; facilitation; Quercus emoryi; safe site; seed dispersal; seedling establishment; shade, effects on seedling recruitment; tree line.

INTRODUCTION

... rather than concentrating on a search for the

ways in which organisms are perfectly suited to their

environments, we might more healthily concentrate

on the nature of the limitations that constrain where

they live. [Investigation of] what goes wrong with

plants when they are grown in communities outside

their normal range ... appear[s] to be the ideal way

to demonstrate the real extent and proximal cause

of the narrow specialization of most plant forms.

--Harper (1982)

Woody plant stature and abundance in savannas and grasslands have increased worldwide within recent history (Archer 1995). These changes in landscape physiognomy have important implications for desertification, livestock production, wildlife habitat conservation, nutrient cycling, and soil erosion (Grover and Musick 1990, Schlesinger et al. 1990, Young and Solbrig 1993). The proximate factors that influence rates, dynamics, and patterns of vegetation change are not well understood, although several mechanisms for changes in woody plant abundance and distribution have been proposed (e.g., fire suppression, herbivory by native or introduced herbivores, and directional climate change; Archer 1994, Polley et al. 1996, Weltzin et al. 1997).

On landscape scales, biotic and edaphic factors and disturbance are important determinants of vegetation pattern (Prentice 1986, Archer et al. 1995). At more local scales, facilitative interactions and biogenic safe sites constitute important controls on terrestrial plant populations (Steenbergh and Lowe 1969, 1977, Jordan and Nobel 1979, Neilson and Wullstein 1983, Franco and Nobel 1989). Amelioration of subcanopy micro-environments by overstory vegetation can be conducive to establishment of plants that are intolerant of surrounding environmental conditions, with ramifications for spatial and temporal patterns of recruitment and distribution of species within plant communities (Clements et al. 1926, Callaway 1992a, Wilson and Agnew 1992, Callaway et al. 1996, Wilson 1998).

Facilitation of conspecific seedling recruitment through provision of biogenic safe sites has been reported for trees and shrubs in many arid and semiarid savanna and woodland systems (Petranka and McPherson 1979, Neilson and Wullstein 1983, Muick and Bartolome 1987, Callaway and D'Antonio 1991, Bragg et al. 1993, Jackson and Van Auken 1997). Microenvironments beneath woody plants within these systems often differ markedly from adjacent grassland zones (see reviews by Vetaas [1992], Belsky and Canham [1994], McPherson [1997], Scholes and Archer [1997], Wilson [1998]). Woody plant canopies alter subcanopy precipitation distribution, solar radiation, evapotranspiration, and ambient and soil temperature regimes (e.g., Tiedemann and Klemmedson 1973, Joffre and Rambal 1988, Belsky et al. 1989, Callaway et al. 1991, Haworth and McPherson 1994, 1995), and concentrate soil nutrients relative to adjacent grassland zones (e.g., Garcia-Moya and McKell 1970, Tiedemann and Klemmedson 1973, 1977, Belsky et al. 1989, Schlesinger et al. 1990, McPherson et al. 1991). However, the relative importance of these different and potentially opposing canopy effects, and their interaction with patterns of seed dispersal, are seldom determined with respect to subcanopy seedling recruitment.

Changes in the distribution and abundance of woody plants within savannas and grasslands are ultimately dependent on recruitment of individuals into the population. In turn, spatial and temporal patterns of plant recruitment represent the integrated effect of biotic and abiotic environments on patterns and processes of dispersal, as well as germination of seeds and emergence, growth, survival, and establishment of seedlings and juveniles (Grubb 1977, Harper 1977, De Steven 1991a, b, Primack and Miao 1992, Bazzaz and Wayne 1994, Herrera et al. 1994, Schupp 1995, Schupp and Fuentes 1995). Thus, factors that alter seed dispersal and seedling establishment and survival are critical to woody plant population demographics (Grubb 1977, Harper 1977, McPherson 1997, Scholes and Archer 1997).

In this paper, we investigate potential constraints on woody plant establishment within oak (Quercus L.) savannas characteristic of the southwestern United States and northwestern Mexico (McClaran and McPherson, in press). Southwestern oak savanna forms an ecotone between temperate oak woodland and adjacent semidesert grassland (Brown 1982). Although this lower tree line shifted downslope into former grasslands within the last millennia, it has apparently remained stable over the last several centuries (Bahre 1991, McPherson et al. 1993, McClaran and McPherson 1995). This contrasts with other savanna systems of the world that have experienced relatively recent shifts in the distribution and abundance of woody plants and grasses (Schlesinger et al. 1990, Young and Solbrig 1993, Archer 1994, McPherson 1997).

Research suggests that Quercus seedling establishment within existing woodlands and savannas in southern Arizona is variable (Sanchini 1981, Borelli et al. 1994). Preliminary surveys suggested that Quercus seedling establishment occurs infrequently in grasslands below current lower tree line and in the grassland phase of low-elevation savannas (Weltzin and McPherson 1995). Variations in Quercus seedling recruitment have been attributed to interannual and seasonal variation in precipitation regimes (Pase 1969, Neilson and Wullstein 1983, McPherson 1992, Germaine and McPherson 1998, Weltzin and McPherson, in press), herbaceous interference (McPherson 1993, Germaine and McPherson 1999), livestock grazing (Bahre 1977), and acorn predation (Hubbard and McPherson, in press).

The goal of this project was to identify potential biotic and abiotic constraints on Quercus seedling recruitment and subsequent distribution within the context of shifts in lower tree line. Because seedling recruitment is constrained by different density-dependent and density-independent processes at successive demographic stages, patterns of seed dispersal, germination, and early seedling establishment may be only partially coupled (Houle 1995, Schupp 1995, Schupp and Fuentes 1995). Therefore, we focused our investigation on multiple demographic stages of Quercus seedling recruitment (cf. De Steven 1991a, b, Herrera et al. 1994, Hill et al. 1995, Kollmann and Schill 1996).

Our objectives were to (1) describe the distribution of Quercus seedlings at and below lower tree line on sites with and without long-term histories of livestock grazing, (2) determine the potential for acorn dispersal from lower tree line into adjacent grassland habitats, and (3) determine effects of soil properties (particle size distribution and soil nutrient status) and micro-climate (created by overstory shade) on seedling recruitment in different habitats (subcanopy vs. grassland zones). Because comparisons of seedling recruitment between subcanopy and grassland zones are potentially confounded by tree and site effects (e.g., Bartolome et al. 1994), we used manipulative experiments to test explanatory hypotheses generated by descriptive surveys (sensu Gurevitch and Collins 1994).

METHODS

Southwestern oak savanna and lower tree line

Research was conducted during 1993-1997 at the lower (and drier) margin of temperate, evergreen oak woodland at the base of the Huachuca Mountains in southeastern Arizona. The ecotone between oak woodland and adjacent semidesert grassland is characterized by Quercus emoryi Tort. (Emory oak)-dominated woodlands and savannas bordered by semidesert grassland dominated by perennial bunchgrasses (Brown 1982, McClaran and McPherson, in press; see Plate 1).

[Plate 1 ILLUSTRATION OMITTED]

Climate at this lower tree line is semiarid, with an average annual temperature of ~20 [degrees] C. Average annual precipitation ranges from 350 to 600 mm and is bimodally distributed, with peaks during the summer "monsoon" (July-September; 50%) and during winter (December-March; 30%). Additional information on climate, soils, and vegetation of lower tree line is provided by McClaran and McPherson (in press; see Plate 1).

Q. emoryi effects on subcanopy microenvironments

Effects of mature Q. emoryi on subcanopy microenvironments are well documented. Q. emoryi tree canopies suppress herbaceous biomass production by as much as 40%, although herbaceous species composition is not affected (Haworth and McPherson 1994). Seasonal temperature extremes are lower beneath trees than in interstitial zones: subcanopy temperatures are relatively cool in summer and warm in winter (Nyandiga and McPherson 1992, Haworth and McPherson 1995). Individual canopies can intercept as much as 70% of incident precipitation, whereas stem flow contributes water to the soil near the tree bole; soil moisture contents (at 10 cm) did not differ between subcanopy and interstitial zones (Haworth and McPherson 1995). N and organic C contents in the top 10 cm of soil are greater beneath mature trees than in interstitial zones (McPherson et al. 1993).

In 1993, six sites at lower tree line were selected subjectively for their physiognomic, elevational, and topoedaphic similarities (Table 1). Three sites (GW, GE, TC) are on Fort Huachuca Military Reservation (FHMR) and represent communities with a 40-yr history of protection from livestock grazing and fire (J. Miller, FHMR, personal communication 1993). The other three sites (MC, HC, AC) are on lands adjacent to FHMR that are administered by the United States Forest Service (USFS), and represent communities with a [is greater than] 100-yr history of livestock grazing (Bahre 1991, D. Bennett, USFS, personal communication 1993). The three grazed sites had not burned for [is greater than] 30 yr prior to the initiation of this study (W. Wilcox, U.S. Forest Service, personal communication 1997).
TABLE 1. Characteristics of the six study sites at lower tree line
of the Huachuca Mountains, southeastern Arizona, USA.

                                          Site

                           Garden Canyon         Garden Canyon
   Characteristic            West (GW)             East (GE)

Slope (%)                        6                     7
Aspect                          ESE                    N
Elevation (m)                  1560                  1550
Latitude                31 [degrees] 30' N    31 [degrees] 29' N
Longitude               110 [degrees] 20' W   110 [degrees] 19' W
Q. emoryi tree canopy
  cover (%)                     23                    29
Total tree canopy
  cover (%)                     23                    32
Herbaceous
  cover([dagger])
  Canopy zone              33 [+ or -] 3         39 [+ or -] 2
Herbaceous
  cover([dagger])
  Interstitial zone        48 [+ or -] 2         36 [+ or -] 3
Herbaceous
  cover([dagger])
  Grassland zone           57 [+ or -] 4         42 [+ or -] 4
Soil subgroup           Pachic                Aridic
  ([double dagger])       Haplustolls           Haplustalfs
Grazing
  history([sections])           LTP                   LTP

                                          Site

                           Tinker Canyon         Manila Canyon
   Characteristic              (TC)                  (MC)

Slope (%)                        6                     6
Aspect                          ENE                   NNW
Elevation (m)                  1550                  1580
Latitude                31 [degrees] 29' N    31 [degrees] 33' N
Longitude               110 [degrees] 18' W   110 [degrees] 26' W
Q. emoryi tree canopy
  cover (%)                     29                    25
Total tree canopy
  cover (%)                     33                    34
Herbaceous
  cover([dagger])
  Canopy zone              32 [+ or -] 2         29 [+ or -] 2
Herbaceous
  cover([dagger])
  Interstitial zone        27 [+ or -] 3         39 [+ or -] 2
Herbaceous
  cover([dagger])
  Grassland zone           43 [+ or -] 2         62 [+ or -] 3
Soil subgroup           Aridic                Ustollic
  ([double dagger])       Haplustalfs           Haplargids
Grazing
  history([sections])           LTP                   LTP

                                          Site

                           Hunter Canyon          Ash Canyon
   Characteristic              (HC)                  (AC)

Slope (%)                        4                     4
Aspect                          NNE                   NE
Elevation (m)                  1500                  1520
Latitude                31 [degrees] 25' N    31 [degrees] 24' N
Longitude               110 [degrees] 15' W   110 [degrees] 14' W
Q. emoryi tree canopy
  cover (%)                     22                    23
Total tree canopy
  cover (%)                     24                    26
Herbaceous
  cover([dagger])
  Canopy zone              28 [+ or -] 3         38 [+ or -] 4
Herbaceous
  cover([dagger])
  Interstitial zone        36 [+ or -] 2         39 [+ or -] 3
Herbaceous
  cover([dagger])
  Grassland zone           64 [+ or -] 2         47 [+ or -] 5
Soil subgroup           Ustollic              Ustollic
  ([double dagger])       Haplargids            Haplargids
Grazing
  history([sections])           LTP                   LTP

([dagger]) Percent cover [+ or -] 1 SE; pooled data for 1993 and
1995 (see text for statistical details).

([double dagger]) From Richardson et al. (1979) and Soil
Conservation Service (1994).

([sections]) LTP, long-term protected; LTG, long-term grazed.


Overstory vegetation at both grazed and ungrazed sites was dominated by Quercus emoryi, with scattered Juniperus deppeana Steud. (alligator juniper), Quercus arizonica Sarg. (Arizona white oak), and Quercus oblongifolia Tort. (Mexican blue oak) (Table 1). Total tree canopy cover ranged 23-34% as determined with the line-intercept method (Canfield 1941).

Herbaceous vegetation within the savannas and grasslands on the ungrazed sites was dominated by late-successional, [C.sub.4] perennial bunchgrasses, including Trachypogon montufari (H.B.K.) Nees. (crinkleawn), Bouteloua curtipendula (Michx.) Tort. (sideoats grama), Andropogon cirratus Hack. (Texas bluestem), and Eragrostis intermedia Hitchc. (plains lovegrass). Dominant [C.sub.3] forbs in the grasslands included Heterotheca subaxillaris (Lam.) Britton and Rusby (camphorweed), and Evolvulus arizonicus Gray. [C.sub.3] graminoids, particularly Sitanion hystrix J. G. Smith (bottlebrush squirreltail) and Cyperus (Tourn.) L. spp., were located beneath some tree canopies within the savanna.

Herbaceous vegetation within the savannas and grasslands on grazed sites was dominated by Bouteloua hirsuta Lag. (hairy grama), Eragrostis lehmanniana Nees. (Lehmann lovegrass), and B. curtipendula. Within the grazed grasslands, canopy cover of Prosopis velutina (Wooten) Sarg. (velvet mesquite) was 19% at MC and was [is less than] 1% at HC and AC, as determined by the line-intercept technique.

Seedling surveys

We conducted intensive surveys to determine patterns of Quercus spp. seedling recruitment at and below lower tree line. Our objectives were (1) to determine Quercus seedling densities in three landscape zones (beneath mature Q. emoryi tree canopies, in interstitial zones between Q. emoryi, and in adjacent, treeless semidesert grassland slightly below current lower tree line) and (2) to compare patterns of seedling distribution between landscapes with different histories of livestock grazing.

In 1993, at each of the six study sites (Table 1), we selected subjectively one 3-4-ha macroplot within the oak savanna, and another macroplot of similar size within the grassland located topographically below but adjacent to the savanna. Grassland macroplots were situated within 50-150 m (horizontal distance) from lower tree line, and within 5-15 m (vertical distance) below lower tree line. Within each savanna macroplot, we selected randomly 20 tree plots for sampling. Tree plots consisted of one to three mature Q. emoryi trees with overlapping canopies that were isolated from neighboring tree canopies by a distance of [is greater than] 1 m. Selected trees had canopies [is greater than] 4 m tall and [is greater than] 3 m in diameter. For each tree plot, we recorded the number of individual trees, height of the tallest tree, stem diameters at 20 cm above ground level for calculation of basal area, and mean canopy diameter (based on measurements of canopy diameters in orthogonal directions) for calculation of canopy area.

At each site, mean canopy area of the 20 tree plots dictated the mean area sampled at 20 plots located at random within both the interstitial zone between mature trees and in the adjacent grassland zone. Each plot within each landscape zone (subcanopy, interstitial, and grassland; n = 20) was surveyed intensively for Quercus seedlings between 18 September and 2 October 1993. We repeated the survey between 16 September and 24 October 1995 using newly selected plots within the macroplots.

In both sample years, individual seedlings ([is less than] 1 m tall) were identified to species, and placed into one of three approximate age classes ([is less than] 1, 1-2, and [is greater than] 2 yr) based on presence or absence of an acorn, perrenating tissue, and size. Soil around seedlings was excavated when necessary to determine whether seedlings originated from root or crown sprouts. When seedlings were located beneath Q. emoryi canopies, we recorded their ordinal aspect relative to the approximate geometric mean of the tree bole(s). For subsequent analyses, aspects between northwest and east were reclassified as a northern exposure, and the balance as a southern exposure. Because Q. emoryi comprised [is greater than] 99% of seedlings, all subsequent analyses and experiments were restricted to this species. Foliar herbaceous cover in each landscape zone at each of the six sites in each year was determined using the point-step method (Evans and Love 1957).

Seed bank surveys

We conducted acorn seed bank surveys to determine the potential for acorn dispersal across lower tree line into adjacent grasslands. Our objective was to determine acorn densities in litter and soil seed banks beneath mature Q. emoryi trees and at increasing distances into the grassland.

Seed bank surveys were conducted at GE and TC (Table 1) in May of 1994 and 1995 after all acorns from the previous year had detached from trees but prior to development of that year's acorn crop. At each site in each year, we selected randomly six mature Q. emoryi trees (mean height [+ or -] 1 SE = 9 [+ or -] 1 m; canopy diameter = 10 [+ or -] 1 m) as independent samples of trees along the lower edge of the savanna ecotone. Selected trees were [is greater than] 4 m from other Q. emoryi individuals above or to the side of the selected tree.

At each tree, we established a transect from tree bole to adjacent grassland. Along each transect and perpendicular to its axis, we positioned seven 0.5 x 1.0 m plots, such that the first plot (plot 1) was located beneath the canopy, its lower edge contiguous with a line parallel to lower tree line and tangent to the canopy edge (which was the designated zero point). Thus, plot 1 was centered at -0.25 m on the transect relative to the canopy edge. Subsequently, plots 2-7 were centered at +0.25, +0.75, +1.25, +1.75, +2.75, and +3.75 m below the canopy edge. Hubbard and McPherson (in press) conducted similar Q. emoryi acorn seed bank surveys, at a nearby site in 1993 and 1995, at 10-m intervals to 50 m distance into the grassland from the canopy edge.

Within each plot, we collected surface organic material (i.e., litter) from two randomly located 0.0625 [m.sup.2] subplots. All remaining litter was then cleared from the plot, and soil within the plot excavated from two depths below the mineral soil surface: 0-5 cm and 5-15 cm. Excavated material was dry-sieved through tiered 13 mm and 6 mm screens constructed of hardware cloth, and the 6-13 mm fraction was collected. Whole acorns and acorn components (buds, cups, and fragments) were separated manually from litter and sieved-fraction soil samples in the laboratory. Whole acorns and acorn components were tallied and converted to density (number/[m.sup.2]). The sum of whole acorns and acorn components (hereafter, referred to as "total") was used as a simple index of acorn presence and potential dispersal. Means of litter seed bank subplots were used in analyses (n = 6). All whole acorns were tested for viability with 2,3,5-triphenyl tetrazolium chloride (Nyandiga and McPherson 1992).

Soil morphology and chemistry

We analyzed soil from beneath Q. emoryi trees and from adjacent grassland below tree line at GW (Table 1) to compare soil physical and chemical characteristics between these two locations. Soil samples were obtained from soil profiles beneath five randomly selected, isolated, mature Q. emoryi tree canopies, north of the tree bole and midway between the bole and the canopy edge. Within the grassland, five sample points were located randomly within a representative 1-ha area. Plots for this and all subsequent field experiments were located within the macroplots at GW established for the seedling survey.

At each point, a sample was collected from each of five depths in the soil profile: 0-5, 20-40, 40-60, 60-80, and 80-100 cm. Samples were transported to the laboratory in plastic bags and were air-dried to constant weight. Samples were then sieved to separate coarse rock fragments ([is greater than or equal to] 2 mm diameter) from finer particles ([is less than] 2 mm diameter).

The proportion of coarse fragments in each sample was determined by weighing, and was converted to a percentage by volume basis (Soil Survey Staff 1994a). Percentage sand and clay of air-dried samples (on a 100-g oven-dried basis) were determined by particle size analysis (Day 1965). We calculated a weighted particle size distribution for the soil profile at each sample point based on the control section (25-100 cm depth; Soil Survey Staff 1994b).

Soil pH for all soil layers was determined with a Mettler DL21 Titrator (Mettler Instruments, Hightstown, New Jersey, USA) using a solution of 10 ml distilled water mixed with a 10-g (air-dried fine fraction) subsample (McLean 1982). Analyses of element concentration were based on the soil samples collected from the 0-5 cm soil layer after samples were pulverized to a fine powder and homogenized thoroughly with a ball mill. Soil organic C was determined by dry combustion (Nelson and Sommers 1982) in a LECO high-frequency induction furnace (LECO, Saint Joseph, Michigan, USA) corrected for inorganic C using a gasometric method (Dreimanis 1962). Available N was determined using 2 mol/L KCl extraction followed by MgO and Devardas alloy steam distillation (Keeney and Nelson 1982). Available P was determined using 0.5 mol/L Na[HCO.sub.3] extraction followed by ascorbic acid color development (Olsen and Sommers 1982).

Soil/microsite effects: seedling emergence

To determine the relative importance of soil source and microsite on Q. emoryi seedling germination and emergence, we conducted emergence trials in the field and in a greenhouse in 1994 and 1995. Acorns used in the study were collected in July of each year from [is greater than or equal to] 20 trees at GW. Collected acorns were visually examined for insects and pathological infection and were sorted by flotation: acorns that floated or had visible insect damage were discarded (sensu Nyandiga and McPherson 1992).

Field trials were conducted at GW, where we established 0.125-[m.sup.2] plots beneath five randomly selected mature Q. emoryi trees, as well as in five randomly selected locations in the grassland below tree line. Plots beneath canopies were located on the north side of the tree bole, midway between the bole and the canopy edge. At each of the 10 plots, we excavated a pit 10 cm deep, and we used the loose soil to fill 20 1-L pots (n = 200 pots). Ten pots from each plot were retained in the field (n = 100), and ten were transported to a greenhouse in Tucson, Arizona, USA (n = 100).

In the field, pots were redistributed into a randomized complete block design; each plot was assigned one pot from each of the 10 plots (n = 10). Pots were placed in random order into the excavated pit, which was then backfilled so that the soil surface within the pot was level with the surrounding soil. Five acorns were planted 1 cm below the soil surface in each pot. Each plot was covered with 5 mm wire mesh to exclude vertebrates. Pots were watered weekly for three weeks and monitored for seedling shoot emergence weekly for 12 wk.

In the greenhouse, pots with subcanopy and grassland soil were arranged in a completely randomized design (n = 50), and five acorns were planted into each pot. Pots were watered and monitored for emergence thrice weekly for eight weeks. Pots were relocated at weekly intervals to minimize edge and gradient effects.

Soil/microsite effects: seedling development

To determine potential interactive effects of soil and microsite on seedling development and survival, we conducted a reciprocal soil transfer in the field. Soils were collected from grassland and subcanopy environments and transferred to the opposite location. Research was conducted within the savanna and adjacent grassland at GW (Table 1) during 1994-1996.

In June 1994, we established plots beneath each of five, randomly selected, isolated clusters of one to three individual, mature Q. emoryi trees (mean cluster height [+ or -] 1 SE = 10.5 [+ or -] 0.7 m; diameter = 9.9 [+ or -] 0.7 m). Plots were positioned north of the northernmost bole, midway between the bole and the canopy edge. Within the grassland, we established five plots at random locations within a representative 1-ha area.

At each plot, we excavated soil in 20-cm increments to 1-m depth. Excavated soil from each plot was used to backfill eight soil columns constructed of 15 cm diameter polyvinyl chloride pipe cut to 1-m length with 6-mm mesh hardware cloth wired to the bottom. Thus, a total of 80 columns were filled with field soil (2 vegetation types x 5 plots/type x 8 columns/plot). Half of the columns were retained in the field and half were transported to the greenhouse for a nutrient amendment experiment (see Methods: Role of soil nutrients).

In the field, columns from each plot were assigned randomly to both grassland and subcanopy plots, with the stipulation that each plot contain columns from two canopy and two grassland plots (n = 10). Columns were placed upright, 10 cm apart, in random order within each pit, which was backfilled such that columns protruded ~2.5 cm above the surrounding soil surface. Q. emoryi seedlings established in a greenhouse were transplanted into each column at the four to six leaf stage on 18 August 1994, and were watered for two consecutive days to facilitate establishment. Seedlings were protected from vertebrate herbivores with 6-mm mesh hardware cloth exclosures. Insecticide (Carbaryl 4L) was applied to all seedlings biweekly during the growing season to minimize invertebrate herbivory.

Seedling survival was monitored monthly until 9 August 1995, whereupon we recorded leaf number, shoot diameter, and height. Columns were then excavated and harvested destructively to determine shoot biomass and root biomass within 20-cm increments. All biomass samples were oven-dried at 60 [degrees] C for 72 h before weighing. We repeated this experiment in 1995-1996, using new plots at the same site; seedlings were transplanted into columns on 14 August 1995 and were harvested destructively on 17 August 1996.

Role of soil nutrients

To determine potential constraints of soil nutrient status on Q. emoryi development, we conducted a nutrient amendment experiment in a greenhouse. Columns of soil collected from subcanopy and grassland plots at GW in June 1994 were arranged in the greenhouse in a randomized complete block design. Columns were supported in two blocks within a framework of commercially available lumber that held pots 10 cm apart. Within each block, half of the pots from each soil source received a full nutrient amendment, and the other half served as unfertilized controls (n = 5). Nutrient amendments in the form of fertilizer dissolved in deionized water were applied monthly at rates equivalent to 12, 10.6, and 10 g [multiplied by] [m.sup.-2] [multiplied by] [yr.sup.-1] N, P, and K, respectively. Micronutrients included B, Cu, Fe, Mn, Mo, Zn, and Cl.

Q. emoryi seedlings established in the greenhouse were transplanted into each column at the four to six leaf stage on 26 August 1994, and were watered with 1.5 L water to facilitate establishment. Thereafter, columns received 0.5 L water/week. Leaf number and shoot height of seedlings were recorded monthly for 10 mo, whereupon columns were harvested destructively on 29 June 1995 for determination of oven-dry shoot biomass, and root biomass within 20-cm increments.

Role of overstory shade

Microenvironmental amelioration by Q. emoryi tree canopies includes numerous driving variables that may interact to alter conspecific seedling establishment. We conducted an experiment designed to separate the effects of overstory shade from subcanopy soil properties on Q. emoryi seedling recruitment at lower tree line during 1994-1997 at GW (Table 1). Within the savanna, we randomly selected 18 mature, individual Q. emoryi trees that were [is greater than or equal to] 6 m from other trees. One of three treatments was assigned randomly to each tree: tree left intact, tree cut at ground level, and tree cut at ground level with artificial shade erected. Within each of the interstitial and grassland zones, we randomly selected 12 2 x 2 m plots within homogeneous stands of herbaceous vegetation. Two treatments (artificial shade or unshaded controls) were assigned at random to six plots within each landscape zone. Therefore, this experimental design encompassed seven treatment combinations (n = 6).

Artificial shade structures were constructed of galvanized steel tubing joined to form a 2 x 2 m square frame, bisected by additional tubing for support. Commercially available shade cloth (rated at 95% light reduction) was attached to the frame. This shade cloth was selected because it reduces transmission of total radiation and photosynthetically active radiation (PAR) by 91%, an amount similar to light reduction by mature Q. emoryi (i.e., 90% on cloudless days in summer; Weltzin and McPherson unpublished data). The frame and shade cloth were permanently wired to four fence posts, such that they were parallel to, and 1 m above, ground level. Poultry netting (2.5-cm mesh) was wired to fence posts and rebar stakes around each plot to form a 60-cm-tall vertebrate exclosure.

On 12 July 1994 and 17 July 1995, just prior to the onset of the summer "monsoon," we planted 49 acorns at 10-cm spacing into each plot. In each year, seedling emergence and survival were monitored at two-week intervals until the end of October. Thereafter, seedling survival for each cohort was monitored monthly until December 1996, whereupon seedlings were monitored every 3-5 mo until experiment termination on 8 October 1997. On each monitoring date, dead seedlings were assigned a probable cause of mortality (e.g., desiccation or defoliation).

At the end of each growing season, standing herbaceous biomass within each plot was determined by clipping three randomly located 0.0625-[m.sup.2] subplots within each plot. Means of subplots within each plot were used in analyses. Soil moisture content at 10 and 50 cm in each plot was determined gravimetrically four times each year through November 1996. At experiment termination, we recorded the height of live seedlings in each cohort.

Statistical analysis

All data were tested for normality with the Shapiro--Wilk W statistic (Shapiro and Wilk 1965). Data not normally distributed (P [is less than] 0.05) were transformed prior to analysis. When transformations failed, we used nonparametric analyses as indicated. First- and second-order interactions were included in all analysis of variance (ANOVA) models (SAS Institute 1989). We used Fisher's protected least significant difference (LSD; Fisher 1960) a posteriori mean separation tests to compare levels within factors for all significant (P [is less than] 0.05) main effects and first-order interactions.

We used a fixed-effects Kruskal--Wallis ANOVA model (Kruskal and Wallis 1952) to evaluate crossed (grazing history, landscape zone, and seedling age class) and nested (site) effects on seedling density for 1993 and 1995. For canopy plots only, we used a fixed effects Kruskal--Wallis ANOVA model to evaluate crossed (livestock grazing history, seedling age class, and exposure) and nested (site) effects on seedling density for 1993 and 1995. We used Pearson product moment correlation coefficients to assess correlations between subcanopy seedling density and tree plot characteristics (number of trees, height of tallest tree, basal area, and canopy area). We used paired t tests (Zar 1996) to determine differences in herbaceous cover between 1993 and 1995 for each zone at each of the six sites.

Whole and total acorn density data were analyzed by year (1994 and 1995) and seed bank depth (litter layer, 0-5 cm, and 5-15 cm) for main and interactive effects of site and distance from canopy edge. Site was treated as a block, and distance from the canopy edge was treated as a repeated measure within a multivariate analysis of variance framework (MANOVAR; Pillai's Trace in SAS procedure GLM; SAS Institute 1989) to account for potential autocorrelation between adjacent plots (von Ende 1993). We used single degree of freedom contrasts ("profile" transformation in SAS; SAS Institute 1989) with Bonferroni adjustments to compare means at successive distances from the canopy edge (von Ende 1993).

Soil physical (percent rock fragments, sand, and clay) and chemical (C, N, and P) data were analyzed together in a MANOVA model (Scheiner 1993) for effect of landscape zone (subcanopy or grassland). We used standardized coefficients of the canonical variables to evaluate the relative importance of response variables to overall differences between soil at each landscape zone. pH data were analyzed with t tests to compare subcanopy and grassland soils at each depth within the soil profile.

We used a fixed-effects Kruskal--Wallis test to evaluate main and interactive effects of microsite (subcanopy or grassland) and soil source (subcanopy or grassland) on seedling emergence in the field. A Mann--Whitney test (Zar 1996) was used to compare differences in emergence between grassland and subcanopy soils in the greenhouse. Seedling emergence data were analyzed separately for each year.

We used Fisher's exact test (Fisher 1958) to evaluate effects of soil source (subcanopy or grassland) and microsite (subcanopy or grassland) on seedling survival in the field for each year. Data for leaf number, shoot height, shoot, root, and total biomass, and root:shoot biomass ratios of surviving plants at experiment termination in each year were analyzed with ANOVA models for main and interactive fixed effects of soil source and microsite. We used MANOVAR models to analyze root biomass in successive 20-cm soil increments.

Data for leaf number, shoot height, shoot, root, and total biomass, and root:shoot biomass ratios of surviving plants at experiment termination in the greenhouse were analyzed for main and interactive effects of block, soil source (subcanopy or grassland) and nutrient amendment (amended or control). Soil source and nutrient amendment were treated as fixed effects, and block was treated as a random effect. We used MANOVAR models to analyze repeated measures of shoot height and leaf number, and root biomass in successive 20-cm soil increments.

We used a fixed-effects ANOVA model and multiple contrasts (Zar 1996) to test specific hypotheses regarding effects of shade treatment and landscape zone on seedling emergence (no. emerged seedlings/no. acorns planted). Proportional hazards regression analysis (SAS procedure PHREG; Allison 1995, SAS Institute 1996) was used to test for differences between survivorship curves of emerged seedlings for each cohort. Within cohort, subsets of survivorship curves were compared to test for differences between specific treatment combinations (equivalent to the multiple contrasts). To determine whether cohorts responded similarly to the different treatments, we compared survivorship curves within treatment that encompassed the duration of the experiment for cohort 2 (i.e., 783 d), and the first 783 d of cohort 1.

Seedling recruitment was defined as the product of seedling emergence and seedling survival at experiment termination as determined by proportional hazards regression analysis. We used Z tests of proportions (Zar 1996) to compare recruitment between treatment combinations equivalent to multiple contrasts. We used fixed-effects ANOVA models and multiple contrasts to evaluate effects of treatment on annual herbaceous standing biomass, and effects of treatment on soil moisture content at each depth. Because some treatments had only one or no surviving seedlings, heights of live seedlings at experiment termination are presented as means [+ or -] 1 SE.

RESULTS

Seedling surveys

Significant three-way interactions between grazing history, landscape zone, and seedling age class in 1993 (P [is less than] 0.0001) and 1995 (P = 0.046) reflected differential response of seedling densities in each age class to grazing history and landscape zone, so we reanalyzed the data by seedling age class.

In 1993, [is less than] 1-yr-old seedlings were located almost exclusively beneath tree canopies, and seedling densities beneath canopies were 16 x greater on ungrazed than on grazed sites (landscape zone x grazing history interaction, P = 0.02; Fig. 1). In 1995, [is less than] l-yr-old seedlings were restricted to tree understories on ungrazed sites (65 plants/ha) and were absent (0 plants/ha) from grassland and interstitial zones on ungrazed sites and from all landscape zones on grazed sites (landscape zone x grazing history, P [is less than] 0.0001). In 1993, 1-2-yr-old seedling densities beneath canopies (224 plants/ha) exceeded (P [is less than] 0.001) those in both interstitial (5 plants/ha) and grassland zones (0 plants/ha), which did not differ (P = 0.85). In both 1993 and 1995, density of [is greater than] 2-yr-old seedlings was greater (P [is less than] 0.007) beneath tree canopies (725 and 752 plants/ha, respectively) than in both interstitial (130 and 64 plants/ha, respectively) and grassland zones (0 plants/ha in both years), which did not differ (P [is greater than] 0.06). Densities of 1-2- and [is greater than] 2-yr-old seedlings did not differ between sites with different grazing histories in either year (P [is greater than] 0.36).

[Figure 1 ILLUSTRATION OMITTED]

In 1993, seedling densities beneath mature Q. emoryi canopies were greater at north (155 plants/ha) than south (56 plants/ha) exposures (P = 0.002). Densities of the youngest seedlings were greatest on ungrazed sites, whereas densities of older seedlings did not differ between grazed and ungrazed sites (grazing history x seedling age class, P [is less than] 0.0001; Fig. 2). In 1995, younger seedlings were relatively uncommon at both exposures, but the oldest seedlings were relatively numerous, especially at northern exposures (age class x exposure, P = 0.04; Fig. 3).

[Figures 2-3 ILLUSTRATION OMITTED]

In 1993 and 1995, subcanopy seedling density was correlated with overstory canopy area (r [is greater than] 0.37, P [is less than] 0.003) and basal area (r [is greater than] 0.38, P [is less than] 0.003) on ungrazed sites, but not on grazed sites (r [is less than] 0.14, P [is greater than] 0.27). Seedling density was correlated with tree height only for ungrazed sites in 1995 (r = 0.33, P [is less than] 0.01). Seedling density was correlated with number of trees per cluster in both years for grazed sites only (r [is greater than] 0.27, P [is less than] 0.04).

Herbaceous cover did not differ between 1993 and 1995 for subcanopy zones at any site (P [is greater than] 0.06), or for interstitial and grassland zones at TC, HC, GE, and MC sites (P [is greater than] 0.13) (Table 1). At AC and GW, herbaceous cover was greater in 1993 than in 1995 for interstitial (P [is less than] 0.04) and grassland (P [is less than] 0.0006) zones (data not shown).

Seed bank surveys

In both 1993 and 1995, whole and total (whole + component) acorn densities in the litter layer were greatest beneath tree canopies and decreased with increasing distance from the canopy edge (P [is less than] 0.01; Table 2). Whole acorn densities in the litter were [is less than or equal] 1 acorn/[m.sup.2] at distances [is less than or equal to] 2.75 m into the grassland, but acorn components were present at distances [is greater than or equal to] 2.75 m.
TABLE 2. Whole and total (whole plus components) mean acorn
densities (no. acorns/[m.sup.2]) in the surface litter layer and
0-5 cm and 5-15 cm below the soil surface, by distance from the
canopy edge at lower tree line in 1994 and 1995 (n = 12).

                               Distance from canopy edge (m)

Year   Seed bank   Component       -0.25           0.25

1994   Litter        Whole      129              71
                     Total     1180             623
       0-5 cm        Whole       34              12
                     Total      260([dagger])    91([dagger])
       5-15 cm       Whole        6               3
                     Total       53([dagger])    18
1995   Litter        Whole      201              86
                     Total      908             435([dagger])
       0-5 cm        Whole       20               8
                     Total      105([dagger])    36([dagger])
       5-15 cm       Whole        2               0
                     Total       13               4

                               Distance from canopy
                                     edge (m)

Year   Seed bank   Component      0.75         1.25

1994   Litter        Whole      18                3
                     Total     247               63
       0-5 cm        Whole       3                1
                     Total      23                8
       5-15 cm       Whole       1                0
                     Total       7                1
1995   Litter        Whole      30([dagger])     13
                     Total     219              111
       0-5 cm        Whole       2                1
                     Total      12               10
       5-15 cm       Whole       1                0
                     Total       2                0

                               Distance from canopy
                                     edge (m)

Year   Seed bank   Component    1.75   2.75   3.75

1994   Litter        Whole         2      1      0
                     Total        39     37      4
       0-5 cm        Whole         1      0      1
                     Total         3      1      2
       5-15 cm       Whole         0      0      0
                     Total         1      0      0
1995   Litter        Whole         1      1      0
                     Total        25     13     11
       0-5 cm        Whole         0      0      1
                     Total         3      3      6
       5-15 cm       Whole         0      0      0
                     Total         0      2      0

Note: Mean densities represent data pooled across blocks (i.e.,
sites TC and GE).

([dagger]) Within rows, designated means were different from the
successive mean (P < 0.05/6 = 0.008).


Whole acorn densities in both the 0-5 and 5-15 cm soil seed banks were unaffected by distance from the tree canopy in either year (P [is greater than] 0.13). In contrast, total acorn densities were greatest under the tree canopy and decreased with increasing distance into the grassland in the 0-5-cm soil seed bank in both years (P [is less than] 0.007) and in the 5-15-cm soil seed bank in 1993 (P = 0.02). No whole acorns were viable.

Soil morphology and chemistry

Soils beneath Q. emoryi and in adjacent grassland differed in their morphological and chemical characteristics (P = 0.01; Table 3). Differences in the volume of rock fragments contributed most to the overall difference between the two soils. Soils beneath tree canopies had almost twice the volume of rock fragments as did soils in the grassland (Table 3). In contrast, sand and clay contents differed little between subcanopy and grassland soils.
TABLE 3. Particle size distribution (means [+ or -] 1 SE; control
section, 25-100 cm) and concentration of organic C (g/kg
soil) and available N and P (mg/kg soil) in soil beneath Q.
emoryi trees and in adjacent grassland at GW (n = 5).

                             Landscape zone

Soil parameter         Subcanopy       Grassland

Rock fragments (%)   26 [+ or -] 2   15 [+ or -] 0
Sand (%)             64 [+ or -] 1   67 [+ or -] 1
Clay (%)             21 [+ or -] 1   20 [+ or -] 0
C (g/kg)             23 [+ or -] 4   12 [+ or -] 1
N (mg/kg)            14 [+ or -] 3    6 [+ or -] 1
P (mg/kg)            19 [+ or -] 2   12 [+ or -] 2


Soil nitrogen and carbon concentrations were about two times greater beneath trees than in adjacent grassland (Table 3). P concentrations also tended to be higher beneath trees. Soil pH ranged 6.0-6.6, and was 3-7% greater in grassland than subcanopy soils at all depths (P [is less than] 0.04), except in the 0-5-cm layer, where pH did not differ (P = 0.12, mean [+ or -] 1 SE = 6.5 [+ or -] 0.1).

Soil/microsite effects: seedling emergence

In the greenhouse, seedling emergence did not differ between grassland and subcanopy soil sources in 1995 (P = 0.99, mean [+ or -] 1 SE = 10 [+ or -] 1%) or 1996 (P = 0.18, 2 [+ or -] 1%). In the field, seedling emergence in 1995 was not affected by main or interactive effects of soil source and microsite (P [is greater than] 0.27, 26 [+ or -] 2%). In 1996, seedling emergence differed only between grassland (5 [+ or -] 2%) and subcanopy (12 [+ or -] 2%) microsites (P = 0.002).

Soil/microsite effects: seedling development

Seedling survival in soil columns differed between grassland and subcanopy microsites in Year 1 (1994-1995; P = 0.003), but not in Year 2 (1995-1996; P = 0.24; Table 4). In contrast, seedling survival in soil columns was not affected by soil source in either year (P [is greater than] 0.24).
TABLE 4. Characteristics of live Quercus emoryi seedlings in
1994-1995 (Year 1) and 1995-1996 (Year 2) at lower tree
line in southeastern Arizona, USA.

                                         Response variable

       Factor                     Survival         Height
Year   ([parallel])     Level       (%)             (cm)

 1     Microsite      Grassland    100(*)     8([sections])
                      Subcanopy     60(*)    12([sections])
       Soil           Grassland     85        8([double dagger])
                      Subcanopy     75       11([double dagger])
 2     Microsite      Grassland     70       16([sections])
                      Subcanopy     90        9([sections])
       Soil           Grassland     70       12
                      Subcanopy     90       12

                                         Response variable

                                                        Shoot
       Factor                          Leaf            biomass
Year   ([parallel])     Level         number             (g)

 1     Microsite      Grassland   25([sections])   0.8([dagger])
                      Subcanopy   14([sections])   0.6([dagger])
       Soil           Grassland   17(*)            0.6([sections])
                      Subcanopy   26(*)            1.0([sections])
 2     Microsite      Grassland   36([sections])   1.4([sections])
                      Subcanopy   10([sections])   0.4([sections])
       Soil           Grassland   20               0.8
                      Subcanopy   23               0.9

                                   Response variable

                                          Root
       Factor                           biomass
Year   ([parallel])     Level             (g)

 1     Microsite      Grassland   1.8([double dagger])
                      Subcanopy   0.9([double dagger])
       Soil           Grassland   1.2(*)
                      Subcanopy   1.9(*)
 2     Microsite      Grassland   1.9([sections])
                      Subcanopy   0.6([sections])
       Soil           Grassland   1.1
                      Subcanopy   1.2

                                   Response variable

                                         Root:
       Factor                            shoot
Year   ([parallel])     Level            ratio

 1     Microsite      Grassland   2.3([double dagger])
                      Subcanopy   1.6([double dagger])
       Soil           Grassland   2.2([dagger])
                      Subcanopy   1.8([dagger])
 2     Microsite      Grassland   1.5
                      Subcanopy   2.0
       Soil           Grassland   1.8
                      Subcanopy   1.8

                                   Response variable

                                         Total
       Factor                           biomass
Year   ([parallel])     Level             (g)

 1     Microsite      Grassland   2.7([double dagger])
                      Subcanopy   1.6([double dagger])
       Soil           Grassland   1.7(*)
                      Subcanopy   2.8(*)
 2     Microsite      Grassland   3.3([sections])
                      Subcanopy   1.0([sections])
       Soil           Grassland   1.9
                      Subcanopy   2.1

Notes: Seedlings were grown in soils collected from beneath mature
Q. emoryi tree canopies and from adjacent grassland that were
transferred to either subcanopy or grassland microsites (n = 10).
For levels within each year/factor treatment combination:
([dagger]) P < 0.10, (*) P < 0.05, ([double dagger]) P < 0.005,
([sections]) P < 0.0005.

([parallel]) Interactions between microsite and soil source were
not significant in either year (P > 0.24), so means represent main
effects.


Interactive effects of soil source and microsite on leaf number, shoot height, shoot, root, and total biomass, and root:shoot biomass ratios at experiment termination were not significant in either year (P [is greater than] 0.24). In both years, seedlings were 50-200% larger and produced 25-200% more above- and belowground biomass in grassland than subcanopy microsites (Table 4). In Year 1, seedlings were 40-70% larger and more productive in soils collected from beneath mature Q. emoryi trees than from the adjacent grassland. In contrast, soil source did not affect seedling development (P [is greater than] 0.48) in Year 2. Root:shoot ratios differed between grassland and subcanopy microsites only in Year 1 (P = 0.005), and were not affected by soil source in either year (P [is greater than] 0.06).

Main and interactive effects of soil source did not affect seedling root biomass distribution within experimental soil columns (P [is greater than] 0.10), except in Year 1, when seedlings produced 50% more biomass in soils collected from subcanopy (0.37 g/20 cm depth increment) than adjacent grassland (0.23 g/increment). In contrast, interactive effects of microsite and depth in soil affected growth during both years (P [is less than] 0.02): seedlings produced more root biomass at shallow depths in grassland than subcanopy microsites in both years (Fig. 4).

[Figure 4 ILLUSTRATION OMITTED]

Role of soil nutrients

Main and interactive effects of soil source and nutrient amendment did not affect leaf number, shoot height, shoot, root, and total biomass, and root:shoot ratios at experiment termination (P [is greater than] 0.11), with the exception of total root weight, which was greater (P = 0.04) in subcanopy (25.4 [+ or -] 2.3 g) than grassland soils (12.2 [+ or -] 1.5 g). Between-subjects effects did not influence incremental root biomass (P [is greater than] 0.22), except for soil source (P [is less than] 0.0001). However, an interaction between soil source and depth increment (P = 0.01) indicated that root biomass at shallow soil depths was greater in subcanopy soils than in grassland soils (Fig. 5). Interactive effects of nutrient treatment with depth were not significant (P [is greater than] 0.17), which indicates that root biomass apportionment with depth was not affected by nutrient amendments.

[Figure 5 ILLUSTRATION OMITTED]

Increases in seedling shoot height and leaf number throughout the experiment were affected only by soil source (P [is less than] 0.0008; P [is greater than] 0.14 for all other between-subject effects). Within-subjects effects for time and its interaction with soil source (P [is less than] 0.007) indicated that seedlings grew taller and produced more leaves (Fig. 6) in subcanopy soils than in grassland soils. Although nutrient amendments tended to accelerate development in grassland soils, interactive effects of soil source, nutrient amendment, and time did not affect height (P [is greater than] 0.23) or leaf number (P [is greater than] 0.11).

[Figure 6 ILLUSTRATION OMITTED]

Role of overstory shade

Seedling emergence.--Seedling emergence was affected by shade treatment, but this relationship depended on cohort and landscape zone (Table 5). Overstory shade (Contrast 1) increased emergence ~3- to 10-fold, depending on cohort. Shade effects were particularly pronounced in the grassland and interstitial zones (Contrasts 2 and 3, respectively), especially for cohort 2. Within the canopy zone, artificial and natural shade (Contrast 4) increased emergence three- to fivefold, depending on cohort. Emergence was two to three times greater beneath trees than in plots where the tree canopy had been removed (Contrast 6). When tree canopies were removed and replaced with artificial shade (Contrast 5), emergence did not differ from intact canopies for cohort 1, but tripled for cohort 2. Emergence did not differ between grassland, interstitial, and canopy zones (Contrasts 7, 8, and 9, respectively) for cohort 1, but emergence was greater in the canopy zone than in the grassland and interstitial zones for cohort 2. Overall, emergence rates ranged within 0-30%, which is consistent with other research with Q. emoryi within the region (Pase 1969, Nyandiga and McPherson 1992, Germaine and McPherson 1999).
TABLE 5. Emergence, survival at experiment termination, and
recruitment of Q. emoryi seedlings planted in 1994 (Cohort 1)
and 1995 (Cohort 2) in selected treatment combinations at lower
tree line in southeastern Arizona, USA.

                                                     Cohort 1

Contrast        Contrast         Treatment          Emergence
 number        description         codes               (%)

   1       Shaded                2, 4, 5, 7   16.3([sections])
           Unshaded              1, 3, 6       6.1([sections])
   2       Grassland, shaded     2            16.3([double dagger])
           Grassland, unshaded   1             4.8([double dagger])
   3       Interstitial,
             shaded              4            20.1([double dagger])
           Interstitial,
             unshaded            3             9.2([double dagger])
   4       Canopy, shaded        5, 7         14.5([sections])
           Canopy, unshaded      6             4.4([sections])
   5       Canopy intact         7            13.6
           Canope cut, shaded    5            15.3
   6       Canopy intact         7            13.6([double dagger])
           Canopy cut,
             unshaded            6             4.4([double dagger])
   7       Grassland             1, 2         10.6
           Interstitial          3, 4         14.7
   8       Interstitial          3, 4         14.7
           Canopy                5, 6, 7      11.1
   9       Grassland             1, 2         10.6
           Canopy                5, 6, 7      11.1

                                                    Cohort 1

Contrast        Contrast         Treatment          Survival
 number        description         codes               (%)

   1       Shaded                2, 4, 5, 7   38.9(*)
           Unshaded              1, 3, 6       2.1(*)
   2       Grassland, shaded     2            63.0([sections])
           Grassland, unshaded   1             7.1([sections])
   3       Interstitial,
             shaded              4            46.6(*)
           Interstitial,
             unshaded            3             0.0(*)
   4       Canopy, shaded        5, 7         17.6
           Canopy, unshaded      6             0.0
   5       Canopy intact         7             2.9(*)
           Canope cut, shaded    5            37.1(*)
   6       Canopy intact         7             2.9
           Canopy cut,
             unshaded            6             0.0
   7       Grassland             1, 2         50.0(*)
           Interstitial          3, 4         31.9(*)
   8       Interstitial          3, 4         31.9([double dagger])
           Canopy                5, 6, 7      15.1([double dagger])
   9       Grassland             1, 2         50.0([sections])
           Canopy                5, 6, 7      15.1([sections])

                                                    Cohort 1

Contrast        Contrast         Treatment          Recruit-
 number        description         codes            ment (%)

   1       Shaded                2, 4, 5, 7    6.3([sections])
           Unshaded              1, 3, 6       0.1([sections])
   2       Grassland, shaded     2            10.0([sections])
           Grassland, unshaded   1             0.3([sections])
   3       Interstitial,
             shaded              4             9.4([sections])
           Interstitial,
             unshaded            3             0.0([sections])
   4       Canopy, shaded        5, 7          2.6(*)
           Canopy, unshaded      6             0.0(*)
   5       Canopy intact         7             0.4([double dagger])
           Canope cut, shaded    5             5.7([double dagger])
   6       Canopy intact         7             0.4
           Canopy cut,
             unshaded            6             0.0
   7       Grassland             1, 2          5.3
           Interstitial          3, 4          4.7
   8       Interstitial          3, 4          4.7([double dagger])
           Canopy                5, 6, 7       1.7([double dagger])
   9       Grassland             1, 2          5.3([double dagger])
           Canopy                5, 6, 7       1.7([double dagger])

                                                     Cohort 2

Contrast        Contrast         Treatment          Emergence
 number        description         codes               (%)

   1       Shaded                2, 4, 5, 7   19.1([sections])
           Unshaded              1, 3, 6       1.8([sections])
   2       Grassland, shaded     2            18.8([sections])
           Grassland, unshaded   1             1.0([sections])
   3       Interstitial,
             shaded              4            17.1([sections])
           Interstitial,
             unshaded            3             0.0([sections])
   4       Canopy, shaded        5, 7         20.2([sections])
           Canopy, unshaded      6             4.4([sections])
   5       Canopy intact         7            10.5([sections])
           Canope cut, shaded    5            29.9([sections])
   6       Canopy intact         7            10.5(*)
           Canopy cut,
             unshaded            6             4.4(*)
   7       Grassland             1, 2          9.9
           Interstitial          3, 4          8.5
   8       Interstitial          3, 4          8.5([double dagger])
           Canopy                5, 6, 7      15.0([double dagger])
   9       Grassland             1, 2          9.9(*)
           Canopy                5, 6, 7      15.0(*)

                                                  Cohort 2

Contrast        Contrast         Treatment        Survival
 number        description         codes            (%)

   1       Shaded                2, 4, 5, 7   30.6
           Unshaded              1, 3, 6       0.0
   2       Grassland, shaded     2            45.9
           Grassland, unshaded   1             0.0
   3       Interstitial,
             shaded              4            32.0
           Interstitial,
             unshaded            3             ...
   4       Canopy, shaded        5, 7         20.1
           Canopy, unshaded      6             0.0
   5       Canopy intact         7             6.0(*)
           Canope cut, shaded    5            25.1(*)
   6       Canopy intact         7             6.0
           Canopy cut,
             unshaded            6             0.0
   7       Grassland             1, 2         43.0
           Interstitial          3, 4         31.7
   8       Interstitial          3, 4         31.7([dagger])
           Canopy                5, 6, 7      18.4([dagger])
   9       Grassland             1, 2         43.0([sections])
           Canopy                5, 6, 7      18.4([sections])

                                                    Cohort 2

Contrast        Contrast         Treatment          Recruit-
 number        description         codes            ment (%)

   1       Shaded                2, 4, 5, 7   5.8([sections])
           Unshaded              1, 3, 6      0.0([sections])
   2       Grassland, shaded     2            8.6([double dagger])
           Grassland, unshaded   1            0.0([double dagger])
   3       Interstitial,
             shaded              4            5.5([double dagger])
           Interstitial,
             unshaded            3            0.0([double dagger])
   4       Canopy, shaded        5, 7         4.1([double dagger])
           Canopy, unshaded      6            0.0([double dagger])
   5       Canopy intact         7            0.6([double dagger])
           Canope cut, shaded    5            7.5([double dagger])
   6       Canopy intact         7            0.6
           Canopy cut,
             unshaded            6            0.0
   7       Grassland             1, 2         4.3
           Interstitial          3, 4         2.7
   8       Interstitial          3, 4         2.7
           Canopy                5, 6, 7      2.8
   9       Grassland             1, 2         4.3
           Canopy                5, 6, 7      2.8

Note: Within contrasts, statistical significance was assessed using
ANOVA (emergence), Wald chi-square tests (survival), or Z tests
(recruitment): ([dagger]) P < 0.10; (*) P < 0.05; ([double dagger])
P < 0.005; ([sections]) P < 0.0005.


Seedling survivorship.--Survival of seedlings in each treatment did not differ between cohorts for the first three growing seasons (P [is greater than] 0.36), with the exception of shaded plots in the grassland (P = 0.04), where mortality rates of seedlings in cohort 2 exceeded those of cohort 1. Therefore, we present survivorship curves for cohort 1 only (Fig. 7). Survivorship curves generally were characterized by (1) low mortality during the first two months after emergence, (2) high mortality during 2-5 mo, and (3) extended periods of low mortality punctuated by periods of high seedling mortality during the annual seasonal drought between April and June.

[Figure 7 ILLUSTRATION OMITTED]

Seedling survivorship was dependent upon both treatment and cohort. Seedling survival was greater in shaded than unshaded plots (Contrast 1) for cohort 1, but not for cohort 2 (Table 5). Within the grassland and interstitial zones, artificial tree shade increased survival rates for cohort 1, but not for cohort 2 (Contrasts 2 and 3, respectively). Within the canopy zone, seedling survival for both cohorts did not differ between shaded (i.e., natural or artificial) and unshaded plots (Contrast 4). Relative to plots beneath intact canopies, seedling survival was not affected by tree canopy removal (Contrast 6). Tree canopy removal followed by provision of artificial shade increased survival for both cohorts (Contrast 5). Regardless of shade treatment, seedling survival rates for both cohorts were much greater in grassland than tree canopy plots (Contrast 9), and to a lesser extent, in interstitial than tree canopy plots (Contrast 8).

Seedling recruitment.--Seedling recruitment (i.e., the number of individuals added to the population) reflected the combined effects of emergence and subsequent survivorship within the different treatment combinations and cohorts (Table 5). Recruitment was much greater in shaded than unshaded plots (Contrasts 1, 2, 3, and 4). Absolute and relative differences in recruitment between shaded and unshaded plots were greatest in the grassland (Contrast 2) and interstitial zone (Contrast 3), especially for cohort 1. Relative to plots beneath intact canopies, seedling recruitment was not affected by tree canopy removal (Contrast 6). However, when tree canopies were removed and replaced with artificial shade (Contrast 5), recruitment rates increased 13- to 14-fold. For cohort 1, recruitment was greater in grassland and interstitial zones than in the canopy zone (Contrasts 8 and 9).

Seedling size.--Where comparisons could be made, seedlings were taller in shaded than unshaded plots (Contrasts 1 and 2), and they were taller in canopy plots where trees had been removed and replaced with artificial shade (Contrast 5; Table 6). Seedlings in the interstitial zone were taller than the grassland or canopy zones (Contrasts 7 and 8), but only for cohort 1.

Herbaceous production.--Herbaceous biomass did not differ between treatments for the first two growing seasons (i.e., 1994, 1995; data not shown), but differences in the third (1996) and fourth (1997) growing seasons reflected longer term effects of shade on herbaceous production (Table 6). In 1996, herbaceous biomass was about 1.5 x greater in shaded than unshaded plots (Contrast 1). Within the grassland and interstitial zones (Contrasts 2 and 3), provision of artificial shade roughly doubled herbaceous biomass. In 1997, these differences were attenuated and significant only for the interstitial zone (Contrast 3). In all years, herbaceous production was two to four times greater in interstitial and grassland plots than in canopy plots (Contrasts 8 and 9), and did not differ between grassland and interstitial plots (Contrast 7).
TABLE 6. Height (mean [+ or -] 1 SE) at experiment termination
(8 October 1997) of live Q. emoryi seedlings planted in 1994
(Cohort 1) and 1995 (Cohort 2), and herbaceous standing biomass at
the end of the 1996 and 1997 growing seasons, in selected treatment
combinations at lower tree line in southeastern Arizona, USA.

                                                Seedling height
                                                     (cm)
Contrast   Contrast               Treatment
 number    description              codes          Cohort 1

   1       Shaded                 2, 4, 5, 7   14.5 [+ or -] 2.0
           Unshaded               1, 3, 6             4.0
   2       Grassland, shaded      2             9.8 [+ or -] 3.4
           Grassland, unshaded    1                   4.0
   3       Interstitial, shaded   4            19.5 [+ or -] 2.1
           Interstitial,
             unshaded             3                   ...
   4       Canopy, shaded         5, 7         10.3 [+ or -] 1.5
           Canopy, unshaded       6                   ...
   5       Canopy intact          7                   8.0
           Canopy cut, shaded     5            17.2 [+ or -] 3.2
   6       Canopy intact          7                   8.0
           Canopy cut, unshaded   6                   ...
   7       Grassland              1, 2          9.0 [+ or -] 3.0
           Interstitial           3, 4         19.5 [+ or -] 2.1
   8       Interstitial           3, 4         19.5 [+ or -] 2.1
           Canopy                 5, 6, 7      14.1 [+ or -] 3.6
   9       Grassland              1, 2          9.0 [+ or -] 3.0
           Canopy                 5, 6, 7      14.1 [+ or -] 3.6

                                                Seedling height
                                                     (cm)
Contrast   Contrast               Treatment
 number    description              codes          Cohort 2

   1       Shaded                 2, 4, 5, 7   10.7 [+ or -] 1.4
           Unshaded               1, 3, 6             ...
   2       Grassland, shaded      2            12.1 [+ or -] 4.4
           Grassland, unshaded    1                   ...
   3       Interstitial, shaded   4            10.1 [+ or -] 1.7
           Interstitial,
             unshaded             3                   ...
   4       Canopy, shaded         5, 7         11.0 [+ or -] 1.7
           Canopy, unshaded       6                   ...
   5       Canopy intact          7                   7.0
           Canopy cut, shaded     5            11.0 [+ or -] 1.7
   6       Canopy intact          7                   7.0
           Canopy cut, unshaded   6                   ...
   7       Grassland              1, 2          8.7 [+ or -] 4.4
           Interstitial           3, 4         10.1 [+ or -] 1.7
   8       Interstitial           3, 4         10.1 [+ or -] 1.7
           Canopy                 5, 6, 7      10.3 [+ or -] 1.5
   9       Grassland              1, 2          8.7 [+ or -] 4.4
           Canopy                 5, 6, 7      10.3 [+ or -] 1.5

                                               Herbaceous biomass
                                                  (g/[m.sup.2])
Contrast   Contrast               Treatment
 number    description              codes             1996

   1       Shaded                 2, 4, 5, 7     184([dagger])
           Unshaded               1, 3, 6        127
   2       Grassland, shaded      2              294(*)
           Grassland, unshaded    1              167(*)
   3       Interstitial, shaded   4              296(*)
           Interstitial,
             unshaded             3              144(*)
   4       Canopy, shaded         5, 7            72
           Canopy, unshaded       6               71
   5       Canopy intact          7               83
           Canopy cut, shaded     5               61
   6       Canopy intact          7               83
           Canopy cut, unshaded   6               71
   7       Grassland              1, 2           230
           Interstitial           3, 4           220
   8       Interstitial           3, 4           220([sections])
           Canopy                 5, 6, 7         72([sections])
   9       Grassland              1, 2           230([sections])
           Canopy                 5, 6, 7         72([sections])

                                               Herbaceous biomass
                                                  (g/[m.sup.2])
Contrast   Contrast               Treatment
 number    description              codes              1997

   1       Shaded                 2, 4, 5, 7   158
           Unshaded               1, 3, 6      142
   2       Grassland, shaded      2            281
           Grassland, unshaded    1            216
   3       Interstitial, shaded   4            260(*)
           Interstitial,
             unshaded             3            117(*)
   4       Canopy, shaded         5, 7          46
           Canopy, unshaded       6             92
   5       Canopy intact          7             50
           Canopy cut, shaded     5             42
   6       Canopy intact          7             50
           Canopy cut, unshaded   6             92
   7       Grassland              1, 2         248
           Interstitial           3, 4         189
   8       Interstitial           3, 4         189([double dagger])
           Canopy                 5, 6, 7       62([double dagger])
   9       Grassland              1, 2         248([sections])
           Canopy                 5, 6, 7       62([sections])

Note: Within contrasts statistical significance was assessed using
ANOVA: ([dagger]) P < 0.10; (*) P < 0.05; ([double dagger])
P < 0.005; ([sections]) P < 0.0005.


Soil moisture.--Soil moisture contents were affected by combined effects of shade treatment, soil depth, and sample date (data are in Appendix B). Soil moisture at 10 cm was greater in shaded than unshaded plots (Contrast 1) on 6 of 12 sample dates (29 September 1994; 15 April, 26 June, and 16 August 1995; 10 April and 10 July 1996). However, within the grassland, interstitial, and canopy zones (Contrasts 2-4), shade increased soil moisture on only 2 of 12 dates (29 September 1994 and 15 April 1995). Removal of the tree canopy increased soil moisture at 10 cm on only three dates, which differed depending on whether the soil was shaded (Contrast 5; 29 September 1994; 31 January and 10 April 1996) or unshaded (Contrast 6; 10 November 1995; 31 January and 13 November 1996). There was no discernible relationship between antecedent precipitation and treatment on soil moisture at 10 cm.

At 50 cm, soil moisture was greater in shaded than unshaded plots (Contrast 1) on 4 of 12 dates (29 September 1994; 15 April and 26 June 1995; and 10 April 1996), which corresponded to periods of little antecedent precipitation. Within the grassland, interstitial, and canopy zones (Contrasts 2-4), shade increased soil moisture on only three dates (29 September 1994, 15 April 1995, and 10 April 1996), two dates (29 September 1994 and 15 April 1995), and one date (29 September 1994) of 12 dates, respectively. Removal of the tree canopy (Contrast 6) increased soil moisture at 50 cm on only two dates (19 August 1994 and 10 November 1995). In contrast, when the tree canopy was replaced with artificial shade (Contrast 5), soil moisture was increased on 7 of 12 dates (29 September 1994; 29 January, 26 June, 10 November 1995; 31 January, 10 April, and 13 November 1996) corresponding to dry periods.

DISCUSSION

Although Q. emoryi seedlings are abundant beneath mature tree canopies at lower tree line, they are infrequent within the grassland phase of the savanna, and are absent from grasslands adjacent to and below current lower tree line (Fig. 1, Plate 1). Landscape-level patterns of seedling distribution resulted from interactions between seed dispersal and habitat-specific response of seedlings at different life history stages to environmental variation (cf. Harper 1977, Houle 1995, Schupp and Fuentes 1995). Results from this study indicate that rates of Q. emoryi recruitment within grasslands below tree line are relatively low and are constrained by relatively low rates of seed dispersal (Table 2), coupled with a low probability of seedling emergence (Table 5). In contrast, large numbers of acorns are dispersed directly to Q. emoryi understories, where they have a higher probability of emergence than in adjacent grassland. Survival rates of emerged seedlings are low, regardless of landscape position.

Q. emoryi acorn distribution and dispersal

Dispersed acorns were located primarily on the soil surface beneath mature Q. emoryi at lower tree line (Table 2). Relatively few whole acorns or acorn components were present in the grassland only 3.5 m from the edge of the canopies of these trees in either year. No acorns were viable at the times of our surveys. Hubbard and McPherson (in press) conducted a similar survey of acorn dispersal at a nearby site in 1993 and 1995, and they found acorns beneath mature Q. emoryi at lower tree line (at densities of about 34 acorns/[m.sup.2]), and at 20 m (3 acorns/[m.sup.2]), 30 m ([is less than] 0.1 acorn/[m.sup.2]), and 50 m (0.4 acorn/[m.sup.2]) distance into grassland adjacent to lower tree line. Results of these two studies indicate that Q. emoryi acorns are dispersed into grasslands adjacent to lower tree line, but at rates considerably lower than beneath mature Q. emoryi.

Acorns are not substantially dispersed by wind, water, or seed morphology (Jensen and Neilsen 1986). Instead, acorns are usually dispersed by vertebrates that drop them during transport or fail to relocate acorn caches (Darley-Hill and Johnson 1981, Hubbard and McPherson 1997). At a nearby Q. emoryi savanna site, vertebrate acorn predators included Mexican Jays (Aphelocoma ultramarina), Acorn Woodpeckers (Melanerpes formicivorus), White-winged Doves (Zenaida asiatica), Peccaries (Dicotyles xacu), Coati (Nasua nasua), and Sonora white-tailed deer (Odocoileus virginianus couesi) (Hubbard and McPherson, in press). Although the relative contribution of these and other species to spatial and temporal patterns of acorn predation are unknown, annual acorn removal from woodlands exceeded 99% in seed cafeteria trials (Hubbard and McPherson, in press). In contrast, once acorns were dispersed to adjacent grassland, their rate of removal was relatively low. Post-removal rates and patterns of acorn predation and dispersal for these vertebrates have not been described. In sum, acorn dispersal into grassland adjacent to lower tree line occurs at rates that are apparently insufficient to enable subsequent seedling recruitment, given current environmental conditions.

Q. emoryi seedling distribution and population dynamics

Beneath tree canopies, the abundance of seedlings less than one year old was much greater on ungrazed than grazed sites (Fig. 1). In contrast, the density of seedlings greater than one year old did not differ between these sites in either year (Fig. 2). Interpretations of the effect of livestock grazing per se on Quercus density are complicated by minor differences in soil characteristics between sites (Table 1). However, comparable abundance of seedlings in older age classes between sites suggests that (1) direct utilization of Q. emoryi seedlings by livestock is rare, infrequent, or sporadic (McClaran et al. 1992); (2) periodic mast production of seed (Sanchini 1981, Sork 1993), or episodic establishment of seedlings during favorable climatic conditions (sensu Blackburn and Tueller 1970, McPherson and Wright 1990), facilitates seedling recruitment into older age classes, irrespective of grazing history or soil characteristics; or (3) patterns of grazing management (e.g., rest rotation) may enable occasional seedling recruitment. Further, the observed age-dependent response of Quercus seedlings to livestock grazing history and soil characteristics may help explain inconsistencies in reported effects of livestock on Quercus recruitment in the southwestern United States, which range from insignificant (Muick and Bartolome 1987, McClaran et al. 1992) to detrimental (Pase 1969, Bahre 1977, McClaran and Bartolome 1989, 1990).

Comparisons of seedling abundance between tree understories and adjacent canopy gaps potentially confound microsite effects with canopy effects on seedling establishment (sensu Bartolome et al. 1994). For example, our observations of low herbaceous cover beneath Q. emoryi relative to adjacent canopy gaps and grasslands (Table 1) are consistent with observations of individual Q. emoryi tree suppression of herbaceous production by about 40% at GW (Haworth and McPherson 1994). Because growth and survival of Q. emoryi seedlings are detrimentally affected by herbaceous interference (McPherson 1993, Germaine and McPherson 1999), concentrations of seedlings beneath tree canopies could be attributed to a relative lack of herbaceous interference.

However, it seems unlikely that differences in herbaceous biomass between subcanopy, interstitial, and grassland zones were the sole contributor to observed differences in seedling densities. First, Germaine and McPherson (1999) reported that Q. emoryi emergence rates were not affected by experimental removal or reductions in herbaceous interference. Second, in this study, net recruitment of seedlings from planted acorns was comparable between unshaded grassland, unshaded interstitial, and intact tree plots (Table 5), even though herbaceous biomass was two to three times greater in grassland and interstitial plots than intact canopy plots (Table 6). Finally, woody-plant seedling survival may be affected less by simple changes in biomass of herbaceous neighbors than by more complex interactions between gross soil moisture supply and demand for that soil moisture by herbaceous neighbors (Davis et al. 1998).

The relative abundance of seedlings less than one year old beneath tree canopies differed 10-fold between 1993 and 1995. Such interannual variations in seedling density are consistent with other observations of Q. emoryi seedling establishment in the region. Sanchini (1981) reported mean Q. emoryi seedling densities of 300, 0, and 309 plants/ha in 1978, 1979, and 1980, respectively. In this study, differences in seedling densities between 1993 and 1995 probably were caused by differences in summer (July-September) precipitation, which was 86% of mean summer precipitation in 1993 and 57% of the mean in 1995. This interpretation is supported by other research that demonstrates that Q. emoryi seedlings are dependent upon the availability of summer precipitation for germination and early establishment in this system (Weltzin and McPherson, in press). Although Q. emoryi seedlings and native grasses compete for water or nutrients (McClaran and McPherson, in press), differences in seedling densities between years could not be attributed to herbaceous interference, which did not differ between years on some sites, and was greater in 1993 than in 1995 on other sites.

Patterns of seedling distribution observed in this study are comparable to other studies that found relatively high Quercus seedling survival rates beneath mature conspecifics relative to adjacent grasslands (Neilson and Wullstein 1983, Muick and Bartolome 1987, Bragg et al. 1993). However, previous research on conspecific seedling association with mature Quercus trees has not investigated effects of tree size on subcanopy seedling dynamics. Results of this study suggest that tree size (i.e., height, canopy area, basal area) is an important factor affecting subcanopy seedling density, but that this effect is attenuated by livestock grazing or soil characteristics.

Soil properties vs. microsite effects

Soil nutrient and organic carbon contents were higher beneath Q. emoryi trees than in adjacent grassland (Table 3), which is consistent with other observations of soil nutrient accumulation beneath woody plants in this (McPherson et al. 1993) and numerous other semiarid systems (see reviews by McPherson [1997], Scholes and Archer [1997], Wilson [1998]). Alternatively, the possibility of distinct edaphic or geologic differences between tree and grass-dominated sites (sensu Bartolome et al. 1994) is suggested by differences in content of rock fragments.

Despite observed differences in soil nutrient contents in the field, experimental nutrient amendments to subcanopy and grassland soils in the greenhouse did not affect above- or belowground growth of Q. emoryi seedlings. This suggests that soil nutrients do not limit seedling growth in either subcanopy or grassland soils, and that nutrient contents in grassland soils are adequate for seedling growth.

In the same greenhouse experiment, seedlings grown in subcanopy soils produced more belowground biomass, as well as producing taller shoots with more leaves at faster rates, than did seedlings grown in grassland soils (Figs. 5 and 6). This suggests that physical or chemical characteristics of grassland and subcanopy soils, other than soil nutrient contents, contribute to observed differences in seedling size and productivity. It is unlikely that soil textures, which were comparable (Table 3), or soil pH, which differed by [is less than or equal to] 7%, contributed greatly to differences in seedling development.

Differences in the volume of rock fragments between subcanopy and grassland soils may have altered the distribution of soil moisture in experimental columns in a manner conducive to seedling development. In particular, the relatively high volume of rock fragments in subcanopy soils may have displaced infiltrating water to deeper depths than in grassland soils, where it could be used more effectively by these tap-rooted seedlings. Although rock fragments in subsurface soils have often been reported as beneficial to juvenile and mature woody plants, especially in arid and semiarid regions (Jackson et al. 1972, Munn et al. 1987, Kadmon et al. 1989, Albaladejo 1990, Kosmas et al. 1994), this hypothesis has not been tested experimentally in this system and has seldom been tested in any other unmanaged system (Poesen and Lavee 1994).

Both microsite and soil source affected growth of seedlings in reciprocally transferred subcanopy and grassland soils in the field. The fact that seedlings were more productive in grassland than subcanopy microsites, in both years and regardless of soil source, suggests the overriding importance of microclimatic conditions to seedling development. Relatively low growth rates beneath tree canopies could have resulted from (1) the mean 90% reductions in subcanopy light intensity by the tree canopy (cf. Parker and Muller 1982, Espelta et al. 1995, Ashton and Larson 1996), or (2) interception of precipitation by the tree canopy. Within this savanna, Q. emoryi trees the size of those used in this study reduce precipitation by up to 70% directly beneath the canopy, depending on precipitation event size (Haworth and McPherson 1995). Potential canopy-induced reductions in evapotranspiration (cf. Tiedemann and Klemmedson 1973, Breshears et al. 1998) were apparently not sufficient to overcome light or water limitations to growth of these seedlings, which were in soil columns isolated from neighboring plants.

In contrast with microsite effects on seedling development, soil source effects were year dependent: seedlings produced more biomass in subcanopy than grassland soils in Year 1, but there was no effect in Year 2. Interannual variation in seedling production probably was caused by differences in growing season (March-October) precipitation, which was 50% of mean in Year 1 and 83% of mean in Year 2. In a dry year, the high proportion of rock fragments in subcanopy soils may have been conducive to infiltration of water to deeper depths than in grassland soils, where it could be used more effectively by these tap-rooted seedlings. In a wetter year, differences in soil particle size distribution would be less important (sensu Poesen and Lavee 1994). Interactive effects of soil rock fragment content and interannual variation in precipitation amount have not been experimentally investigated.

Overstory shade affects seedling demography

The results of the artificial-shade experiment, coupled with the observed concentration of seedlings at northern aspects, illustrate the importance of overstory shade on Q. emoryi recruitment. Overstory shade was important at all stages of seedling development investigated: the provision of artificial or natural shade increased seedling emergence three-fold for cohort 1 and as much as 19-fold for cohort 2, survival as much as 19-fold, and recruitment 30-60-fold. However, shade effects on seedling demography were dependent upon where, in relation to lower tree line, acorns were experimentally "dispersed." Seedling emergence was greatly enhanced with the provision of artificial shade within the grassland and interstitial zones. This suggests that, given sufficient dispersal of acorns, seedling emergence is constrained by physical environments created by a lack of overstory shade. Similarly, Q. emoryi acorns placed beneath trees had viability rates more than twice those placed in adjacent grassland zones (Nyandiga and McPherson 1992).

Survivorship curves suggest that cohorts responded similarly to interannual processes that contributed to seedling mortality. In contrast, intra-annual patterns of survival for both cohorts were similar, which illustrates the importance of the premonsoon drought and the subsequent monsoon on seedling demography. Intra-annual patterns of seedling survival were comparable to those observed by Germaine and McPherson (1999).

Recruitment of Quercus seedlings in semiarid regions is often observed to be greater in shaded microsites associated with tree or shrub canopies than unshaded microsites (Bray 1960, Griffin 1971, 1980, Neilson and Wullstein 1983, Muick and Bartolome 1987, Callaway and D'Antonio 1991, Callaway 1992a, Johnson 1992, Espelta et al. 1995). In addition, the importance of shade for seedling establishment in California oak savannas has been demonstrated experimentally (Muick 1991, Williams et al. 1991, Callaway 1992a). Although beneficial effects of shade have been attributed to direct effects on seedling physiology (e.g., photosynthesis and transpiration; Williams et al. 1991, Callaway 1992b), mechanisms by which shade benefits seedlings are poorly understood (but see Davis et al. [1998]).

Elucidation of mechanisms by which overstory shade benefits seedlings is complicated by indirect effects of shade on seedling microenvironments that may also affect seedling performance. For example, shade-induced reductions in stomatal conductance and transpiration of coexisting grasses may increase the availability of soil water (Amundson et al. 1995), which in turn may be utilized by Quercus seedlings (cf. Espelta et al. 1995, Policy et al. 1997). Or, reductions in subcanopy herbaceous biomass caused by overstory shade (e.g., Callaway et al. 1991, Haworth and McPherson 1994) may minimize detrimental effects of herbaceous interference on Quercus seedlings (Gordon and Rice 1993, McPherson 1993, Koukoura and Menke 1995). Ultimately, environmental factors that reduce soil water availability may reduce seedling survival by intensifying competition with herbaceous neighbors, whereas factors that increase water availability may increase seedling success by decreasing competition with neighboring plants (Davis et al. 1998).

Soil moisture limitations and drought often limit oak seedling survival in semiarid or seasonally arid environments (Pase 1969, Griffin 1971, Neilson and Wullstein 1983, Borchert et al. 1989, Gordon et al. 1989, Welker and Menke 1990, Gordon and Rice 1993, Germaine and McPherson 1998). Accordingly, the distribution of oaks is considered dependent upon gradients of moisture availability and interspecific variation in soil water utilization (Griffin 1973, 1977, Matsuda and McBride 1986, Pigott and Pigott 1993). However, results of this study suggest that soil moisture limitations may be influenced by soil particle size distribution and microenvironmental amelioration by adult tree canopies (cf. Neilson 1993).

Safe sites and seed-seedling conflicts

The obvious distribution of mature Q. emoryi trees at lower tree line, and patterns of seed and seedling demography, indicate that mature oak trees form safe sites (Harper et al. 1961) conducive to conspecific seedling establishment, relative to the grassland phase between savanna trees and the grassland below lower tree line. However, differences in suitability between these potential habitats lead to patch-dependent and stage-dependent differences in seed viability and germination (Nyandiga and McPherson 1992), and seedling emergence, growth, and survival (Germaine and McPherson 1999). For example, although Q. emoryi tree understories form safe sites for seeds, seedlings did not consistently exhibit habitat preferences. Thus, patch suitabilities for seeds and seedlings were neither concordant nor discordant, but were uncoupled (Houle 1995, Schupp 1995, Schupp and Fuentes 1995). Although an uncoupled and heterogeneous response of seeds and seedlings to environmental variation can be expressed in terms of recruitment to a given demographic stage, this approach overlooks important mechanisms that constrain stage-specific transition probabilities (Harper 1977, Herrera et al. 1994, Houle 1995).

Our research focused on Q. emoryi seedling recruitment within only the first four years, and did not include extreme climatic or disturbance events. Therefore, predictions of the ultimate distribution of adult plants on the landscape are constrained without more specific investigation of the transition from juvenile to adult plants. However, because mortality is usually concentrated during early stages of recruitment (Harper 1977), distribution patterns of adults are probably more closely aligned with the distribution of older seedlings than younger seedlings or dispersed seeds.

Seed-seedling conflicts arise when safe sites for seeds are not necessarily safe sites for seedlings (Schupp 1995). Although seed-seedling conflicts can result from differential response of seeds and seedlings to physical environments alone (e.g., Sork 1985), habitat-dependent rates of seed predation or seedling herbivory often form the context for seed-seedling conflicts (Schupp 1995). In this context, our ability to assess seed-seedling conflicts was constrained by intentional exclusion of herbivores from experimental plots. In fact, Q. emoryi acorns are more likely to escape predation in grasslands than woodlands (Hubbard and McPherson, in press). In addition, episodic vertebrate herbivory can affect survival and development of established Q. emoryi seedlings (Germaine and McPherson 1999).

Dynamics of lower tree line

We suggest herein that biogenic facilitation and self-enhancing feedback mechanisms constrain spatial and temporal shifts in the oak woodland-semidesert grassland ecotone of western North America. Similarly, spatial stability of savannas and ecotones in both humid and arid regions of the world has been attributed to positive feedbacks between trees and tree seedling establishment (e.g., Menaut et al. 1990, Skarpe 1991, Wilson and Agnew 1992). These feedback effects may be augmented or attenuated by other processes that affect seedling establishment (e.g., competition, herbivory, fire; Berkowitz et al. 1995, Scholes and Archer 1997, Malanson 1997, Germaine and McPherson 1999).

Wilson and Agnew (1992) review the role of positive feedback mechanisms as constraints on plant population dynamics and develop a conceptual model of the vegetation positive-feedback switch, or "switch" (sensu Odum 1971). A switch occurs when an environment becomes more suitable for an extant population as a result of that population's persistence through time. Consequently, initial differences in vegetation are magnified, and vegetation boundaries on spatial scales from individual plants to landscapes are sharpened or enhanced (Wilson and Agnew 1992, Malanson 1997). Switches also constrain temporal boundaries: vegetation dynamics (e.g., shifts in ecotone position) can be accelerated (when invading species alter environmental conditions in their favor) or delayed (by prolonging initial patch composition or delaying the response to climate change; Wilson and Agnew 1992).

Results from this research indicate that overstory shade, seed dispersal, and seedling establishment comprise a positive-feedback switch that sharpens spatially and stabilizes temporally the oak woodland-semidesert grassland ecotone: Q. emoryi seedlings can establish under conspecific adults, but, without sufficient seed dispersal and overstory shade, they are constrained from adjacent grassland only meters away. This positive-feedback switch is one-sided in that Q. emoryi can establish within grassland, given sufficient seed dispersal coupled with environmental conditions conducive to seedling establishment. This is consistent with recent suggestions that the relative abundance of trees and grasses in most savannas is inherently unstable in the absence of disturbance (Ludwig et al. 1997, Scholes and Archer 1997).

However, the positive-feedback switch operating at this ecotone, albeit one sided, probably delays both upslope and downslope shifts in this ecotone barring extreme climatic episodes or disturbance (cf. Kullman 1989, Noble 1993). Once established, Q. emoryi at lower tree line should persist because they are long-lived (Sanchini 1981) and are capable of vegetative reproduction following top removal by fire or drought (Bahre 1991, Babb 1992, Caprio and Zwolinski 1995). During unfavorable years, seedling recruitment and survival rates are buffered by biogenic amelioration of subcanopy environments. Soil resource partitioning may further contribute to the stable coexistence of mature trees and grasses (e.g., Walter 1954, Walker and Noy-Meir 1982, Weltzin and McPherson 1997). Although periods of reduced summer precipitation during the last century have been associated with periods of oak top kill or mortality at lower tree line (Leopold 1924, Marshall 1957, Hastings and Turner 1965), Bahre's (1991) analysis of photographs and other historical evidence indicates that the position of lower tree line has been stable since at least 1872. Reconstruction of historic and earlier distribution of Q. emoryi is complicated, in part, because stem age can not be reliably determined with current dendrochronological techniques (McPherson 1992; T. W. Swetnam, personal communication).

McPherson et al. (1993) and McClaran and McPherson (1995) used soil organic carbon isotopes in their investigation of spatial and temporal shifts in oak savanna-semidesert grassland ecotones of southern Arizona. They concluded that lower tree line shifted downslope into adjacent grassland about 700-1700 yr BP, which coincides with a period of particularly high summer precipitation in the region (the "Medieval Warm" period, 645-1295 yr BP; Davis 1994). Recent experimental research further illustrates the importance of summer precipitation to Q. emoryi seedling emergence, survival, and recruitment at the savanna-grassland ecotone (Weltzin and McPherson, in press).

However, conditions conducive to downslope shifts in tree line apparently occur infrequently in southern Arizona. First, the relative position of the savanna-grassland ecotone has been stable over the last millennium (McClaran and McPherson 1995). Second, population size class structures (as proxy for age) exhibit normal distributions (Sanchini 1981), which suggests that processes of recruitment and mortality have been relatively constant for the last two centuries (the approximate life span of Q. emoryi). Furthermore, analyses of available climate records for the region indicate few directional trends in seasonal precipitation over the past century (Bahre 1991). In sum, these lower tree lines exhibit considerable biological inertia, in terms of relative slope position, to all but climatic variations that occur on the scale of centuries to millennia (cf. Neilson and Wullstein 1983, Kullman 1989).

Recruitment of Q. emoryi within grasslands, given sufficient dispersal, is likely dependent on the formation of safe sites conducive to seedling emergence and survival. Safe sites could be established within grasslands by either autogenic or allogenic processes, or a combination of both processes. Autogenic shifts in tree line would involve spatial progression of biogenic safe sites, wherein seedlings establish on the downslope side of trees at lower tree line and, while growing to reproductive maturity, extend the original safe site to include their subcanopy. Distributional shifts of lower tree line would be limited to relatively small spatial and long temporal scales. In contrast, allogenic shifts in tree line could occur when climatic conditions simulate, or negate the importance of, conspecific, biogenic safe sites. Resultant shifts in tree line could occur across larger spatial scales over shorter temporal scales. For example, a series of wet summers may provide soil moisture sufficient for seedling establishment without conspecific overstory shade, with concomitant increases in Q. emoryi recruitment and a downslope shift in lower tree line (Weltzin and McPherson, in press). Results from this study, coupled with those of other research described above, suggest that although downslope shifts in lower tree line in southern Arizona have occurred (McPherson et al. 1993), this ecotone is relatively stable and is resistant to decadal or even century-scale climatic perturbation (e.g., the regional drought of the 1950s). Stability and biological inertia are the result of a one-sided positive-feedback switch generated by self-enhancing feedback mechanisms of overstory shade, seed dispersal, and seedling establishment, coupled with strong abiotic constraints beyond the current ecotone. Furthermore, observed downslope shifts in lower tree line (McPherson et al. 1993) are less likely the result of spatial progression of autogenic safe sites than episodic and infrequent allogenic processes that simulate, or negate the importance of, conspecific, biogenic safe sites.

ACKNOWLEDGMENTS

Seedling and seed bank surveys were completed only with the sharp eyes and nimble fingers of R. A. Abolt, H. Germaine, A. Hubbard, S. Lowe, R. Matson, M. Meixner, S. Merrigan, S. Miller, A. Nicholas, K. Rojahn, K. Snyder, K. Suedkamp, J. Villanueva-Diaz, A. Wigg, and personnel from the Fort Huachuca Military Reservation (FHMR) Office of Game Management (OGM). Some of these people, as well as L. Abbott, D. Angell, J. Atchley, R. Brooks, K. Clayton, G. Fisher, P. Ford, P. R. Meixner, M. O'Dea, B. Orr, P. Roller, J. Stone, G. Soroka, K. Weltzin, and P. Weltzin assisted with experiment establishment, maintenance, monitoring, and disassembly. E. Pendall described soil profiles at GW. This research could not have been conducted without the assistance and patient facilitation of Sheridan Stone, FHMR-OGM, and the help of J. Miller and J. Murray, FHMR, and D. Bennett, U.S. Forest Service Sierra Vista Ranger District. B. Steidl provided statistical advice. C. Canham, L. Graumlich, J. Klemmedson, M. McClaran, B. Steidl, T. Thompson, D. Williams and two anonymous reviewers provided helpful comments on earlier drafts. Research was supported by USDA-CSRS National Research Initiative Grant #92-37101-7435 and McIntire-Stennis project ARZT-139017-M-12-118. J. F. Weltzin was supported in part by a UA/NASA Space Grant Graduate Student Fellowship, a Flinn Foundation Biology.21 Graduate Fellowship, a William G. McGinnies Scholarship in Arid Lands Studies, and a Graduate College Fellowship.

LITERATURE CITED

Albaladejo, J. 1990. Impact of degradation processes on soil quality in arid Mediterranean environments. Pages 193-214 in J. Rubio and J. Rickson, editors. Strategies to combat desertification in Mediterranean Europe. Commission of the European Communities, Report EUR 11175.

Allison, P. D. 1995. Survival analysis using the SAS system: a practical guide. SAS Institute, Cary, North Carolina, USA.

Amundson, R. G., A. R. Ali, and A. J. Belsky. 1995. Stomatal responsiveness to changing light intensity increases rain-use efficiency of below-crown vegetation in tropical savannas. Journal of Arid Environments 29:139-153.

Archer, S. 1994. Woody plant encroachment into southwestern grasslands and savannas: rates, patterns and proximate causes. Pages 13-68 in M. Vavra, W. Laycock, and R. Pieper, editors. Ecological implications of livestock herbivory in the West. Society for Range Management, Denver, Colorado, USA.

--. 1995. Tree-grass dynamics in a Prosopis--thornscrub savanna parkland: reconstructing the past and predicting the future. Ecoscience 2:83-99.

Archer, S. A., D. S. Schimel, E. A. Holland. 1995. Mechanisms of shrubland expansion: land use or [CO.sub.2]? Climatic Change 29:91-99.

Ashton, M. S., and B. C. Larson. 1996. Germination and seedling growth of Quercus (section Erythrobalanus) across openings in a mixed-deciduous forest of southern New England, USA. Forest Ecology and Management 80: 81-94.

Bahre, C. J. 1977. Land-use history of the Research Ranch, Elgin, Arizona. Journal of the Arizona Academy of Science 12(supplement 2): 1-32.

--. 1991. A legacy of change. University of Arizona Press, Tucson, Arizona, USA.

Bartolome, J. W., B. H. Allen-Diaz, and W. D. Tietje. 1994. The effect of Quercus douglasii removal on understory yield and composition. Journal of Range Management 47: 151-154.

Bazzaz, F. A., and P. M. Wayne. 1994. Coping with environmental heterogeneity: the physiological ecology of tree seedling across the gap-understory continuum. Pages 349-390 in M. M. Caldwell and R. W. Pearcy, editors. Exploitation of environmental heterogeneity by plants. Ecophysiological processes above- and belowground. Academic Press, San Diego, California, USA.

Belsky, A. J., R. G. Amundson, J. M. Duxbury, S. J. Riha, A. R. Ali, and S. M. Mwonga. 1989. The effects of trees on their physical, chemical, and biological environments in a semiarid savanna in Kenya. Journal of Applied Ecology 26:1005-1024.

Belsky, A. J., and C. D. Canham. 1994. Forest gaps and isolated savanna trees: an application of patch dynamics in two ecosystems. BioScience 44:77-84.

Berkowitz, A. R., C. D. Canham, and V. R. Kelly. 1995. Competition vs. facilitation of tree seedling growth and survival in early successional communities. Ecology 76: 1156-1168.

Blackburn, W. H., and P. T. Tueller. 1970. Pinyon and juniper invasion in black sagebrush communities in east-central Nevada. Ecology 51:841-848.

Borchert, M., F. W. Davis, J. Michaelsen, and L. D. Oyler. 1989. Interactions of factors affecting seedling recruitment of blue oak (Quercus douglasii) in California. Ecology 70: 389-404.

Borelli, S., P. F. Ffolliott, and G. J. Gottfried. 1994. Natural regeneration in encinal woodlands of southeastern Arizona. Southwestern Naturalist 39:179-183.

Bragg, W. K., A. K. Knapp, and J. M. Briggs. 1993. Comparative water relations of seedling and adult Quercus species during gallery forest expansion in tallgrass prairie. Forest Ecology and Management 56:29-41.

Bray, J. R. 1960. The composition of savanna vegetation in Wisconsin. Ecology 41:721-732.

Breshears, D. D., J. W. Nyhan, C. E. Heil, and B. P. Wilcox. 1998. Effects of woody plants on microclimate in a semiarid woodland: soil temperature and evaporation in canopy and intercanopy patches. International Journal of Plant Sciences 159:1010-1017.

Brown, D. E., editor. 1982. Biotic communities of the American southwest--United States and Mexico. Desert Plants 4:1-342.

Callaway, R. M. 1992a. Effects of shrubs on recruitment of Quercus douglasii and Quercus lobata in California. Ecology 73:2118-2128.

--. 1992b. Morphological and physiological responses of three California oak species to shade. International Journal of Plant Science 153:434-441.

Callaway, R. M., and C. M. D'Antonio. 1991. Shrub facilitation of coast live oak establishment in central California. Madrono 38:158-169.

Callaway, R. M., E. H. DeLucia, D. Moore, R. Nowak, and W. H. Schlesinger. 1996. Competition and facilitation: contrasting effects of Artemisia tridentata on desert vs. montane pines. Ecology 77:2130-2141.

Callaway, R. M, N. M. Nadkarni, and B. E. Mahall. 1991. Facilitation and interference of Quercus douglasii on understory productivity in central California. Ecology 72: 1484-1499.

Canfield, R. H. 1941. Application of the line interception method in sampling range vegetation. Journal of Forestry 39:388-394.

Clements, F. E., J. Weaver, and H. Hansson. 1926. Plant competition: an analysis of the development of vegetation. Carnegie Institute, Washington, D.C., USA.

Darley-Hill, S., and W. C. Johnson. 1981. Acorn dispersal by the blue jay (Cyanocitta cristata). Oecologia 50:231-232.

Davis, M. A., K. J. Wrage, and R B. Reich. 1998. Competition between tree seedlings and herbaceous vegetation: support for a theory of resource supply and demand. Journal of Ecology 86:652-661.

Davis, O. K. 1994. The correlation of summer precipitation in the southwestern USA with isotopic records of solar activity during the Medieval Warm Period. Climatic Change 26:271-287.

Day, P. R. 1965. Particle fractionation and particle-size analysis. Pages 545-567 in C. A. Black, editor. Methods of soil analysis, part 1. Physical and mineralogical properties. Agronomy Monographs 9. American Society of Agronomy and Soil Science Society of America, Madison, Wisconsin, USA.

De Steven, D. 1991a. Experiments on mechanisms of tree establishment in old-field succession: seedling emergence. Ecology 72:1066-1075.

--. 1991b. Experiments on mechanisms of tree establishment in old-field succession: seedling survival and growth. Ecology 72:1076-1088.

Dreimanis, A. 1962. Quantitative gasometric determination of calcite and dolomite by using Chittick apparatus. Journal of Sedimentary Petrology 32:520-529.

Espelta, J. M., M. Riba, and J. Retana. 1995. Patterns of seedling recruitment in west-Mediterranean Quercus ilex forests influenced by canopy development. Journal of Vegetation Science 6:465-472.

Evans, R. A., and R. M. Love. 1957. The step point method of sampling--a practical tool in range research. Journal of Range Management 10:208-212.

Fisher, R. A. 1958. Statistical methods for research workers. Thirteenth edition. Hafner, New York, New York, USA.

--. 1960. The design of experiments. Seventh edition. Hafner, New York, New York, USA.

Franco, A. C. and P. S. Nobel. 1989. Effect of nurse plants on the microhabitat and growth of cacti. Journal of Ecology 77:870-886.

Garcia-Moya, E., and C. M. McKell. 1970. Contribution of shrubs to the nitrogen economy of a desert-wash plant community. Ecology 51:81-88.

Germaine, H. L., and G. R. McPherson. 1998. Effects of timing of precipitation and acorn harvest date on emergence of Quercus emoryi. Journal of Vegetation Science 9:157-160.

Germaine, H. L., and G. R. McPherson. 1999. Effects of biotic factors on emergence and survival of Quercus emoryi at lower tree line, Ecoscience 6:92-99.

Gordon, D. R., and K. J. Rice. 1993. Competitive effects of grassland annuals on soil water and blue oak (Quercus douglasii) seedlings. Ecology 74:68-82.

Gordon, D. R., J. M. Welker, J. W. Menke, and K. J. Rice. 1989. Competition for water between annual plants and blue oak (Quercus douglasii) seedlings. Oecologia 79:533-541.

Griffin, J. R. 1971. Oak regeneration in the upper Carmel Valley, California. Ecology 52:862-868.

--. 1973. Xylem sap tension in three woodland oaks of central California. Ecology 54:152-159.

--. 1977. Oak woodland. Pages 383-415 in M. Barbour and J. Major, editors. Terrestrial vegetation of California. John Wiley, New York, New York, USA.

--. 1980. Animal damage to valley oak acorns and seedlings, Carmel Valley, California. Pages 242-245 in T. R. Plumb, technical coordinator. Proceedings of the symposium on the ecology, management, and utilization of California oaks. U.S. Forest Service General Technical Report PSW-44.

Grover, H. D., and H. B. Musick. 1990. Shrubland encroachment in southern New Mexico, USA: an analysis of desertification processes in the American Southwest. Climatic Change 17:305-330.

Grubb, P. J. 1977. The maintenance of species richness in plant communities: the importance of the regeneration niche. Biological Review 52:107-145.

Gurevitch, J., and S. L. Collins. 1994. Experimental manipulation of natural plant communities. Trends in Ecology and Evolution 9:94-98.

Harper, J. L. 1977. Population biology of plants. Academic Press, London, UK.

--. 1982. After description. Pages 11-24 in E. I. Newman, editor. The plant community as a working mechanism. The British Ecological Society, Oxford, UK.

Harper, J. L., J. N. Clatworthy, I. H. McNaughton, and G. R. Sagar. 1961. The evolution of closely related species living in the same area. Evolution 15:209-277.

Hastings, J. R. and R. M. Turner. 1965. The changing mile. University of Arizona Press, Tucson, Arizona, USA.

Haworth, K., and G. R. McPherson. 1994. Effects of Quercus emoryi on herbaceous vegetation in a semiarid savanna. Vegetatio 112:153-159.

Haworth, K., and G. R. McPherson. 1995. Effects of Quercus emoryi trees on precipitation distribution and microclimate in a semiarid savanna. Journal of Arid Environments 31: 153-170.

Herrera, C. M., P. Jordano, L. Lopez-Soria, and J. A. Amat. 1994. Recruitment of a mast-fruiting, bird-dispersed tree: bridging frugivore activity and seedling establishment. Ecological Monographs 64:315-344.

Hill, J. D., C. D. Canham, and D. M. Wood. 1995. Patterns and causes of resistance to tree invasion in rights-of-way. Ecological Applications 5:459-470.

Houle, G. 1995. Seed dispersal and seedling recruitment: the missing link(s). Ecoscience 2:238-244.

Hubbard, J. A., and G. R. McPherson. 1997. Acorn selection by Mexican jays: a test of a tri-trophic symbiotic relationship hypothesis. Oecologia 110:143-146.

Hubbard, J. A., and G. R. McPherson. In press. Do seed predation and dispersal limit downslope movement of the semidesert grassland/oak woodland ecotone in the American Southwest? Journal of Vegetation Science.

Jackson, J. T, and O. W. Van Auken. 1997. Seedling survival, growth and mortality of Juniperus ashei (Cupressacae) in the Edwards Plateau region of central Texas. Texas Journal of Science 49:267-278.

Jackson, L. P., I. V. Hall, and L. E. Aalders. 1972. Lowbush blueberry seedling growth as affected by soil type. Canadian Journal of Soil Science 52:113-115.

Jensen, T S., and O. F. Neilson. 1986. Rodents as seed dispersers in a heath-oak wood succession. Oecologia 70: 214-221.

Joffre, R., and S. Rambal. 1988. Soil water improvement by trees in the rangelands of southern Spain. Acta Oecologia 9:405-422.

Johnson, P. S. 1992. Oak overstorey/reproduction relations in two xeric ecosystems in Michigan. Forest Ecology and Management 48:233-248.

Jordan, P. W., and P. S. Nobel. 1979. Infrequent establishment of seedlings of Agave deserti (Agavaceae) in the northern Mojave Desert. American Journal of Botany 66:1076-1084.

Kadmon, R., A. Yair, and A. Danin. 1989. Relationship between soil properties, soil moisture, and vegetation along loess-covered hillslopes, northern Negev, Israel. Catena Supplement 14:43-57.

Keeney, D. R., and D. W. Nelson. 1982. Nitrogen--inorganic forms. Pages 643-698 in A. L. Page, editor. Methods of soil analysis, part 2. Chemical and microbiological properties. Second edition. Agronomy Monographs 9. American Society of Agronomy and Soil Science Society of America, Madison, Wisconsin, USA.

Kollmann, J., and H.-P. Schill. 1996. Spatial patterns of dispersal, seed predation and germination during colonization of abandoned grassland by Quercus petraea and Corylus avellana. Vegetatio 125:193-205.

Kosmas, C., N. Moustakas, N. G. Danalatos, and N. Yassoglou. 1994. The effect of rock fragments on wheat biomass production under highly variable moisture conditions in Mediterranean environments. Catena 23:191-198.

Koukoura, Z., and J. Menke. 1995. Competition for soil water between perennial bunch-grass (Elymus glaucus B.B.) and blue oak seedlings (Quercus douglasii H. and A.). Agroforestry Systems 32:225-235.

Kruskal, W. H., and W. A. Wallis. 1952. Use of ranks in one-criterion analysis of variance. Journal of the American Statistical Association 47:583-621.

Kullman, L. 1989. Recent retrogression of the forest-alpine tundra ecotone (Betula pubescens Ehrh. ssp. tortuosa (Ledeb.) Nyman) in the Scandes Mountains, Sweden. Journal of Biogeography 16:83-90.

Leopold, A. 1924. Grass, brush, timber and fire in southern Arizona. Journal of Forestry 22:1-10.

Ludwig, D., B. H. Walker, and C. S. Holling. 1997. Sustainability, stability, and resilience. Conservation Ecology (1):7 [online: <http://www.consecol.org/voll/iss1/art7>].

Malanson, G. P. 1997. Effects of feedbacks and seed rain on ecotone patterns. Landscape Ecology 12:27-38.

Marshall, J. T. 1957. Birds of the pine-oak woodland in southern Arizona and adjacent Mexico. Pacific Coast Avifauna Number 32, Cooper Ornithological Society, Berkeley, California, USA.

Matsuda, K., and J. R. McBride. 1986. Difference in seedling growth morphology as a factor in the distribution of three oaks in central California. Madrono 33:207-216.

McClaran, M. P., L. S. Allen, and G. B. Ruyle. 1992. Livestock production and grazing management in the encinal oak woodlands of Arizona. Pages 57-64 in P. F. Ffoliott, G. J. Gottfried, D. A. Bennett, V. M. Hernandez C., A. Ortega-Rubio, and R. H. Hamre, technical coordinators. Ecology and management of oak and associated woodlands: perspectives in the southwestern United States and northern Mexico. U.S. Forest Service General Technical Report RM-218.

McClaran, M. P., and J. W. Bartolome. 1989. Fire related recruitment in stagnant Quercus douglasii populations. Canadian Journal of Forest Research 19:580-585.

McClaran, M. P., and J. W. Bartolome. 1990. Comparison of actual and predicted blue oak age structures. Journal of Range Management 43:61-63.

McClaran, M. P., and G. R. McPherson. 1995. Can soil organic carbon isotopes be used to describe grass-tree dynamics at a savanna-grassland ecotone and within the savanna? Journal of Vegetation Science 6:857-862.

McClaran, M. P., and G. R. McPherson. In press. Oak savanna in the American southwest. In R. C. Anderson, J. S. Fralish, and J. Baskin, editors. Savanna, barren and rock outcrop plant communities of North America. Cambridge University Press, New York, New York, USA.

McLean, E. O. 1982. Soil pH and lime requirement. Pages 199-224 in A. L. Page, editor. Methods of soil analysis, part 2. Chemical and microbiological properties. Second edition. Agronomy Monographs 9. American Society of Agronomy and Soil Science Society of America. Madison, Wisconsin, USA.

McPherson, G. R. 1992. Ecology of oak woodlands in Arizona. Pages 24-33 in P. F. Ffoliott, G. J. Gottfried, D. A. Bennett, V. M. Hernandez C., A. Ortega-Rubio, and R. H. Hamre, technical coordinators. Ecology and management of oak and associated woodlands: perspectives in the southwestern United States and northern Mexico. U.S. Forest Service General Technical Report RM-218.

--. 1993. Effects of herbivory and herbs on oak establishment in a semiarid temperate savanna. Journal of Vegetation Science 4:687-692.

--. 1997. Ecology and management of North American savannas. University of Arizona Press, Tucson, Arizona, USA.

McPherson, G. R., T. W. Boutton, and A. J. Midwood. 1993. Stable carbon isotope analysis of soil organic matter illustrates vegetation change at the grassland/woodland boundary in southeastern Arizona, USA. Oecologia 93:95-101.

McPherson, G. R., G. A. Rasmussen, D. B. Wester, and R. A. Masters. 1991. Vegetation and soil zonation associated with Juniperus pinchotii Sudw. trees. Great Basin Naturalist 51:316-324.

McPherson, G. R., and H. A. Wright. 1990. Establishment of Juniperus pinchotii in western Texas: environmental effects. Journal of Arid Environments 19:283-287.

Menaut, J. C., J. Gignoux, C. Prado, and J. Clobert. 1990. Tree community dynamics in a humid savanna of the Cote-d'Ivoire: modelling the effects of fire and competition with grass and neighbors. Journal of Biogeography 17:471-481.

Muick, P. C. 1991. Effects of shade on blue oak and coast live oak regeneration in California annual grasslands. Pages 21-24 in B. Standiford, editor. Symposium on oak woodlands and hardwood rangeland management. U.S. Forest Service General Technical Report PSW-126.

Muick, P. C., and J. W. Bartolome. 1987. An assessment of natural regeneration of oaks in California. Report to California Department of Forestry, Forest and Rangeland Resource Assessment Program, California, USA.

Munn, L., N. Harrington, and D. R. McGirr. 1987. Rock fragments. Pages 259-282 in R. D. Williams and G. E. Schuman, editors. Reclaiming mine soils and overburden in the western United States. Analytic parameters and procedures. Soil Conservation Society of America, Ankeny, Iowa, USA.

Neilson, R. P. 1993. Transient ecotone response to climatic change: some conceptual and modeling approaches. Ecological Applications 3:385-395.

Neilson, R. P., and L. H. Wullstein. 1983. Biogeography of two southwest American oaks in relation to atmospheric dynamics. Journal of Biogeography 10:275-297.

Nelson, D. W., and L. E. Sommers. 1982. Total carbon, organic carbon and organic matter. Pages 539-580 in A. L. Page, editor. Methods of soil analysis, part 2. Chemical and microbiological properties. Second edition. Agronomy Monographs 9. American Society of Agronomy and Soil Science Society of America. Madison, Wisconsin, USA.

Noble, I. R. 1993. A model of the responses of ecotones to climate change. Ecological Applications 3:396-403.

Nyandiga, C. O., and G. R. McPherson. 1992. Germination of two warm-temperate oaks, Quercus emoryi and Quercus arizonica. Canadian Journal of Forest Research 22:1395-1401.

Odum, H. T. 1971. Environment, power, and society. Wiley-Interscience, New York, New York, USA.

Olsen, S. R., and L. E. Sommers. 1982. Phosphorus. Pages 404-430 in A. L. Page, editor. Methods of soil analysis, part 2. Chemical and microbiological properties. Second edition. Agronomy Monographs 9. American Society of Agronomy and Soil Science Society of America. Madison, Wisconsin, USA.

Parker, V. T., and C. H. Muller. 1982. Vegetational and environmental changes beneath isolated live oak trees (Quercus agrifolia) in a California annual grassland. American Midland Naturalist 107:69-81.

Pase, C. P. 1969. Survival of Quercus turbinella and Quercus emoryi in an Arizona chaparral community. Southwestern Naturalist 14:149-156.

Petranka, J. W., and J. K. McPherson. 1979. The role of Rhus copallina in the dynamics of the forest-prairie ecotone in north-central Oklahoma. Ecology 60:956-965.

Pigott, C. D., and S. Pigott. 1993. Water as a determinant of the distribution of trees at the boundary of the Mediterranean zone. Journal of Ecology 81:557-566.

Poesen, J., and H. Lavee. 1994. Rock fragments in top soils: significance and processes. Catena 23:1-28.

Polley, H. W., H. B. Johnson, H. S. Mayeux, and C. R. Tischler. 1996. Are some of the recent changes in grassland communities a response to rising [CO.sub.2] concentrations? Pages 177-195 in Christian Korner and F. A. Bazzaz, editors. Carbon dioxide, populations, and communities. Academic Press, San Diego, California, USA.

Polley, W. H., H. S. Mayeux, H. B. Johnson, C. R. Tischler. 1997. Viewpoint: atmospheric [CO.sub.2], soil water, and shrub/ grass ratios on rangelands. Journal of Range Management 50:278-284.

Prentice, I. C. 1986. Vegetation responses to past climatic variation. Vegetatio 67:131-141.

Primack, R. B., and S. L. Miao. 1992. Dispersal can limit local plant distribution. Conservation Biology 6:513-519.

Richardson, M. L., S. D. Clemmons, and J. C. Walker. 1979. Soil survey of Santa Cruz and part of Cochise and Pima Counties, Arizona. Soil Conservation Service, U.S. Department of Agriculture, Washington, D.C., USA.

Sanchini, P. J. 1981. Population structure and fecundity patterns in Quercus emoryi and Quercus arizonica in southeastern Arizona. Dissertation. University of Colorado, Boulder, Colorado, USA.

SAS Institute. 1989. SAS/STAT User's Guide, version 6. Fourth edition. SAS Institute, Cary, North Carolina, USA.

--. 1996. SAS/STAT software: changes and enhancements through Release 6.11. SAS Institute, Cary, North Carolina, USA.

Scheiner, S. M. 1993. MANOVA: Multiple response variables and multispecies interactions. Pages 94-112 in S. M. Scheiner and J. Gurevitch, editors. Design and analysis of ecological experiments. Chapman and Hall, New York, New York, USA.

Schlesinger, W. H., J. F. Reynolds, G. L. Cunningham, L. Huenneke, W. M. Jarrell, R. A. Virginia, and W. G. Whitford. 1990. Biological feedbacks in global desertification. Science 247:1043-1048.

Scholes, R. J., and S. R. Archer. 1997. Tree-grass interactions in savannas. Annual Review of Ecology and Systematics 28:517-544.

Schupp, E. W. 1995. Seed-seedling conflicts, habitat choice, and patterns of plant recruitment. American Journal of Botany 82:399-409.

Schupp, E. W., and M. Fuentes. 1995. Spatial patterns of seed dispersal and the unification of plant population ecology. Ecoscience 2:267-275.

Shapiro, S. S., and M. B. Wilk. 1965. An analysis of variance test for normality (complete samples). Biometrika 52:591-611.

Skarpe, C. 1991. Spatial patterns and dynamics of woody vegetation in an arid savanna. Journal of Vegetation Science 2:565-572.

Soil Conservation Service. 1994. Soil survey of Fort Huachuca, Cochise County, Arizona. Prepared for the U.S. Department of the Army. U.S. Department of Agriculture, Soil Conservation Service, Washington, D.C., USA.

Soil Survey Staff. 1994a. National soil survey handbook. U.S. Department of Agriculture, Soil Conservation Service. Washington, D.C., USA.

--. 1994b. Keys to soil taxonomy. Sixth edition. U.S. Department of Agriculture, Soil Conservation Service. Washington, D.C., USA.

Sork, V. L. 1985. Germination response in a large-seeded neotropical tree species, Gustavia superba (Lecythidaceae). Biotropica 17:130-136.

--. 1993. Evolutionary ecology of mast-fruiting in temperate and tropical oaks (Quercus spp.). Vegetatio 107/108: 133-147.

Steenbergh, W. F., and C. H. Lowe. 1969. Critical factors during the first years of life of the saguaro (Cereus giganteus) at Saguaro National Monument. Ecology 50:825-834.

Steenbergh, W. F., and C. H. Lowe. 1977. Ecology of the saguaro. II. Reproduction, germination, establishment, growth, and survival of the young plant. National Park Service Science Monograph Series number 8. National Park Service, Washington, D.C., USA.

Tiedemann, A. R., and J. O. Klemmedson. 1973. Effects of mesquite on physical and chemical properties of the soil. Journal of Range Management 26:27-29.

Tiedemann, A. R., and J. O. Klemmedson. 1977. Effect of mesquite trees on vegetation and soils in the desert grassland. Journal of Range Management 30:361-367.

Vetaas, O. R. 1992. Micro-site effects of trees and shrubs in dry savannas. Journal of Vegetation Science 3:337-344.

von Ende, C. N. 1993. Pages 113-137 in S. M. Scheiner and J. Gurevitch, editors. Design and analysis of ecological experiments. Chapman and Hall, New York, New York, USA.

Walker, B. H., and I. Noy-Meir. 1982. Aspects of the stability and resilience of savanna ecosystems. Pages 556-590 in B. J. Huntley and B. H. Walker, editors. Ecology of tropical savannas. Springer-Verlag, Berlin, Germany.

Walter, H. 1954. Die verbuschung, eine erscheinung der subtropischen savannengebiete, und ihre okologischen urscachen. Vegetatio 5/6:6-10.

Welker, J. M., and J. W. Menke. 1990. The influence of simulated browsing on tissue water relations, growth, and survival of Quercus douglasii (Hook and Am.) seedlings under slow and rapid rates of soil drought. Functional Ecology 4:807-817.

Weltzin, J. F., S. Archer, and R. K. Heitschmidt. 1997. Small-mammal regulation of vegetation structure in a temperate savanna. Ecology 78:751-763.

Weltzin, J. F., and G. R. McPherson. 1995. Potential effects of climate change on lower tree lines in the southwestern United States. Pages 180-193 in L. F. DeBano, G. J. Gottfried, R. H. Hamre, C. B. Edminster, P. F. Ffolliott, and A. Ortega-Rubio, technical coordinators. Biodiversity and management of the Madrean Archipelago: the sky islands of southwestern United States and Northwestern Mexico. U.S. Forest Service General Technical Report RM-264.

Weltzin, J. F., and G. R. McPherson. 1997. Spatial and temporal soil moisture resource partitioning by trees and grasses in a temperate savanna, Arizona, USA. Oecologia 112: 156-164.

Weltzin, J. F., and G. R. McPherson. In press. Implications of precipitation redistribution for shifts in temperate savanna ecotones. Ecology.

Williams, K., S. D. Davis, B. L. Gartner, and S. Karlsson. 1991. Factors limiting the establishment of a chaparral oak, Quercus durata Jeps., in grassland. Pages 70-73 in Symposium on oak woodlands and hardwood rangeland management. U.S. Forest Service General Technical Report PSW-126.

Wilson, J. B., and A. D. Q. Agnew. 1992. Positive-feedback switches in plant communities. Advances in Ecological Research 23:263-336.

Wilson, S. D. 1998. Competition between grasses and woody plants. Pages 231-254 in G. R Cheplick, editor. Population biology of grasses. Cambridge University Press, Cambridge, UK.

Young, M. D., and O. T. Solbrig. 1993. The world's savannas: economic driving forces, ecological constraints and policy options for sustainable land use. Parthenon Publishing Group, Carnforth, UK.

Zar, J. H. 1996. Biostatistical analysis. Prentice Hall, Upper Saddle River, New Jersey, USA.

APPENDIX A

A color version of Plate 1, showing a representative woodland-grassland ecotone, is available in ESA's Electronic Data Archive: Ecological Archives M069-005-A1.

APPENDIX B

Supplemental data relating to the effect of shade, soil depth, and sample date on soil moisture contents are available in ESA's Electronic Data Archive: Ecological Archives M069-005-A2.

JAKE F. WELTZIN(1) AND GUY R. MCPHERSON

School of Renewable Natural Resources, University of Arizona, Tucson, Arizona 85721 USA

Manuscript received 7 July 1998; revised 23 October 1998; accepted 16 November 1998.

(1) Current address: Department of Ecology and Evolutionary Biology, University of Tennessee, Knoxville, Tennessee 37996 USA. E-mail: jweltzin@utk.edu
COPYRIGHT 1999 Ecological Society of America
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1999 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Weltzin Jake F.; McPherson, Guy R.
Publication:Ecological Monographs
Geographic Code:1USA
Date:Nov 1, 1999
Words:18970
Previous Article:RESPONSES OF TUNDRA PLANTS TO EXPERIMENTAL WARMING: META-ANALYSIS OF THE INTERNATIONAL TUNDRA EXPERIMENT.
Next Article:PLANT REMOVALS IN PERENNIAL GRASSLAND: VEGETATION DYNAMICS, DECOMPOSERS, SOIL BIODIVERSITY, AND ECOSYSTEM PROPERTIES.
Topics:

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters