Diversity and Distributions (2001) 7, 1-14

: .. .. Effects of physical disturbance and granivory on establishment of native and alien riparian trees in Colorado, U.S.A.

GABRIELLE L. KATZ', JONATHAN M. FRIEDMAN' and SUSAN W BEATTY' I Geography Department, University of Colorado, Boulder, CO 80309, 'US Geological Suney, Midcontinent Ecological Science Center, F/. Coflins, Colorodo, U.SA.

Abstract. In western North Ameriea, the alien Elaeagnus angustifalia L. invades riparian hab­itats usually dominated by pioneer woody species such as Populus del/aides Marshall ssp. manilifera (Aiton) Eekenwalder. We conducted manipulative field experiments to compare the importanee of physieal disturbanee and granivory for seedling establishment of these two species. We planted seeds of both species in disturbed and undis­turbed study plots, and used exclosures, seed dish rrials and live-trapping to assess the role of granivory. Seedling establishment of both species was increased by physical disturbance and seeds of both species were subject to granivory. How­ever, the relarive importance of these two factors differed between species. For P delta ides , lack of physical disturbance prevented seedling

establishment in uncleared subplots, but granivory did not prevent seedling establishment outside of exelosures. For E. angustifolia, granjvory prevented seedling establishment outside of exelosures, but lack of physieal disturbance did not pre­vent establishment in uncleared subplots. The lesser dependence on disturbance may enable E angustifolia to invade areas characterized by low levels of fluvial disturbance, such as floodplains along regulated rivers, where P del/oides recruit­ment does not occur. Populations of granivorous rodents may affect the susceptibility of riparian ecosystems to invasion by E. angustifolia.

Key words. Biological invasions, disturbance, Elaeagnus, Populus, riparian vegetation, seedling establishment.

INTRODUCTION

Physical processes play an important role in both limiting and enabling seedling establishment of floodplain species worldwide. Climatic variability, streamflow and local abiotic variables related to geomorphic position (e.g. soil moisture and rates of erosion and deposition) have been used to explain spatial and temporal patterns of floodplain rree recruitment in a variety of contexts (Streng elol.. 1989; Baker, 1990; Kalliola el 01., 1991; Sacclu & Price, 1992; Scott el al., 1997; Mahoney & Rood, 1998; Stromberg, 1998; Cooper el aI.,

Correspondence: Gabrielle L Kalz, Geography Depanmenl, Campns Box 260, University of Colorado. Boulder, CO 80309, USA. E-mail: kaLzg@.samiam coJorado.edu

1999). Although these physical variables may play the dominant role in structuring western riparian plant communities, biotic interactions undoubtedly influence community structure as well (Sacchi & Price, 1992; Naiman & Rogers, 1997; Andersen & Cooper, 2000). On some regulated rivers, the relative importance of biotic interactions may be enhanced due to the reduced frequency or inten­sity of flood disturbance (Lesica & Miles, 1999; Andersen & Cooper, 2000).

Granivory is a biotic process which has an important influence on vegetatIOn dynamies in many settings (Smith, 1987; Brown & Reske, 1990; Asquith el al., 1997; Reader, 1997). Because granivores may preferentially consume seeds of certain species, or groups of species (e.g, grass seeds. Brown & Heske, 1990; or larger seeds,

© 200l Blackwell Science Ltd. hLtpJfwww.blackwell-sc.ence.comfddi

2 G. Katz,). Friedman and S. W Beatty

Reader, 1993), granivory potentially influences the rate and direction of vegetation change. Granivory can therefore influence processes such as colonization, succession (Davidson, 1993; Reader, 1997) and alien plant invasion (Bossard, 1991; Steele el aI., 1999).

Despite the demonstrated importance of granivory in other enVironments, studies of biotic interactions in riparian settings have focused largely on seedling eompetition (Cooper ec 01., 1999), herbivory a nd habitat alteration by native mammals (Naiman & Rogers, 1997; Lesica & Miles, 1999; Andersen & Cooper, 2000) and the impacts of introduced livestock (Stromberg, 1997; Auble & Scott, 1998; Belsky el aI., 1999). Tn studies of riparian seedling establishment in western North America, post-establishment biotic processes such as competition and herbivory are generally thought to overprint initial establishment patterns determined largely by abiotic factors (Sacehi & Price, 1992; Cooper et al., 1999; Andersen & Cooper, 2000). In this study, we investigate the possibility that granivory may be an addi­tional limiting factor operating prior to riparian tree seedling establishment. We also address the question of whether granivory, through its influ­ence on propagule availability, may influence alien riparian plant invasion.

Elaeagnus anguslifolia L. (Russian olive) is a non-native Eurasian tree that invades riparian habitats in much of western North America (Olson & Knopf, 1986). Introduced for ornamental pur­poses by 1900 (Hansen, 1901), and planted widely for wildlife habitat and windbreaks beginning in the 1930s (Read, 1958), E. angustlfolia is now found on streambanks, in riparian forest under­storeys, along riparian forest edges and in moist meadows (Currier, 1982; Knopf & Olson, 1984; Howe & Knopf, 1991). Although E. angustl/olia has some value as wildlife habitat, avian spe­cies nchness and diversity tend to be lower in E. anguSlifolia stands compared to native ripar­ian forests (Knopf & Olson, 1984). Where E. angustifolia Invades native riparian forest under­storeys, it can ultimately dominate the sites once the overstorey trees die (Currier, 1982). Because E. anguslifolia is actinorhizal (Miller & Baker, 1985), its invasion potentially alters ecosystem nutrient levels through the addition of nitrogen-rich organic matter to floodplain soils (de Castro et al., 1990; Vitousek. 1990; Simons & Seastedt, 1999).

The native woody vegetation in the habitats that E. angustifolia invades is usually dominated by pioneer species, typically members of the Family Salicaceae (cottonwoods and willows) Populus del/aides Marshall ssp. monilifera (Aiton) Eckenwalder (plains cottonwood), the primary riparian cottonwood of the southern Great Plains reglOn of North America, possesses many traits characteristic of bottomland pioneer species (White, 1979). It produces abundant small seeds (e.g. mean seed mass of 6.04 x 10-4 g in eastern Colorado, Friedman el al., 1995), which are dis­persed by wind and water in the early summer (Braatne et al., 1996). The seeds are not dormant, and remain viable for only a few weeks after dispersal (Braatne el aI., 1996). Ideal seedhng establishment sites are moist and free of com­peting vegetation and groundcover (Friedman et al., 1995). Such disturbed patehes are typically created by erosion and sediment deposition accom­panying natural flow fluctuations.

In contrast, E. angustijolia possesses a repro­ductive strategy more typical of a late succes­sional species. In the autumn, it produces edible single-seeded fruits which can be dispersed by birds and mammals. The seeds are relatively large (0.08 g, Young & Young, 1992), and are dorm­ant when dispersed, generally requiring a period of cold moist after-ripening for germination to occur (Hogue & LaCroix, 1970). Seeds produced in a given year may germinate across multiple growing seasons, whenever conditions at a par­ticular site become suitable (Shafroth et al., 1995). Observations that E. angustifolia invades Into wet meadows (Currier, 1982) and beneath forest canopies (Campbell & Diek-Peddie, 1964; Lesica & Miles, 1999) suggest that seedlings are tolerant of a broad range of moisture and light conditions, although little published research has addressed this issue (Shafroth et aI., 1995; Lesica & Miles, 1999). Like other vertebrate­dispersed seeds, those of E. ang-ustifolia may be subject to significant post-dispersal granivory (Hulme, 2997).

Populus recruitment along formerly meander­ing ehannels has been curtailed downstream of dams on many western rivers because dam opera­tion often results in reduced downstream flow variability and reduced creation of suitable seed­ling establishment siles (Fenner et al., 1985; Bradley & Smith, 1986; Johnson, 1992; Friedman

© 200 I Blackwell SCience Ltd, Diversity afld DistributiOfls, 7, 1-\4

3 Riparian tree seedling establishment

er al., 1998). Some authors have argued that such regulated systems are especially vulnerable to invasion by E. angusti/olia because its larger seeds allow establishment III the absence of physical disturbance (Shafroth el 01., 1995; Lesica & Miles, 1999). However, although it has been demonstrated that P del/aides seedling establishment is limited largely by the availability of bare moist estab­lishment sites (Friedman el al., 1995), it is not clear to what extent E. angusri/olia establishment depends upon similar sites. Controlled studies are necessary to test whether the two species actually differ in their dependence on disturbance.

The goals of this study were to use manipulat­ive field experiments to investigate the influ­ence of (l) physical disturbance, and (2) granivory on seedling establishment of P del/oides and E. angus/i/o/ia. We planted seeds of both species in disturbed and undisturbed plots and used exclosures, seed dish trials and live-trapping to assess the importance of granivory.

METHODS

Study area

This study was conducted in Boulder County, Colorado on the south shore of Teller Lake (40°2' north, 105°8' west), an irrigation pond owned by the City of Boulder. Teller Lake is located in the southeru Great Plains physiographic region of North America (Osterkamp et ai., 1987), close to the foothills of the Colorado Front Range of the Rocky Mountains at elevation 1585 m. The climate in this area is semi-arid and continental; average annual precipitation is 46.9 em (Colorado Climate Center, unpublished dala). The mean num­ber of days between the last spring frost and the firsl autumn frost is 154 (Colorado Climate Center, unpublished data).

Our experiment was conducted within 50 m of the edge of Teller Lake, in an area that was an agricultural field until 1993. The vegetation at the study site was dominated by herbs, especially iva xanlhi/o/ia and the alien Kochia scoparia (nomenclature follows Great Plains Flora Associ­ation, 1986; alien designations follow Weber & Wittman, 1996). Other abundant herbs included the aliens Salsolo australis, Xanthium slrumarium, and COl1yo/yulus anensis, and the natives Ambrosia psilostachya and Dislich/is stricta.

Study deSIgn

The experimental design consisted of 15 4 x 4 m study plots. The plots were arranged in three lines of five, perpendicular to the slope of the lake shore. Each plot was fenced (5 x 8 cm wire mesh), and contained a central circular sprinkler mounted at a height of 55 cm. All plots were irrigated to eliminate drought stress. Water was pumped from Teller Lake to the study plots up to three times daily. Irrigation was adjusted in order to ensure that the soil surface was satur­ated in the plots without producing run-off. The average daily duration and depth of irrigation were 23 min and 0.8 cm.

The subplot treatments were designed to com­pare the effects of physical disturbance on seed­ling establishment of the two target species. Each plot was divided into four 2 x 2 m quarters or subplots (Fig. I). In two randomly assigned sub­plot quarters per plot, Rodeo® herbicide was applied with a backpack sprayer (30 mL Rodeo® per 3.8 L of water) in May 1998, and all vegeta­tion and plant litter was then removed by hand from 10 to 16 June 1998 (= 'cleared' treatment). Throughou t the experiment, the cleared sub­plots were maintained by hand weeding as needed. In the remaining two subplots, vegetation and litter were left intact (= 'uncleared' treatment). Vegetation in the uncleared SUbplots was kept at a height of 45 em by trimming at 2-week intervals.

Ground surface light levels were measured at midday (between 11 : 00 and 14 : 00 h) in all sub­plots between l5 and 25 September 1998 using aLI-COR LI-190SA Quantum Sensor. In each subplot, light levels were measured at two loca­tions, with four light readings taken at the cor­ners of a 10 x 10 cm square configuration at each location. Within eaeh large plot (n = 15), the mean ground surface light level for each treat­ment type (cleared vs. uncleared) was calculated, and these values were compared using analysis of variance (SAS Institute Inc., 1989).

Within each plot, one randomly assigned cleared subplot and one randomly assigned uncleared subplot received E. angustlfolia seeds, and the remaining two subplots received co[\onwood seeds (Fig. 1). All seeds used in the experiments were collected from adult trees in Boulder County, Colorado that appeared to have established

© 2001 Blackwell Science LId, Diversify and DWl'ibuliollS, 7, 1-14

4 G. Katz,). Friedman and S. W Beatty

2m • Exciosure cage

Fig. I Sample plot design. Cleared subplot quarters are white, and uncleared subplots are shaded. Only the centre 1 x I m square of each subplot was censused for seedling density.

from naturally dispersed seed. Ripe E. anguslifolia fruits were eolleeted from at least 30 naturalized

trees during the faU of 1997. Fruits were stored

at room temperature until March 1998, at which

time the outer pulp was removed from the

seeds by hand maceration. On I April 1998, the

seeds were mixed with moist sand and perlite

and refrigerated until plan ting. Catkins of p dellOides were initiaJly collected from at least 20 trees on 3, 8, 9 and 16 June 1998. For a second

seeding (see below). catkins were colleeted on 12 July 1998. Following the proeedure of Friedman er al. (1995), catkins were collected that con­

tained capsules that were partially open. The

catkins were stored in paper bags and allowed

to open further after collection. Within 48 h of

collection, the seeds were cleaned in a sct of soil

sieves by directing air from a vacuum cleaner

through the sieves. Cleaned seeds were refrigerated until needed.

Seed germinabihty was tested in the laboratory

by placing seeds of each species on moistened

filter paper in Petri dishes at room temperature. Because E. angustifolia does not exhibit synchron­ous germination (Shafroth el al., 1995), germina· bility was measured as the cumulative number of germinants after 4 weeks and 6 weeks. Because

P deltoides seeds are not dormant and may germ­

inate immediately upon dispersal, P delloides

© 2001

germinability was measured as the number of

germinants after I week. A seed was considered

to have germinated when both cotyledons had

emerged and were free of the seed coat.

Experiment I: effects of physical disturbance with granlvory

In the first experiment, seeds were applied on 19 June 1998. E. angustifolia seeds were broad­cast by hand onto the subplots. at a density of 133 seeds/m'. P deltoides seeds were sprinkled onto

the subplots at a density of 267 seeds/m2, using plastic eontainers with perforated lids. Due to

an observed lack of E angustifolia establishment

in the subplots by mid-July, an additional 100

E. angusti/olia seeds/m2 were pressed under the

soil surface of each E. angu.wjolia subplot on 24 and 25 July 1998. An additional 250 P deftoides seeds/m2 were sprinkled onto each P deltoides subplot on 27 July 1998. Therefore, overall seeding densities in the subplots were 233 seeds/m2 for

E. angustifolia, and 517 seeds/m' for P delloides. P de/roides and E. angustifolia seedlings in the subplots were counted between 29 September and 2 October 1998. Seedling density distributions in the cleared vs. uncleared subplots were compared

using a non-parametric paired sign test (SPSS Inc., 1990).

Blackwell Science Ltd, Diversify and Distributions, 7. t-14

5 Riparian tree seedling establishment

Experiment 2: effects of physical disturbance without granIvory

In the second experiment, two exclosures were placed at random locations within each E angustifolia subplot (Fig. I). The exclosures were designed to prevent access by all vertebrates, especially small granivorous mammals and birds. The 60 rectan­gular exclosures were constructed of 6 mm mesh metal hardware cloth and measured LO x 10 em square x 35 cm tall, with closed bottoms and lids which could be fastened closed. They were buried to a depth of 10 cm, and reached 25 cm above the ground surface. During exclosure installation, care was taken to minimize disturbance of sur­rounding soil and vegetation. Exclosure locations were adjusted so thal installation did not require removal of any seedlings of the target species established in experiment I. On 8 August 1998 10 E. angus/lfoNa seeds were pressed into the soil surface within each exclosure. On 18 September 1998 all seedlings growing within the exclosures were counted. A paired t-test for two sample means ",;as used to compare seedling density inside the cxclosures in the cleared and uncleared subplots, paired by plot (SPSS Inc., L990). The test was performed on the number (sum) of seedlings in both exclosures per subplot.

Midday ground surface light levels were meas­ured at the exc)osure eorners in all E. angustifolia subplots between 15 and 25 September 1998. The mean exclosure light level was calculated for each subplot, and these values were compared to the mean light levels at random points described above for experiment I. Using plot (n = 15) as a random effect, and disturbanee status (cleared vs. uncleared) and presence of an exclosure (exclos­ure vs. random location) as fixed treatments,. we used analysis of variance to assess the effeets of the two fixed factors on grouud surface light levels (SAS Institute Inc., 1989). To assess whether light levels were significantly different at exclosures vs. random points within each clearing treatment, we conducted a paired I-test on the mean light levels in each E. angusllfolia subplot (SAS Institute Inc., 1989).

Characteristics 0/ granivory We investigated granivory levels in the study plots using seed dish trials. Seed dish stations were set up in the 30 E. anguslr/olia subplots. Each

station consisted of a pair of seed dishes, with one dish placed inside an exclosure (described above), and the other plaeed outside of it. The seed dishes were constructed from l.5 mm mesh plastie screen stretched around 7 cm diameter Ball® jar lid rings. For the P delloides seed trials, a circular piece of weed cloth was plaeed in the bottom of each dish to prevent seeds from falling through the screen. Dishes were pinned to the ground, and a wooden craft stick was bent over the lip of each dish to ensure invertebrate access. Ten seeds were placed in each dish at dusk and left out overnight. At dawn of the next day, overnight granivory rates were recorded and any missing or damaged seeds were replaced. At dusk, the dishes were collected and daytime gran ivory rates were recorded. The E. angustlfolia seed trial was con­ducted 8-9 August 1998, and the P del/oides trial was conducted 9-10 August 1998.

Data analysis followed Hulme (1994) by exam· ining two components of granivory: (1) seed encounter (the percentage of dishes from which at least one seed was exploited), and (2) seed exploitation (the proportion of seeds visibly dam­aged or removed from a dish once it was encoun­tered). In this study, a single value of percent seed encounter was calculated for each treatment combination. Average seed exploitation was then calculated for each treatment.

To assess the kinds of small mammals present in the study area, we used a 5 x 5 grid of Sherman live-traps spaced 3 m apart. Trapping was eon­ducted for four consecutive nights starting 24 August 1998. The traps were provided with polyester bedding, baited with sweet oats, and opened at approximately 17: 00 h in the evening. Traps were checked between 06 : 30 hand 08 : 00 h the following morning, and left closed throughout the day. Captured animals were weighed and identified aceording to species, sex, reproductive condition and maturity. Iudividuals were marked by clipping fur in distinctive patterns and were then released.

RESULTS

Experiment I: effects of physical disturbance with granivory

Germinability of the P delloides seeds was 81% for the batch pi anted on 19 June ]998, and

© 2001 Blackwell SCIence Lld. Diversity and DlSlflbU!lonJ. 7, J-14

6 G. Katz,}. Friedman and S. W Beatty

85% for the batch planted on 27 July 1998. For E. angus/i/olia, 17% of the seeds had germinated after 4 weeks, and 21 % had germinated after

6 weeks. P deltoides seedling establishment occurred

only in cleared subplots. At the end of the growing season, the mean number of P de/toides seedlings in the 15 cleared P delta ides 1 m' su b­plots was 6 (median = 2, standard deviation = 8.5, range of 0-28). In contrast, there were no P deltoides seedlings in any of the 15 subplots in which the pre-existing groundcover had been left intact. The effect of clearing on P della ides seed­ling establishment was signifieant (P < 0.001).

No E. angustlj'olia seedlings were established in either the cleared or the uncleared E. angustifolia subplots. By I August 1998, all E. angustifo/ia seeds visible on the ground surface appeared to have been chewed.

The clearing treatment influenced light levels significantly at the ground surface (F =2527.15, P < 0.0001, dJ. = 1,14, MSE = 22.53). In Septem­ber 1998, mean ground surface light levels were 94.9% of full sunlight in the cleared subplots, and 7.7% in the uncleared subplots. In the cleared subplots, the average ground surface light levels were slightly less than full sun due to minor shading effects of the plot fcnces and vegetation outside of the study plots.

Experiment 2: effects of physical disturbance without granivory

In September 1998, a total of 197 E. angustifo/ia seedlings were found inside the 60 exclosures, an establishment rate of 32%. E. angustljolia seed­hngs occurred in both cleared and uncleared subplots, although physical disturbance had a sig­nificant positive effect on establishment (T = 7.45, P = 0.005, d.f. = 14; Table I). In September 1998, 160 seedlings had established within the exclos­ures in the cleared subplots aud 37 seedlings had established in the uncleared subplots. The mean number of E. angustljo/ia seedlings estab­lished in the exclosures in the cleared subplots was 10.6, compared to 2.5 in the exclosures in the uncleared subplots (Table I). This difference was due primarily to delayed and lower rates of germination in the uncleared subplots, rather than to initial germination and subsequent seed­ling mortality (G. Katz, unpublished data).

Table I E. anguslifolia seedling establishment inside exelosures at the end of the growing season. Data are the sum of seedling densities LO the two exelosures per subplot, n = 15

Subplot

Cleared Uneleared

Mean 10.7' 2.5' Median 10 2 Std deviation 3.3 2.6 Minimum o o Maximum 16 9

Aeeording to a paired I-test for two sample means, values with different superseripts are significantly different (P = 0.005).

Ground surface light levels in the E. angustljofia

subplots were affected by the clearing treat­ment (F= 1186.61, P<O.OOOI, dJ.= 1,14, MSE= 80.85), the presence of an exclosure (F = 9.03, P = 0.0094, d.f. = 1,14, MSE = 29.86) and the interaction of the 'clearing' and 'exc)osure' treatments (F = 12.94, P = 0.0029, d.f. = 1,14, MSE = 59.02). In the cleared subplots, average light levels were slightly lower at the exc)osures

than at random points due to small shading effects of the exc)osures (Table 2). The uncleared subplots exhibited the opposite pattern, with average tight levels slightly higher at the exclosures compared to random points (Table 2). While in the cleared subplots this difference was significant (T = -5.37, P < 0.0001, d.f. = 14), this was not the case iu the uncleared subplots (T = 1.07, P = 0.3042, dL = 14).

Granivory

Granivory rates

In the seed dish trial, granivory rates were very high for seeds of both E. angustijolia and P deltoides outside the exclosures (Table 3). Over­night dish encounter rates were 100% outside of the exclosures for both species in both the cleared and uncleared subplots. In all cases exeept one, all 10 seeds were exploited within each seed dish encountered overnight. Outside the exclos­ures, chewed seed hulls were found at 16 of 30

(12001 Blackwell SCIence Ltd, Diversity and Dis/nb/arolls, 7. 1-14

7 Riparian tree seedling establishment

Table 2 August 1998 mid-day ground surface light levels at exclosures and random points wtthin E. anguslilolla subplots. Values are expressed as percentage of full sunlight measured at the time of data collection, and are based on the means of eight measured points per treatment eombination, n =15

Treatment"

Cleared subplots Uncleared subplots

Random pt Exelosure Random pt Exelosure

Mean 96.1' 84.7 b 9.0' 11.9' Std deviation 2.7 6.8 9.7 9.3

• According to ANOVA, light levels were signifieantly a/rected by the clearing treatment CP < 0.0001), the presence of an exelosure (P =0.0094), and their interaction CP =' 0.0029). Within each clearing treatment, values with different superscripts are signifiean tly different (P < 0.05) according to a paired I·test.

Table] Granivory. Tabulated values are 'seed encounter', percentage of seed dishes with;" 1 seed damaged or removed. Values in parentheses are 'seed exploitation', mean pereentage of seeds damaged or removed within an encountered dish; n =' 15 dishes per treatment

Cleared subplots Uncleared subplots

Exelosure Outside Exclosure Outside

Overnight E. anguslifolia P dellOides

Daytime E. angus/i/olia P de/tardes

°C") 13.3(10)

DC") 26.7(55.0)

100CIOO) 100(100)

0(") 80(96.7)

6.67(10) 0(")

0(·) 0(")

100(100) 100(99.3)

53.3(lOO) 73.3(7l.8)

" Seed exploitation is undefined because no dishes were encountered.

E. ongusli!olio stations and 10 of 30 P deltoides stations. Inside the exclosures, where seeds were

not accessible to vertebrates, very little overnight

granivory occurred.

Daytime gran ivory patterns were more com­

plex (Table 3). In the cleared subplots, there was

no daytime granjvory of E. angustifolia seeds inside

or outside of the exclosures. In the uncleared

subplots, eight of the 15 E. angustifo/ia seed

dishes were encountered. In all eight cases seed

exploitation was 100%, and chewed seed hulls

remained at three dishes. There was some day­

(ime granivory of P deltoides seeds in both the

cleared and uncleared subplots. In the cleared

P deltoides subplots, seed dishes were encoun­

tered inside four exclosnres (Table 3). Outside the

exclosures in the cleared subplots, L2 of the 15

P del/aides seed dishes were encountered. At

two of the dishes outside the exclosures, chewed

seed hulls remained. No daytime P deltoides gran ivory occurred inside exclosures in the

uncleared subplots, but outside of the exclosures

1I of the 15 seed dishes were encountered.

Chewed seed coats remained at three dishes.

Rodent Populo/ion dara During the four nights of live-trapping, 52

different individuals were captured (Table 4);

28 deer mice (Peromyscus maniculatus), 17 house

mice (A1us musculus), four meadow voles (Micro/us pennsy[vanicus) and three prairie voles (l\t.ficrolu~·

ochrogaster). Trap success increased from 17

© 2001 Blackwell SCIence LId, Diversity arid DlSlnb"lIofls, 7, 1-14

8 G. Katz,}. Friedman and S. W Beatty

Table 4 Number of individuals captured (and number of recaptures) in the 5 x 5 Sherman live-trap grid over four nights

Night I Night 2 Night 3 Night 4

M. musculus 7(0) 5(3) 11(4) 8(7) P manicufatus 10(0) 14(6) 13(6) 14(11) M. ochrogosler 0(0) 1(0) 2(1) 1(1) M. pennsylvanicus 0(0) 2(0) 0(0) 3(0)

Tolal captures t7(0) 22(9) 26(11) 26(19)

Individuals the first night, to 26 individuals on the last two nigh ts. Trap success rates over 25 occur when traps capture more than one individual.

DISCUSSION

Effects of physical disturbance

Physical disturbance was necessary for P. dellOides seedling establishment. In the presence of grani­vores, P. del/aides seedlings became established in the cleared subplots, but not in the subplots with intact groundcover. This result is consist­ent with other published studies demonstrat­ing that P del/aides establishment is limited to disturbance patches that are moist and free of competing vegetation (Regehr e/ of., 1975;

Johnson eI al., 1976; Friedman el 01., 1995; Auble & Scott, 1998). The relative importance of supple­mental moisture and disturbance depends on background levels of moisture and light. In arid regions or on dry, sparsely vegetated surfaces, moisture scareity tends to be more limiting than light (Zimmerman, 1969; Cooper el 01., 1999). In humid regions or on wet, densely vegetated sur­faces, light availability tends to be more import­ant (Johnson, 1965). In our study, the presence of competing vegetation and litter prevented P del/aides establishment even though moisture was abundant.

For E. angus/ifolia, no seedlings became estab­lished in either the cleared or uncleared subplots in the presence of granivorcs. The most plausible explanation for this lack of establishment is that intense granivory prevented [he sown seeds from germinating. Granivory therefore prevented the expression of any effect of physical disturbance on E. anguslifolia establishment in experiment I.

© 2001

Although some granivory of sown P. delloides seeds probably also occurred, it did not prevent estabJishmen L

When protected from vertebrate granivores, E. angustifolia seedlings did become established and the effect of physical disturbance could be assessed. The absence of physical disturbance did not prevent E. angustifolia seedling estab­lishment, although seedling numbers were sig­nificantly higher in cleared than in uncleared subplots. Therefore, whereas P. dellOides recruit­ment depends heavily on the presence of physically disturbed patches (Friedman el 01., 1995; Auble & Scott, 1998), this constraint is less severe for E. angustifolia. It could be argued that the slight increase in light due to the presence of exclos­ures (Table 2) explains why E. angustlfolia but not P. del/oides was able to become established in the uncleared subplots. We reject this hypothesis, however, because the effect of exclosures on light levels was small compared to both the effect of the clearing treatment and to the background vanability.

In the only other experimental study compar­ing the seedling establishment requirements of E. angustifolia and P. deltoides, Shafroth el of. (1995) found that while shade decreased growth of P delloides seedlings more than E. anguSllfolia seedlings, it had no effect on seedling survival to the end of the first growing season for either spe­cies. There are several possible reasons for the stronger effect of shade observed in our study. First, Shafroth et af. (1995) used shade cloth to create their shade treatment, and although the rated shading effect of the cloth was similar in magnitude to the shade levels observed in the uncleared subplots here, the shade efrect they ereated was less severe because they did not block lateral indirect light (P. Shafrolh, personal

Blackwell Science Ltd, Diversity and Distribu(ions, 7, l-t4

9 Riparian tree seedling establishment

communication). Secondly, their study was conducted in experimental planters filled with coarse sand, so there was no litter effect limiting P. delloides seedling establishment. Thirdly, it is possible that other faetors associated with intact groundcover, such as the presence of arthropod herbivores (FacelJi, 1994), play an important role in limiting P deltoides seedling establishment. The present study and Shafroth et al. (I995) both suggest that E. angus/i/olia is better able than P delioides to become established under pre­eXisting vegetation and litter, a conclusion also supported by field observations (Christensen,

1963; Olson & Knopf, 1986). The observed contrast in the effect of dis­

turbance on establishment of E. angusli/olia and P deI/aides may be due, in parl, to differenees in seed size. Because E. angusli/olia produces much larger seeds than P de/toldes, it is likely that difTerences in initial seedling provisioning play a role in determining their differing seedling establishment patterns. Large seed size has been associated with the ability of seedlings to grow and survive in the shade (Grime & JefTrey, 1965; Saverimultu & Westoby, 1996), and with the ability of seedlings to establish within intact groundcover (Reader, 1993). Large seed size may be generally advantageous whenever seedlings experience carbon deficits during early develop­ment, such as might occur following temporary shading, flooding, herbivory, pathogen infection or drought (Streng et af., 1989; Armstrong & Westoby, 1993; Saverimuttu & Westoby, 1996).

Seed dormancy and longevity may a!so facilit­ate invasion by E. anguslijolia into sites with

pre-existing vegetation. First, E. angusti/olla may be able 10 develop a soil seed bank, which would allow seedlings to become established whenever conditions at a site were appropriate, regardless of season or year (Shafroth el al., 1995). This mech­anism mighl be especially important in habitats with low levels of sediment erosion and deposi­tion, such as meadows or along regulated river reaches. Seeondly, E. anguSlljoila seedlings may be able to avoid some of the detrimental efTects of shading and drought by becoming established early in the growing season before leaf expansion of herbaceous or woody overstorey plants. Because E. angusl/jolia seeds are dispersed during the autumn and winter, and only require 9-12 weeks under cold moist conditions to break dormancy

(Hogue & LaCroix, (970), many dispersed seeds will probably be germinable by early the following spring. Early germination could be advantage­ous, provided seedlings are capable of tolerating relatively cool early spring temperatures. We have observed radicle extension of E. angustljolia seeds under refrigeration, and autumn-sown seeds com­monly germinate in April in an outdoor nursery in Fort Collins, Colorado, USA (R. Moench, personal communication), suggesting the poten­tial for considerable cold-tolerance of seedlings. However, we know of no studies that have

examined the timing of E. angusli/olia seedling emergence under field conditions.

E. angustljo/ia's large dormant seeds, its relat­ive freedom from reliance on physieal disturb­ance for seedling establishment, and its ability to invade into intact vegetation, contrast strongly with the attributes of Tamarix spp. (saltcedar), the other widespread woody invader of riparian habitats in the western United States. Like native cottonwoods and willows, Tamarix produces large numbers of small, short-lived seeds which are dispersed by wind and water (Warren & Turner, 1975; Brock, 1994). Tamarix germination and seedling establishment generally occur most effectively on the same bare, moist alluvial sub­strates whieh provide suitable safe sites for the native pioneer species (Shafroth et af., (998). Because of the pronounced difTerences in the regeneration niches occupied by E. angllstijolia and Tamal'ix, these species can be expected to invade under different circumstanees.

Effects of granivory

Although seeds of both species were exploited in this study, granivory prevented E angusli/o/ief seedling establishment outside the exclosures, and this was not the case for P de/toides. Several lines of evidence suggest that the high E. angustifolia granivory rate was attributable to rodents, most probably mice. In the seed dish trial, almost no E. angusti/olia seeds were encountered inside exclosures. However, very high eneountcr and exploitation rates were observed outside the exclosures, especially at night. Some daytime granivory of E. angust1jofia seeds oceurred, but only in the uncleared subplots. Taken together, these observations suggest strongly that noc­turnal rodents were primarily responsible for

© 2001 Blackwell Science Ltd, Di'€l'Jlly and DisrribufiOlls, 1. 1-14

lOG. Katz,}. Friedman and S. W Beatty

utilizing the E. anguslijolia seeds in this study. A secondary source of granivory appears to have come from diurnal rodents which were active in the uncleared snbplots where intact vegetation provided shelter. Observations of characteristically chewed E. angusrifolia sced hulls on the ground surface in the subplots and in the seed dishes, the occurrence of rodent faeces in the seed dishes and the holes dug to access E anguslijofia seeds buried in experiment 1 all provide support­ing evidencc for granivory by rodents in the srudy plots. It is unlikely that birds were respons­ible for the granivory because birds do not forage nocturnally, and would not be expected to prefer sceds in the uncleared subplots during the day.

The rodent community composition in this study was typical of ripanan rodent commu­nities in the region. In a study of small mammal populations at 15 riparian and floodplain sites in Boulder County, Colorado, P maniculalus, 1\1' museuills, M. pennsy!vanicus and M. ochrogasler accounted for 62.8%, 14.6%, 13.3% and 3.8% of captures, respeetively (D. Armstrong, personal communication). In Colorado and Idaho, M. musculus, P Il1anicula/Us and AI. achrogasler were more abundant in Russian-olive stands than in native riparian forest (Knopf & Olson, 1984).

All four of the rodent species trapped in this study arc potenlial granivores. However, the two mouse species, P maniculatus and Jyf. musculus, most probably had the greater impact on seed availabiltty. Several lines of evidence support this assenion. First, the voles, M. ochrogaSler and M. pennsylvanicus, are primarily herbivorous, although their diets do include varying propor­tions of supplementary arthropods, seeds and roots (Reich, 1981; Stalling, 1990; Fitzgerald el al., 1994). In contrast, P maniculatus and M. musculus are more commonly omnivorous, with seeds often comprising a large proportion of their intake (Fitzgerald el al., 1994). Because the two mouse species are generalists and are opportunistic in their diets, they would be more likely to con­sume the seeds made available in this study. Secondly, the voles were present in relatively low numbers (seven individuals captured) compared to the mice (45 individuals captured), which may have limited their impact. Thirdly, miee tend to be nocturnal, while voles can be aetive during the day or at night.

© 2001

Seeds of P deltoides appeared to be exploited by a wider array of granivores than were seeds of E. angustifolia. While E. angustifolia seeds were exploited almost exclusively outside exclosures, P deltoides seeds were exploited both inside and outside. Because P delloides seeds are small, it is plausible that invertebrates removed at least some seeds from the dishes. The observation that day­time gran ivory occurred in the cleared sub­plots for P delloides, but not for E. anguslifolia, is further evidence for the existence of some invertebrate P deltoides secd exploitation; inver· tebrates may forage diurnally and do not avoid open areas (Hulme, 1997), while many granivorous rodents tend to be nocturnal (Fitzgerald et al., 1994), or to forage only in sheltered areas during the day. In our experiment, P deltoides grani­vory rates in the seed dish trial may havc been artificially high due to a 'dish effect'. That is, placing the small P deltoides seeds in the seed dishes may have made them more detectable to granivores, and this increase would not have been as great for the larger E. ang!lstifolia seeds (Reader, 1997).

The ecological effects of seed removal depend upon the fate of exploited seeds (Hulme, 1997). For E. anguslijolia interactions with animals are varied, since vertebrates are important secd dis­persers as well as potential seed consumers. In this study, most of the observed E. ongustifolia seed removal probably consisted of local seed consumption, and therefore prevented germina­tion of the exploited seeds. The presence of empty chewed seed hulls in the subplots and in the seed dishes demonstrates that many E. anguslijolia seeds were eonsumed in silu. Although some E. angustifolia seeds may have been cached in either above·ground or subterranean caches (Stalling, 1990; Fitzgerald et aI., 1994), from which germination could possibly occur, the home ranges of the roden t granivores present in the study area are sufficiently small that at least some of these caches would probably have been located within the study plots, where no germination was observed. However, in other situations, drupe exploitation by larger verteb­rates such as birds or omnivorous mammals (e.g. coyotes, Canis lalrans) may result more commonly in dispersal, since whole fruits are consumed and seeds pass intact through the animals' digestive systems.

Blackwell Scienee Ltd, Diyersily and Disrribur!ons, 7, 1-14

Riparian tree seedling establishment I I

CONCLUSION AND IMPLICATIONS

The factors limiting seedling establishment vary spatially and temporally, but for each species a certain suite of factors may provide the crucial limits to regeneration. Because they are prolific seed producers, because their seeds are dispcrsed over a large spatial area, and because their seeds contain little stored energy, P. delLOides seedling estabhshment IS typically limited by the availabil­ity of suitable bare, moist safc sites during the seed dispersal period (Friedman et 01., 1995). In this study, lack of physical disturbance prevented P del/aides seedling cstablishment, but grani­vory did not. In contrast, granivory prevented E.. angus/ijolia seedling cstablishment, but laek of physical disturbance did not. For both spccies, herbivory may further limit seedling survival after germination. Like Andcrsen & Cooper (2000), we observed herbivory of P deltoide,\' seedlings, apparently by voles, in our study plots,

The prevention of E. anguslijolia seedling estab­lishment by granivores in thiS study contradicts the argument that invasivc allen species are suc­cessful, at least in pan, because they are free of predators in their new range (Crawley, 1987). The results reported here suggest that opportun­istic generalist native predators can exert a sig­nificant impact on the spread of an invasive alien species, a finding which is consistcnt with a model proposed by Lonsdale (1993) in which heavy granivory limited the rate of spread of an invasive species. Based on the observations rcported here, it is possible that granivory influences the distribution and rate of spread of E angus/ifolia in western North America

One of the most widespread human impacts on riparian ecosystems in wcstern North America is thc alteration of river and floodplain hydro­logy. Below dams, the natural river flow regime is usually changed, resulting in a variety of effeets to stream and riparian ecosystems (Ligon el 01., 1995; PofT et af., 1997). For riparian Populus, reduced flood disturbance and reduced rates of channel movement have resulted in a lack of regeneration downsrream of dams on many meandering rivers (Reily & Johnson, 1982; Bradley & Smith, 1986; Friedman e/ af., 1998). Compared Lo Populus, E. anguslifolia seedling establish­ment is not reliant on the creation of fluvial disturbance patches. Therefore, E. ongus/ijolia

can invade downstream of dams, despite reduced fluvial disturbance (Shafroth et at., 1995; Lesica & Miles, 1999).

The contrast in the importance of physical dis­turbance for seedling establishment of the two species in this study provides a counter-example to the generalization that biological invasions are facilitated by physical disturbance (Fox & Fox, 1986; Hobbs & Huenneke, 1992). Here, the native woody species, P. del/oides, is more dependent on physical disturbance for establishment than is the alien invader. Our results suggest that altera· tions of natural disturbance regimes, rather than only increased disturbance levels or new kinds of disturbance (Hobbs & Huenneke, 1992), may be important to biologieal invasions. This may be especially true in 'dynamic systems', where native species depend on physical disturbance at crucial life history stages, and invaders may takc advant­age of damped disturbance regimes (Lesica & Miles, 1999).

Andersen & Cooper (2000) and Lesica & \1ilcs ([ 999) have argued that human alterations of nver hydrology may alter herbivore com­munitics, and thereby impact native riparian vegetation indirectly. In both cases, the authors suggested that stabilized flow regimes down­stream of dams might allow for larger or more persistent populations of native herbivores such as beaver (Castor canadensis, Lesica & Miles, 1999) and voles (ll1icrotus montanus, Andersen & Cooper, 2000), which can be detrimental to native floodplain tree establishment or survival. These 'plant-herbivore-hydroperiod' interactions may be an important aspect of the efTects of dams on downstream ripanan forests (Andersen & Cooper, 2000).

Our results suggest that the indirect impacts of altered river flow regimes on riparian for­ests could include plant-granivore-hydroperiod effects. Such efTects will only occur if altered flow regimes influenee downstream granivore popula­tions. At present, it is not clear that any consist­ent pattern exists regarding the effccts of river regulation on small mammal populations, and it is likely that these effects will depend heavily on the hydrological and geomorphic context, as welt as on the specific animal taxa involved (Ellis el al., 1997; Andersen & Cooper, 2000), In cases where population densities of granivorous mammals are increased downstream of dams, increased

© 2001 Blackwell Science Lld, Diversity and DlStf/blltions, 7, 1-14

I 2 G. Katz,). Friedman and S. W Beatty

granivory could slow or limit E. angustl/olia invasion. However, such effects elearly do not prevent the spread of E. angustifolia at the land­scape scale, where widespread invasion of ripar­ian habitats on both regulated and un regulated rivers is occurring.

ACKNOWLEDGMENTS

We thank the City of Boulder Open Space Department for permission'to use the Teller Lake study site, for providing irrigation water, and for conducting the small mammal census. We are gratefUl to Lynn Reidel for her interest and help with (his project, and to Bernadette Pfau and Mark Miller for inval uable field assistance. D. Andersen and P. Shafroth provided con­structive reviews of an earlier manuscript. This research was supported by the United States Envir­onmental Protection Agency STAR Graduate Fellowship Program, and by the Midcontinent Ecological Science Center of the United States Geological Survey.

REFERENCES

Andersen, D.C. & Cooper, D.J. (2000) Plant­herbivore--hydroperiod interactions: effects of native mammals on floodplalO tree rceruitment. Ecological Applicalions 10, 1384-1399.

Armstrong, D.P. & Westoby, M. (1993) Seedlings from large seeds tolerate defoliation belleI': a lest using phylogenetically independent contrasts. Eco­logy 74, 1092-1 toO.

Asquith, N.M., Wright, S.l & Clauss, MJ. (1997) Does mammal community composition control recrui tmen t in neotropical forests? Evidence from Panama. Ecology 78, 941-946.

Auble, G.T. & Seott, M.L. (1998) Fluvial disturb­ance patches a.nd cottonwood recruitment along the upper Missouri River, Montana. Wetlands 18, 546-556.

Baker, WL. (1990) Climatie and hydrologic etTects on the regeneration of Populus ongustifolia James along the Animas River, CoLorado. Journal of Biogeography 17, 59-73.

Belsky, Al, Matzke, A. & Uselman, S. (1999) Survey of livesLOck influences on slream and riparian eco­systems in the western United States. Journal of Soil and Water Conservation 54, 419-431.

Bossard, c.e (1991) The role of habitat disturb­ance, seed predation and ant dispersal on estab­lishment of the exotie shrub eytisus scoparius in California. American Midland Naturalist 126, 1­13.

Braatne, lR., Rood, S.B. & Reilman, PE. (1996) Life history, ecology, and conservation of ripar­ian cottonwoods in North America. Biology of Populus and irs implications for managemenr and conservalton (ed. by R.F. Stettler, R D. Bradshaw, P.E. Heilman Jr and T.M. Hinckley), pp. 57-85. NRC Researeh Press, Ottawa.

Bradley, C.E. & Smith, D.G. (1986) Plains cotton­wood recruitmcnt and survival on a prairie mean­dering river floodplain, Milk River, southern Alberta and northern Montana. Canadian Journal of Botany 64, 1433-1442.

Brock, lH. (1994) Tamarix spp. (salt cedar), an invasive exotic woody plant in arid and semi· arid riparian habitats of wcstern U.S.A. Ecology and management of invasive nverside plants (ed. by L.C de Waal, L.E. Child, P.M. Wade & lH. Brock), pp. 27-44. Wiley, New York.

Brown, lR. & Reske, E.l (1990) Control of a desert-grassland transition by a keystone rodent guild. Science 250, 1705-1707.

Campbell, C.l & Dick-Peddie, WA. (1964) Com­parison of phreatophyte communities on the Rio Grande in New Mexico. Ecology 45, 492-502.

Christensen, 8.M. (1963) Naturalization of Russian olive (Elaeagnus angustifolia L.) in Utah. American Midland Naturalist 70, 133-137.

Cooper, D.l, Merritt, D.M., Andersen, D. & Chimner, R.A. (1999) Factors controlling the establishment of Fremont cottonwood seedlings on the Upper Green River, USA. Reguloted Rivers: Research and Management IS, 419-440.

Crawley, M.l (1987) \\!hat makes a eommunity invas­ible? Colonization, succession, and stability (ed. by M.J. Craw Icy, Pl. Edwards and A.l Gray), pp. 429-453. Blackwell, Oxford.

Currier, PJ. (1982) The floodplain vegetarion of the Platte River: phyrosociology, forest development, and seedling establishment. Botany. Iowa State University, Ames, Iowa.

Davidson, D.W (1993) The effects of herbivory and granivory on terrestrial plant suceession. Oikos 68, 23-35.

de Castro, ED., Aranda, Y. & Schmitz, M.E (1990) Aeetylene-redueing activity and nitrogen inputs in a bJutT of Elaeagnus angustifolia L. Orsis 5, 85-89.

Ellis, L.M., Manuel, l, Molles, C. & Crawford, es. (1997) Short-tcrm elrects of annual flooding on a population of Peromyscus leucopus in a Rio Grande Riparian forest of central New Mexico. American Midland Naturalist 138, 260-267.

Facelli, lM. (1994) MultipLe indireet effects of plant litter atTect the establishment of woody seedlings in old fields. Ecology 75, 1727-1735.

Fenner, P, Brady, WW & Patton, D.R. (1985) Effects of regulated water /lows on regeneration of Fremont cottonwood. Journal of Range Manage­ment 38, 135-138.

© 200} Blackwell Science LId, Di'ersi'y and Diu,iburionJ, 7, J- 14

Riparian tree seedling establishment 13

FItzgerald, J.P., Meaney, e.A. & Armstrong, D.M. (1994) Mammals of Colorado. University of Colorado Press, Niwot, Colorado.

Fox, M.D. & Fox, B.1. (1986) The susceptibIlity of natural communities to invasion. The ecology of biological imasions: an Ausfralian Perspective (ed. by R.B. Groves and J.J. Burdon), pp. 57-66. Australian Academy of Science, Canberra.

Friedman, lM., Osterkamp, WR., Scott, M.L. & Auble, G.T (1998) Downstream effects of dams on channel geometry and bottomland vegetation: regional patterns in the Great Plains. Wetlands 18, 619-633.

Friedman, IM.M.L., Scott & Lewis, WM., Jr (1995) Restoration of riparian forest using irrigation, artificial disturbance, and natural seedfall. Environ­mental Monagement 19, 547-557.

Great Plains Flora Association (1986) Flora of the Greal Plarl);'. University Press of Kansas, Lawrenee, Kansas.

Grime, J P. & Jeffrey, D.W. (1965) Seedling estab­lishment in vertical gradients of sunlight. Journal of Ecology 53, 621-634.

Hansen, N.E. (1901) Ornamentals for South Dakota. South Dakota Agricultural Experiment Station, Brookings, South Dakota.

Hobbs, R 1 & Huenneke, L.F. (1992) Disturbance, diversity, and invasion: implications for conserva­tion. Conservation Biology 6, 324-337.

Hogue, E.l & LaCroix, L.J. (1970) Seed dormancy of Russian-olive (E. anguslifolia L.). Journal of Ihe American Society of Horticultural Science 95, 449-452.

Howe, WH. & Knopf, F.L. (1991) On the imminent decline of Rio Grande cottonwoods in central New Mexico. Southwestern Naturalist 36, 218-224.

Hulme, PE. (1994) Post-dispersal seed predation in grassland: its magnitude and sources of variation. Journal of Ecology 82, 645-652.

Hulme, P.E. (1997) Post-dispersal seed predation and the establishment of vertebrate dispersed plants in Mediterranean scrublands. Oecologia 111,91-98.

Johnson, R.L. (1965) Regenerating cottonwood from natural seedfall. JOr<rnal of Forestry 63, 33-36.

Johnson, we. (1992) Dams and riparian forests: ease stUdy from the upper Missouri river. Rivers 3, 229-242.

Johnson, We., Burgess, R.L. & Keammerer, WR. (1976) Forest overstory vegetation and environ­ment on the Missouri River floodplain in North Dakota. Ecological Monographs 46, 59-84.

Kalljo!a, R., Salo, J., Puhakka, M. & Rajasilta, M. (1991) New site formation and eolomzing vegeta­tion in primary suecession on the western Amazon floodplains. Journal of Ecology 79, 877-901,

Knopf, F.L. & Olson, TT (1984) Naturalization of Russian-olive: implications to Rocky Mountain wildlife. Wildlife Soclely Bulletin 12, 289-298.

Lesiea, P. & Miles, S. (1999) Russian olive invasion into cottonwood forests along a regulated river in north-eentral Montana. Canadian Journal of Botany 77, 1077-1083.

Ligon, F.K., Dictrieh, WE. & Trush, W.J. (1995) Downstream ecological effeets of darns: a geo­morphic perspective. Bl:oscience 45, 183-/92.

Lonsdale, WM. (1993) Rates of spread of an invad­ing species - Mimosa pigra in northern Australia. Journal of Ecology 81, 513-521.

Maboney, lM. & Rood, S.B. (1998) Streamflow requirements for cottonwood seedling recruit­ment - an integrative model. Wel1ands 18, 634­645.

Miller, LM. & Baker, D.D. (1985) The initiation, development, and structurc of root nodules in Elaeagnus angustifolia L. (Elaeagnaceae). Prolo­plasma 128,107-119.

Naiman, R.J. & Rogers, K.H. (1997) Large animals and system-level characteristies of river corridors Bioscience 47, 521-529.

Olson, TE. & Knopf, FL. (1986) Naturalization of Russian-olive in the western United States. Western Journal of Applied Forestry 1, 65-69.

Osterkamp, WR., Fenton, M.M., Gustavson, Te., Hadley, R.F, Holliday, Y.T, Morrison, R.B, & Toy, Tl (1987) Great Plains. Geomorphic Systems of North America (ed. by WL. Graf), pp. 163­120. Geological Society of America, Boulder, Colorado.

Poff, N.L., Allan, J.D., Bain, M.B., Karr, J.R., Prestegaard, KL., Richter, B.D., Sparks, R.E. & Slromberg, Ie. (1997) The natural flow regime: a paradigm for river conservation and restoration. Bioscience 47, 769-784.

Read, R.A. (1958) The Great Plains shellerbelt in 1954. Great Ploins Agric"ltural Coul1cil Publica­tion 16.

Reader, R.1. (1993) Control of seedling emergence by ground eover and seed predation in relation to seed size for some old-field species. Journal of Ecology 81, 169-175.

Reader, R.J. (1997) Potential effects of granivores on old field succession. Canadian Journal of Botany 75, 2224-2227.

Regehr, D.L., Bazzaz, EA. & Boggess, WR. (1975) Photosynthesis, transpiration and leaf conduet­ance of Populus deltoides in relation to flooding and drought. Phalosynrhetica 9, 52-61.

Reich, L.M. (1981) Microtus pennsylvanictls. Mam­ma!J£ln Species 159, 1-8.

Reily, PW & Johnson, We. (1982) The effects of allered hydrologic regime on tree growtb along tbe Missouri River in North Dakota. Canadian Journal of Botany 60, 2410-2423.

Sacehi, e.F. & Price, P.W (1992) The relative roles of abiotic and biotic factors in seedling demography of Arroyo Willow (Salix lasiolepis: Salieaceae). American Journal of Botany 79, 395-405.

© 2001 Blackwell Science Lid, D"ersily and Disrriburions, 7,1-14

14 G. Katz. j. Friedman and S. W Beatty

SAS Institute Inc. (l989) SASISTAT User's Guide, Version 6. SAS Institute Inc., Cary, Ne.

Saverimutlu, T. & Westoby, M. (1996) Seedling lon­gevity under deep shade in relation to seed size. Journal of Ecology 84. 681-689.

Scott, M.L., Auble, G.T & Friedman, J.M. (l997) Flood dependency of cottonwood establishment along the Missouri River, Montana, USA. Eco­logical Applica/ions 7, 677-690.

Shafroth, PB., Auble, G.T & Scott, M.L. (1995) Germination and establishment of native plains cottonwood (Populus del/oides MarShall subsp. Inonilifera) and the exotic Russian-olive (Elaeagnus angustijolia). Conservation Biology 9, 169-175.

Shafroth, PB., Auble, G.T, Stromberg, 1.e. & Patten, D.T (l998) Establishmcnt of woody ripar­ian vegetation in relation to annual patterns of streamflow, Bill Williams River, Arizona. Wetlands 18, 577-590.

SImons, S.B. & Seastedt, TR. (I 999} Decomposition and nitrogen release from foliage of cottonwood (Populi,s deltO/de.I') and Russian-olivc (Eleaegnus angustijolia) in a nparian eco~ystem. Southwestern Na/ur(l/isl 44, 256-260.

Smith,TJ. III (1987) Seed predation in relation to tree dominanee and distribution in mangrove for­ests. Ecology 68, 266-273.

SPSS Inc. (1990) SPSS Base System User's GUIde. SPSS Inc., Chicago, IL.

Stalling, D.T (1990) Microtus ochrogasler. Mammalian Species 355. 1-9.

Steele, O.e., Ewel, K.e. & Goldstein, G. (1999) The importance of propagule predation in a forest

of non-indigenous mangrove trees. Wellands 19, 705-708.

Streng, D.R., Glitzenstein, 1.S. & Harcombe, P.A. (1989) Woody seedling dynamics in an east Texas floodplain forest. Ecological Monographs 59, 177­204.

Stromberg, 1. (1998) DynamiCS of Fremont cotton­wood (Populus fremon/il) and saltcedar (Tamarix chinensis) populations along the San Pedro River, Arizona. Journal of Arid Environments 40, 133­155.

Stromberg, 1.C. (1997) Growth and survivorship of Fremont cottonwood, Goodding willow, and salt cedar seedlings after large floods in central Arizona. Grea/ Basin Naturalist 57, 198-208.

Vitousek, P.M. (1990) Biological invasions and ccosys­tem processes: towards an integration of population biology and ecosystem stUdies. Oikos 57, 7-13.

Warren, D.K. & Turner, R.M. (1975) Saltcedar (Tamarix chinensis) seed production, seedling establishment, and response to inundation. Journal of the Arizona Academy of Science 10, l35-144.

Weber, WA. & Wittman, R.e. (1996) Colorado flora: ea,·tern slope. University Press of Colorado, Niwot.

White, P.S. (1979) Pattern, process, and natural dis­turbance in vegetation. Bo/anical Review 45, 229­299.

Young, 1.A. & Young, e.G. (l992) Seeds of woody plants in North America. Discorides Press, Port­la.nd, Oregon.

Zimmerman, R.e. (1969) Plant ecology of an orid basin, Tres Alamos-Reding/on area southeastern Arizona. US Geological Survey, Washington, De.

© 200t Blackwell Science Ll.d, Diversity and Distributions, 7, 1-14