Great Basin Naturalist

Volume 50 Number 1 Article 6

3-31-1990

Seed production and seedling establishment of a Southwest riparian tree, Arizona walnut (Juglans major)

Juliet C. Stromberg Arizona State University, Tempe

Duncan T. Patten Arizona State University, Tempe

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

Recommended Citation Stromberg, Juliet C. and Patten, Duncan T. (1990) "Seed production and seedling establishment of a Southwest riparian tree, Arizona walnut (Juglans major)," Great Basin Naturalist: Vol. 50 : No. 1 , Article 6. Available at: https://scholarsarchive.byu.edu/gbn/vol50/iss1/6

This Article is brought to you for free and open access by the Western North American Naturalist Publications at BYU ScholarsArchive. It has been accepted for inclusion in Great Basin Naturalist by an authorized editor of BYU ScholarsArchive. For more information, please contact [email protected], [email protected]. Creat ~ln Naturalist 50(!), 1990, pp. 47-56

SEED PRODUCTION AND SEEDLING ESTABLISHMENT OF A SOUTHWEST RIPARIAN TREE, ARIZONA WALNUT (jUGLMo1S MA]OR)

Juliet C. Strombergl and Duncan T. Pattenl

ABSTRACT.-A four-year studyofflve populations has revealed influences on seed production and seedlingestablish­ ment of the Southwest riparian tree Juglans major. Germination is abundant after production of large seed crops (masts), hut masts are produced infrequently. Within years, germination is stimulated by summer rains, enabling seedlings to establish on riparian terraces as well as streambanks. Traits such as capacity for dormancy dUring summer drought allow some seedlings to survive on terraces, but abundant rainfall is essential for high rates ofseedlingsuccess. Ranges ofmoisture toleraJJce vary amongseeds collectedfrom different populations, suggesting that ecotypes mayexist between riparian sites with dissimilar moisture regimes. Population-based differences are associated, in part, with differences in seed size.

Arizona walnut Uuglans major [Torr.] Hel· central Arizona; germination "safe sites" for ler) is a member ofthe "Interior riparian de· walnut may be more restricted in drier ripar· ciduous forest," an assemblage of trees ian sites such as along ephemeral streams. For thatgrow along streams in the Interior South· example, seed burial may become a prerequi­ west (Brown 1982). Walnut sometimes domi· site ofgennination as soil moisture decreases. nates this community (Szaro 1989), but more as is true for certain oaks (Barrett 1931). often it occurs at-relatively low densities and Gernlination and establishment of walnut frequencies. TIlis distribution pattern, while or other riparian species may also vary as a - indicating that establishment occurs infre· function ofgenetic differences hetween popu· quently, does not reveal the stage where re· lations. The existence of riparian ecotypes is generation is limited. Sudworth (1934) sug· not a new idea (e.g., Hook and Stubbs 1967), gested low rates of seed production and high and riparian populations may differ between rates of seed predation by tree sqUirrels as isolated Southwest watersheds with different possible natural causes of infrequent estab· hydrologic characteristics. Other studies in lishment; seedling mortality from cattle graz· this series on walnut reproduction have iden· ing also may playa role (Rucks 1984). tmed differences between populations in flo­ Lack ofsuitable germination and establish· ral ratios and seed weight (Stromberg 1988), . ment sites may limit recruitment, but little is indicating that]. mnjOf' shows either plastic' known regarding the relationship between responses or genetically based responses to these requirements and ]. major. Certain environmental differences within and be· generalizations have been made aboutthedis· tween riparian sites. The objectives of this tribution ofwalnut trees, but the habitat char· study were to: (1) identil)o factors that limit acteristics of trees may differ from those in seed production and seedling recruitment of which the seedlings established. A study that ]. major between sites and between years; specifically addressed riparian seedlings re· (2) determine how gennination requirements vealed that seedlings of]. major, a facultative vary between microsites; and (3) determine riparian species, establish in many microsites' how seedling growthresponse to soil moisture throughout the riparian zone. This is in con· differs between populations. trast to seedlings of obligate riparian trees (Alnus oblongifolia such as Arizona alder METHODS Torr.) that occur only immediately a<;ljacent to the stream (Larkin 1987). Larkin conducted Seed production, gennination, and seed­ berstudy, however, in a wet. montane area of ling survival of]. mnjor were studied through

lCenler for Environmental Studie$, Ari~Qna State University. Tempe. Arizona 85287.i20L

47 48 J. C. STROMBERG AND D. T. PATIEN [Volume 50 field and greenhouse experiments and field buried 2 em deep. Sites were planted with observations. Five sites were selected in cen­ their own seeds (except fur Workman Creek tral Arizona: Hitt Wash, ao ephemeral stream and Rock Creek), and seedling size and sur­ near Prescott National Forest; Rock Creek, an vival were recorded through 1985. Microsite intermittent stream in the Mazatzal Moun­ soil moisture was recorded monthly with a tains; and perennial streams (South Fork dew-point psychrometer, and canopy cover Walnut Creek in Prescott National Forest and was estimated visually (Stromberg 1986). two sites along Workman Creek in the Sierra Influences on germination and seedling sur­ Ancha Mountains) (Stromberg 1988). Eleva­ vival were analyzed through multiple regres­ tions vary from 1,100 m at Rock Creek to sion analysis ofgermination and survival per­ 2,100 m at Aztec Peak centages with average seasonal values for microsite moisture level, canopy cover, and Field Observations herbaceous cover. Twenty trees were selected at each site, Greenhouse Experiments and samples of 20 shoots per tree were se­ Nuts were collected in 1983 from trees at lected for monitoring ofseed production from Rock Creek, Hitt Wash, and Walnut Creek to 1983 to 1986. Average values per tree were test for influence of seed weight and seed calculated for seed production per shoot and source on germination and seedling growth. for flower production, flower abortion, seed Nuts were sown in the greenhouse in loam soil weight, and seed viability (Stromberg 1988). in polyethylene-lined pots under four water­ A mast (large seed crop) is operationally de­ ing regimes. One regime was watered every fined in this study as a crop 33% larger than other day to maintain saturated conditions. average for the study. Others were watered at less-frequent inter­ Demography of natural seedlings was stud­ vals, producing average soil moisture con­ ied to determine when, where, and how many tents by dry weight of 80%, 40%, and 20%. seeds germinate, and to compare survival Seeds were sown on the surface and buried rates between rnicrosites. One walnut tree 2 em deep. Seven nuts from each of24 parent near the stream and one on the adjacent trees were sown per treatment. Germination terrace with nearby seedlings were selected percent and speed and seedling survival were at each site. Each marked the center of an monitored. Eight weeks after emergence, 8-m-radius circular plot. All seedlings in the plants were harvestedfor measurement ofdry plots were tagged and scored monthly from weight, stem height, and root length. At each May to October of 1983, 1984, and 1985 for soil moisture level, regression analysis was stem height, leaf length and number, and used to test for inlluence of seed weight, a mortality. Twenty seedlings in a plot at Hitt continuous variable. Duncan's multiple range Wash were excavated to measure depth ofthe test was used to test for effects ofseed source nut in the soil profile. on germination and seedling growth rates. Field Experiments RESULTS Seeds were sown in field exclosures in fall 1982 and 1983 to test for influence ofmicrosite Seed Production (soil moisture and canopy cover) and burial Seed production differed substantially be­ depth on germination and survival. Exclo­ tween sites and years. All sites had large an­ sures protected seedlings from trampling and nual variation in seed production, with annual predation. Four microsites (streamside, open coefficients of variation ranging from 100 to canopy; terrace, open canopy; streamside, 140 among sites. Masts were produced in one closed canopy; terrace, closed canopy) were or two of the four years among sites (Fig. 1). selected per site, with three replicate exclo­ Large crops of seeds, irrespective ofviability, sures per microsite. Exclosures were con­ were produced more frequently. Some excep­ structed from 1.3-cm hardware cloth, and tional trees produced four consecutive, large were 0.6 X 0.6 X 1 m high, 'vith 15 cm of seed crops. Number oflarge flower crops also mesh below ground. Seeds were planted varied substantially between sites, ranging 36 per exclosure in 6 rows in 3 planting from one at Walnut Creek to three at Hilt treatments: surface sown, partially buried, or Wash. 1990] R,PARIAN SEED PRODUCTION 49

b 2 0 ::r: C/) a::: 1.6 w a.. C/) a::: 1.2 w 3: 0 ...J 0.8 lL. w ...J ~ 0.4 w lL. 0 0.6

bo 0.5 ::r: C/) a::: 0.4 w 0... ~ 0.3 w w C/) 0.2 w ...J ~ > 0.1

o 1 234 1 234 1 234 1 234 1 234 A K W H R

Fig, 1. Site means for female (lowers and viable nuts prnclut:ed per shoot for ]uglans major from 1983 through 1986 at Aztec Peak (A), Workman Creek (K), 'Walnut Creek (W), Hitt Wash {H}, and Rock Creek (H). 50 J. C. STROMBERG AND D. T. PATTEN [Volume 50

TABLl~ L Average rates of seed. production (expressed as viable seeds produced per 100 shoots) and abundance of Juglan.s major seedlings per 200 m- at five sites in Arizona.

Seeds Seedlings Seeds Seedlings Seeds Seedlings 1983 1984 1984 1985 1985 1986 Aztec Peak 11 o o o 2 o Workman Creek 9 2 3 1 3 o Walnut Creek 38 52 o o 14 23 Hitt Wash 46 207 10 48 6 24 Rock Creek 10 3 10 1 2 I

Masts were produced at all sites in 1983, they were produced. The mast year of 1983 with abundant viable nuts maturing on trees was followed by abundant seedlings in 1984, (Fig. 1). Seed productionpatterns in following whereas the largely inviable seeds produced years differed among sites. Seed crops were during dry 1984 resulted in low or no recruit­ low in different years among sites, and crops ment in 1985. Rainfall in the year of germi­ failed at different reproductive stages. The nation may also have increased seedling most common causes of crop failure were re­ numbers. This is suggested by the greater ductions in numbers of female flowers and abundance of seedlings in wet 1983 than in production of inviable, low-weight seeds; dry 1984 (e.g., 132 vs. 52 at Walnut Creek), high rates of abortion were a less frequent despite production ofmasts in 1982 and 1983. cause. At Walnut Creek, for example, viable Within years, seeds germinated during wet nut production was lowest in 1984, a result of seasons (Table 2). Most germinated during few flowers and inviable seeds (3% viable). the late summer (August-September) rains. Hitt Wash trees produced a large crop of vi­ Seeds germinated infrequently in May and able seeds only in 1983; seed production was only during weI springs or in wet streambank limited at sequentially earlier reproductive microsites; none germinated during the July stages from 1984 to 1986. In 1984 many flow­ dry period. ers were produced and matured, but seed Germination increased with microsite soil viabilities were low (28% viable). In 1985 moisture, although moisture did not exclu­ flowers were again abundant, but many (96%) sively regulate germination (I" = .24, df = 14, were ahorted. In 1986 Hitt Wash finally had a P < .04; values for exclosures). Few seeds sharp decline in female flowering, resulting in germinated on open terraces, particularly at a low crop size. Rock Creek trees showed a the low-elevation sites (Rock Creek, Hilt third pattern. Following the mast of 1983, Wash) (Table 3) where soils were drier (Table moderate numbers offlowers and viable seeds 4). Canopy cover was positively associated were produced in 1984. This was followed by witb germination percentages within these large declines in flower number in 1985. Mod­ drier microsites, and dense herbaceous cover erate numbers ofviable seeds were produced was negatively associated with germination in 1986. rates. Together. soil moisture, canopy cover, and herbaceous cover explained 45% of the Field Germination variation in germination percentages between Natural seedling abundance differed sub­ microsites. stantially between sites and years. These dif­ Few surface-sown nuts germinated in any ferences were related, in part, to seed produc­ microsite (Table 3). Burial increased germina­ tion rates. Seedlings were abundant only at tion rates within all microsites at all sites sites that produced many seeds (Hitt Wash except Workman Creek, where burial in satu­ and Walnut Creek) (Table 1). Few seed­ rated soil decreased germination. Partially lings were present at Rock Creek, despite huried seeds germinated in moderate num­ moderate seed production, because of seed bers in streambanks and under canopy; com­ predation by Arizona gray squirrels (Sciuros plete burial was necessary for germination in arizotu!l1sis; see Stromberg 1988). Annual open terraces. Lack ofburial limited germina­ seedling abundance was influenced by size of tion ofnatural seed populations. For example, the prior year's crop of viable seeds, since many seeds were ungerminated on the soil seeds generally germinated the year after surface at Hilt Wash terrace. Excavation of 1990] RIPARIAN SEED PRODUCfION 51

TABLE 2. Jugums major seedling recruitment by month (May through October) in streamside and terrace plots at 2 sites in Arizona. Plots are 200 m .

1983 1984 1985 Site Plo... M JJ A S 0 M J J A S 0 M J J A S 0 Workman Terrace 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Creek Streamside 0 0 0 2 00 000100 00000 0 WaJnut Terrace 0 0 0 53 44 0 0 0 0 18 12 0 0 0 0 0 0 0 Creek Streamside 0 0 0 22 13 0 0 0 0 14 8 0 0 0 0 0 0 0 Hitt Wash Terrace 6 3 0 92 43 2 0 0 o 105 48 1 0 0 0 21 7 0 Streamside 14 5 0 3-5 13 0 0 0 0 47 25 1 2 0 0 13 5 0 RockC,eek Temu..'C 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Streamside 0 0 0 3 0 0 0 0 0 1 0 0 0 0 0 1 0 0

TABLE 3. Gennination percentages ofJuglans1l1ajor sown at three planting depths in four microsites. Average seed viability "vas 60%. Values are means and standard deviations for 3 groups of12 nuts. Plantingdepthsare: buried at 2.cm, partially buded, and surface SO",1l. Sites are: Aztec Peak (A), \Vorkman Creek (K), Walnut Creek (W), Hitt Wash (H). and Rock Creek (R). NA = not avaiJable.

Streambank, canopy Strcambank, open Site Buried Partiall)' buried. Surface Buried Partially buried SUrfdCC A 0 [0) 0 (OJ 0 [0] 0 [0) 0 [OJ 0 (0] K 4 [6) 25 [OJ 8 [OJ NA NA NA W NA NA NA 29 [6) 13 [6J 8 [0] H 45 (30) 3 [5J 0 (OJ 46 [6) 0 (OJ 0 (0] '-1 R 50 [18J 5 l'" 0 (OJ 50 (15] 0 (OJ 0 [OJ

Terrace, canopy Terrace, open Site Buried Partially buried Surface Buried Partially buried Surface A 0 [OJ 0 [OJ 0 [OJ 0 [OJ 0 [OJ 0 [0] K 21 (6J 25 [OJ 8 (0] 50 (33J 0 10J 0 [OJ w :J.5 (19J 16 [IlJ 3 (5) 28 (16J 0 [OJ 0 (OJ H 53 [5J 25 [l4J 8 (9J 0 (OJ 0 [OJ 0 (0) R 21 (12J 0 [OJ 0 (0] 4 (8J 0 (OJ 0 (0)

TABLE 4. Soil water potential at30em (-MPa) for terrace (T) and streamside (5) open canopy microsites during May, Jul}', and September of1983, 1984, and 198.5. Sites are: Aztec Peak (A), Workman Creek (K), Walnut Creek (W), Hitt Wash (H), and Rock C,.ek (R).

1983 1984 1985 Site Microsite May Jul Sep May Jul Sep May J'" Sep Mean [± SDJ A T 0.0-2 0.15 0.01 0.68 0.83 0.06 0.11 0.79 0.06 0.30 [O.33J 5 0.01 0.16 0.00 0.54 0.63 0.03 0.06 0.54 0.03 0.22 10.25] T 0.03 0.19 0.01 1.03 1.36 0.06 0.24 1.16 0.06 0.46 [0.52J " S 0.01 0.17 0.00 0.78 0.96 0.02 0.12 0.93 0.04 0.34 (O.40J w T 0.03 0.24 0.02 1.18 1.26 0.05 0.42 1.33 0.04 0.51 [0.54] S 0.02 0.18 0.00 0.84 0.96 0.02 0.18 1.02 0.02 0.36 (O.42J H T 0.03 0.32 0.01 1.65 1.84 0.15 0.48 1.80 0.12 0.71 [0.76J S 0.03 0.22 0.02 1.45 1.63 0.05 0.26 1.58 0.06 0.59 [0.69] R T 0.04 0.43 0.20 1.86 2.04 0.13 0.64 1.83 0.14 0.81 (O.80J S 0.03 0.32 0.02 1.45 1.63 0.05 0.26 1.58 0.06 0.80 [O.68J Mean 0.03 0.26 0.04 1.27 1.44 0.07 0.32 1.39 0.07 + SO 0.01 0.09 0.06 0.38 0.39 0.04 0.16 0.3-5 0.04 52 J. C. STROMBICRC AND D. T. PATIEN [Volume 50

TABl.E 5. Gc::mination percentages forJuglallS major bu,ried in soil OT sown on lhe surf.lCe at three soil water contents (by weight) in the W"ccnhouse. No seeds ~enninated at 20% soil moisture. Seeds arc from streambanks (5) or terraces mfrom Walnut Creek (W), Hitt Wash (H), or Rock Creek (R). Values are means and standard deviations.

Satm3tecl 80% moisture 40% moisture Seed source Buried Surface Buried Surface Buried Surface VI' 8 21 [21] 50 [7] 50 [7] 0 [OJ 7 [7] 0 [OJ T 0 [OJ 50 [29J 42 [181 0 10) 21 (21) 0 [OJ H s ,- [6) 28 [31J 6..1 [291 0 [OJ 24 [3.1j 0 [OJ T 0 [OJ 18 [12J 61 [16J 0 [0) 19 [27] 0 [OJ It s 0 [0] 22 [9] :39 [II] 0 [0] l3 (0) 0 [0] ,T 0 [0] 14 [7] 48 [15] 0 [0] 16 [5] 0 [OJ

seedlings revealed that only 5% of the seed­ example, seeds weighing3 g genninated in 16 lings were from surface nuts, whereas 75% days, compared with 38 days for those weigh- . ~ were from partially buried nuts and 20% were mg I g. buried at depths of 9 to 15 em in pocket gopher (Tlwmomys spp.) caches. Seedling Survival in the Field Seedling mortality had a major role in limit­ Greenhouse Gernlination ing regeneration of]. major. Only 1 of 374 Juglans major germinated abundantly in natural 1983 seedlings among the study sites soil with a moisture content of 80-100% by was alive as of f.'I11 1985. Mortality was high in weight{Table5). Few seeds germinated in the the year of germination and in the two years dry soil or saturated soil. Planting depth influ­ following (Fig. 2). Most seedlings died in enced germination at all soil moisture con­ June, typically a dry month, or in winter. tents. In the saturated soil, surface·sowll Precipitation was sparse in early 1984 (Table seeds germinated in substantially higher per­ 6), and seedlings that germinated in J.983 had centages than buried seeds. In contrast, seeds high winter mortality; only 7% of Hitt Wash required burial for germination at all other and 9% of Walnut Creek 1983 natural fall co­ moisture contents. horts survived until spring 1984. Many that Gennination rates in saturated soil varied did survive remained dormant througb the between populations and with seed weight. spring and summer dl'Ooght of 1984, with­ Seeds collected fi'om trees along the banks of holding leaf-out until the late summer rains, perennial Walnut Creekgemlinated in higher In contrast, more than 20% of 1984 cohorts at percentages than seeds fi'om sites with hoth sites survived to spring of 1985. ephemeral (Hit! Wash) or intermittent (Rock Seedling survivorship varied between Creek) stream !low (Table 5). Many cohorts microsites (Table 7) as well as between years. from these latter sites did not germinate in A large part of the variation between mi­ saturated soil. Cermination percentage in sat­ crosites was attributable to soil moisture. Sur­ urated soil also tended to increase as seed vivorship after two years increased with soil weight (g) per cohort decreased (y = 31.8 ._- moisture between exclosures (1'2 = .31, df = 2 5.3x, r ~ .15, df = 22, P < .05). Seed weight 14, P < .05), with seedlings on open strcam­ was not related to germination in dry or moist banks having highest survival. Seedlings that soil. did survive in dry microsites had high mor­ Speed ofgermination varied with soil mois­ tality ofbuds and shoots during their first and ture level and with seed weight. Seeds in second winters, and grew slowly. On terraces, moist soil (80-100%) germinated rapidly (16 42% of year-old survivors originated new ± 10 days), whereas seeds in drier soil (40% spring growth from basal buds, 49% from lat­ by weight) germinated 2-8 ± 12 days after eral huds, and only 9% from the terminal bud. planting. Although variance in germination On streambanks, in contrast, 15% regrew speed wa.s high within a cohort, median ger­ from basal buds, 58% from lateral, and 31 % mination speed increased significantly with from terminal buds. Second-year seedlings on seed weight for a cohort (y = -0.5 + 5.5x, r' the open terrace thatregenerated from a basal = .34, df ~ 22, P < .01, in 80% moisture). For bud had average stem height of2.5 em, with 1990] RIPARIAN SEED PRODUCTION 53

6 a HIIT WASH + WALNlJT CREEK ...... 5 c '-"- ~ 4 C/) Z w a 3 <.:) z - 2 a-l w w C/) 1

A SON D J F MA M J J A SON D J F MA MJ J A S 0 1983 1984 1985 Fig. 2. Seedling survivorship for Juglans major that germinated in fall 1983 at Hitt Wash (HW) and Walnut Creek (We). Values are log., ofseedlings remaining each month.

TABLE 6. Precipitation (em) from 1982 to 1986 at a tributed to seedling mortality. Interestingly, climatic station near the Walnut Creek study site. Aver­ seedlings did not have greatest herbivory at age annual precipitation is 39 em. sites where adult herbivory was high (see 1982 1983 1984 1985 1986 Renaud 1986). Rather, seedlings in specific microsites had greatest leaf loss from herhi­ January-June 21 22 3 14 16 July-December 19 31 37 31 30 vores. Seedlings on terraces, under canopy, Total 40 53 40 44 46 had greatest herbivory damage; leafarea con­ sumed by October 1983 was 33% ± 15, com­ pared to < 9% for all other microsites. as few as two leaves 1 cm long. These seed­ Effects of cattle grazing were included in lings did not markedly increase in size from the study out of necessity because of the al­ 1983 to 1985. Seedlings in moist, open areas most ubiquitous presence of cattle In South­ grew rapidly, as evidenced by a two-year west riparian areas. Two sites, Hitt Wash and streambank seedling in full sun that reached Walnut Creek, had heavy to moderate cattle 44 cm tall. grazing. Seedlings in exclosures at both sites Flooding killed some streambank seedlings had substantially higher survival rates than but did not impact terrace seedlings. Fall those in simil~r unprotected areas; however) floods killed 10% of Walnut Creek and 5% of this was also tiue for the ungrazed sites (Table Hitt Wash 1983 streambank seedlings; mor­ 7). Adverse impacts ofcattle on seedlings in­ tality rates from winter/springfloods were not cluded trampling and grazing. At Hitt Wash, quantified. Physical impacts of floods in­ 22% of 213 natural seedlings had broken or cluded stem breakage, coverage with debris, eaten stems, as did 13% of 130 seedlings at and scouring of seedlings. Several seedlings Walnut Creek. Ability ofseedlings to recover resprouted after stem topping. from trampling and grazing varied between Herbivory damage from insects also con- sites; 40% of all stem-damaged seedlings at 54 J. C. STIlOMBERC AND D. T. PAITEN [Volume 50

TABLE 7. Survival percentages one, two, and four yean; after ~crmination for /uglans major seedlings in cxclosures and natural areas in four mierosites.

Exclosures Natural areas Microsite 1 yr 2yr 4yr 1yr 2yr 4 yr Stream, open 51 28 16 0 0 0 Stream, canopy 22 15 0 7 0 0 Terrace. open 6 0 0 5 0 0 Terrace, canopy 14 0 0 3 0 0

TABLE 8. Dry weight, root length, shoot height, root:shoot weight ratio, and mortality of eight~weekJttgulU$major seedlings in the greenhouse in soil at three moisture contents. Asterisk (*) indicates significant difference between Hitt Wush (H) and Walnut Creek (W). Values arc means and standard deviations.

Moisture Seed Dry weight 'Root Shoot RS Mortalit.y treatment source (g) (em) (em) ratio (%) 40% Ii 0.35 [0.05] 25 [21 9 (2) 1.4 [0.31 0 [OJ w 0..14 [0.051 29 [31 II (2] 1.1 (0.3] 0 [01 H 0.83 [0.14] 34 [71 16 [2J 0.9 [0.2] 0 [OJ w 0.76 [O.12J 34 [6J 15 [2J 0.9 [02] 0 [OJ Saturated Ii 0.12 [0.02J' 4 [2J' 7 [2J' 0..1 [0.11' 80 (221' w 0.27 [O.04J 13 [3J 10 [3] 0.7 [0.21 13 [61

Hitt Wash regenerated a new stem during the ing abundance ofseedlings between years and same year ofbreakage, whereas none did so at sites. TIle extent offluctuation in annual see<.l Walnut Creek production byJ. major is similar to values for other mast-cropping trees (Silvertown 1980), Greenhouse Seedling Survival and frequency ofmast production is similar to All seedling cohorts had greatest growth in other ]uglandaceae (Nixon et aJ. 1980, Sork intermediate soil moistures and poorest in sat­ 1983, Waller 1979). Seedlings were ahundant urated soil (Table 8). Hoot growth in particular only after mast years, substantiating the view was low in saturated soil, and root-to~shoot that infrequent production ofviahle crops lim­ ratios were low compared to drier soils. Some its regeneration (Sudworth 1934). Rainfall, cohorts, however, grew better than others in which varies considerably between years in saturated soil. Similar to results for germina­ the Southwest, appears to have an important tion rates, seedling growth and survival in influence on malit production. The evidence saturated soil were related to seed weight and for this, although based on a limited number secd source. Size and weight of seedlings in of years of observation, comes from associa­ saturated soil increased significantly with de­ tions detected between rainfall and the repro­ creasing seed weight among cohorts (e. g., ductivestages thatare critical to production of seed weight in g = 26.5 - 4.5 * root length in successful masts-flower pmduction, associ­ cm, r" ~ .49, df= ll, P < .01). With regard to ated with abundant prior and present year seed source, cOhorts from the perennially rainfall, and seed weight, which increases saturated streambanks of Walnut Creek had with abundant spring rainfall (Stromberg greater root development and survivorship 1988). Seed numher between sites is limited in saturated soil than did cohorts from drier variously by low soil moisture and high pre­ HittWash. dispersal seed predation (e.g., Hock Creek) and insect herbivory (e.g., Workman Creek) DISCUSSION (Stromberg 1988). Low rates ofg;crmination also limit recruit­ Although a survey offive populations is not ment to some degree. Whereas moisture representative of a species as a whole, this for germination and establishment of some study has highlighted factors influencing re­ obligate ripa,;an tTees is provided by stream cruitment of J. majo,·. Low and fluctuating flow (Fenner et aI. 1985), these processes availability ofseeds plays a large role in limit- in ]. major, and perhaps other facultative 1990] RJPARI.W SEED PROnUCnO"i 55 riparian b'ces, are influenced in large part hy and seedlings as a result ofecotypic differen­ rainfall. Recent documentation offuirly high tiation in morphology or physiology (Hook regeneration rates for walnut in the South­ and Stubbs 1967), and differences in tolerance west (Larkin 1987, Medina [986) may be a of flooding and saturated soil are common result of long-term moisture cycles, Arizona among trees that grow on sites with different being in an abovt~-average cycle at the time of flooding histories (MeGee et al. 1981). Isola­ tbese studies. tion of Southwest riparian populations within Although moist areas provided optimum "'"mountai.n islands" with distinct moisture safe sites lor germination and seedling estah­ regilnes may have led to development of Iishment, some]. major seedlings established physiological and Illorphological ecotypes on terraces and along ephemeral streams. (Little 1950, Thornber 1915). Seedlings v.;ere somewhat drought-tolerant-, Whcreas small seed size and tolerance of as indicated by high survival in greenbouse saturated soil are associated with wet riparian drought conditions, high root:shoot ratios in sites, the large seeds produced by walnuts on drier conditions, and ahility to survive sum­ drier sites (Stromberg 1988) may increase sur­ mer drought via dormancy, Nevertheless, vival of seedlings stressed by fa<:tors such as seedling numbers in dr,ier riparian sites were drought or grazing. Although purely specula­ low. Recnoitment may be abundant only after tive, the greater ability ofHitt 'iVash seedlings a sequence ofseveral wet years-two for pro­ to recover from trampling and grazing com­ duction ofa large viahle seed crop, another for pared with Walnut Creek seedlings Illay have abundant germination, and onc or two more been a t.'onsequence oflarger seed size. Large for high survivorship. The low frequency of seeds have large cotyledons that remain at­ such a sequence may explain the uniform age tached to seedli.llgs lor up to a year after ger­ structure ofadult walnut populations at some mination (Stromberg 1988), allowing young low-elevation sites (Strnmberg 1988). seedlings to regenerate stems (Wetzstein et An additional factor that may limit estab­ al. 1983). In any case, dilferences in seedling lishment of seedlings in dry sites is lack nf responses between walnut populations high­ hurial. Processes that bury seeds include light the need for study ofmany populatinns to deposition of flood debris (rare on terraces), thoroughly understand reproductive dynam­ caelling by pocket gophers (rare except in ics ofany riparian species. sandy soils), trampling by large animals, and possibly caching by squirrels. Although tree ACK 'OWL~UCMENTS squirrels commonly cache walnuts (Stapanian and Smith 1978), there is conflicting evidence This study was supported iu part by a coop­ about the frequency at which they cache nuts erative agreement with the U.S. Forest Ser­ of]. major. In parts of thcir range where vice (RM-80-133-CA). winters arC mild, sqUirrels immediately con­ sume gathered nuts (Brown 1984). This be­ LITERATURE CITED havior may contribute to the decline in ahun­ dance ofwalnut at low elevations. BAIIHE"IT. 1...1.1931. InOllenccofforest litter on the germi­ TIlis study suggests that gennination and nation and early sllrvival ofchestnut oak, Ouercus establishment requirements of]. major differ 1n000tana Willd. E(:o]ogy 12: 476-484. BROWN, D. E.. EO. 1982. Bioticcommunities ofthe Ameri­ between populations, as well as between can Soutlnvcst-United States and Mexico. Des­ microsites. Specifically, populations from pe­ ert Plants 4: 1-342. rennial stream sites appear to be more toler­ __. 1984. Arizona's trce squirrels. Ariwna Came and ant of saturated soil at the seed and seedling Fish Department, Phoenix. 114 pp. FE~NER. stages; this should be vcrifled on a larger sam­ Po, W. \V. BI\ADY. AND D. R. PATION. 1985. gtJecls ofregulated water !lows on regenemtion of ple of populations. The g.reater germination Fremont cottonwood. Journal of Range Manage­ and seedling survival in saturated soil h)r ment 38: 135···138. seeds from such sites may be a consequenee of [l(X)J(. DD.. A~"D R. M. CHAWFOllD, EDS. 197!:l. Plant life lower oxygen demands of their smaller seeds in anaerobic euvimnmcnts. Ann Arbor Science (Sb'omberg 1988) or of physiological adapta­ Publishers, Ann Arbor, Michigan. 564 pp. HooK, D. D., i\I\D J. STUBBS. 1967. Physiographic seed tions (Hook aod Crawford 1978). Tolerance of source \ollriation in tupelo Ktlms grown in various moisture level is knowa to vary among seeds water reW-mes. Pages 61-67 in Proc't:cdings of 56 J. C. STROMBERG AND D. T. PA:n'EN [Volume 50

the Ninth Southern Conference on Forest Tree SlJ-V"RRTOWN.J. W.1980. The evolutiol1ary ecology ofmast Improvement, Knuxville, Tennessee. Committee seeding in trees. Linnean Society Biological Jour­ on Southern Forest Tree Improvement, Macon, nal 14: 235-250. Ceor~a. SORK. V. L. 1983, Mast fmiting in hiCKories and availa­ LARKIN, G. J J987. Factors influencing distribution and bility of nuts. American Midland Naturalist 109: regeneration of riparian species along mountain 81-88. stream" in central AriY-OM. Unpublished thesi.~, STAP,\NIAN. M, A.• AND C. C. SMlTIt. 1978. Amodel for seed Arizona State University, Tempe. 84. pp. scatterhoarding; coevolution of fox squirrels and Lm'LI~, E. 1..., JR. 1950. Southwestern trees: a g:uide to the black walnut.. Ecology 59: 884-896. nativ(~ spl:lcies of New Mexico and Arizona. Agri­ S11{(')MBI::JIC. J. C. 1988. Re-production and establishment cultu,....1 Handbook 9. United States Department of Arizona walnut (Juglans majur). Unpublished ofAgriculture, Washington. D.C. dissertation. Arizona State University, Tempe. MecKE, A. B.• M. It SCHMIERHACIi. AND F. A. BAZZAZ. 1&1 pp. 1981. Photosynthesis and growth ill populations SUDworrm. G. B. 1934. Poplars. principal cree willows, of PO'pulus deltoides from contrasting habitats. and waluut,s ofthe Rocky Mountains region. Tech~ American Midland Naturalist 105: 30.5-311. nical Bunetin 420. United States Deparbnent of M~DlNA. A. L. 1986. Riparian plant communities of the Agriculture, Washington, D.C. Fort Bayard watershed in southwest New Mexico. SZARO. R. C. 1989. Il..iparian forest and s(;rubland commu­ Southwestern Naturalist 31: 345-359. nity types of Arb..ol\ft and New Mexico. Desert NIXON, G M., M. W. MCCLAIN, AND L. P. HANSEN. 1980. Plants 9: 1-138. Six years of hickory seed yields in sOl1theastern Ohio. THORNBER, 1- J. 1915. Walnut culture in Arizona. Univer~ Journal ofWi.ldlife Management 44: 534-539. sHy of Arizona Agricultural Experiment Station H'~NAUU, O. 1986. The impact ofherbivory on the repro­ llulletin 76: 469-503. ductivl~ wal~ and defense allocatioll ofthe Arizona WALwn, D, M. 1979. Models of mast fruiting in trees. nuL Unpublished thesis, Arizona State Univel'· Journal ofTheoretical Biology 80: 223-232. sity, Tempe. 54 pp. W~I·l.';;TF.IN. H. Y.. D. SPAJIJ(S, Ai"lD C. A. LANe. 1983, HUCKS. M. C, 1984. Compositiun and trend of riparian Cotyledon detachment and growth ofpecan seed~ vegetation 011 five perennial strea,ms in southeast­ lings. HortScience 18: 331-333. ern Arizona. Pages 97-107 in It E. Wamer and K. M. Hendrix, cds., California ripnrian systems: ecology, conservation, and pl'Oduetivc manage­ Received 30]Utll! 1989 ment. Univenity ofCalifornia Press, Berkeley. Accepted 1]anum"!! .1990