Although Illinois was not part of the multilocation effort, our goal was to contribute to this study using the limited number of grafted varieties available on the market in 1998.

Information is limited on the early developmental aspects of known varieties in a field setting. Seedlings are sensitive to ultraviolet light (Peterson, 1991), yet pawpaws grow and set fruit best in full sunlight (Willson and Schemske, 1980). Cultural studies have focused primarily on propagation, including seedling production and grafting techniques (Callaway, 1992, 1993; Layne, 1996; Pomper et al., 2002a, 2002b, 2002c). In recent years, PPF distributed seed for rootstock purposes to commercial nurseries that was harvested from open-pollinated half-sib trees in rows 6 and 7 at the PPF orchard located at the Western Maryland Research and Education Center, Keedysville, Md. The same seedling-source rootstock was also used in the 12 PRVT plantings (Pomper et al., 1999). The primary goal of this study was to compare the early vegetative and reproductive characteristics of varieties grafted onto seedling rootstock with similar genetic background, and to demonstrate the potential of pawpaw as an alternative and sustainable fruit crop for small-scale orchards for southwestern Illinois.

Site preparation for this study began in Spring 1997, and data were recorded through Spring 2001. The site was located near Belleville, Ill., in USDA Hardiness Zone 6A (lat. 38circ28'N, long. 90circ2'W) and the soil was a 113A Oconee silt loam, (fine, montmorillonitic, mesic Udollic Ochraqualfs). Soil nutrient and physical parameters were measured 15 Apr. 1997. Soil test results were pH-5.9; phosphorus 30.3 kg·ha-1 (27 lb/acre); potassium 214.1 kg·ha-1 (191 lb/acre); organic matter 2.2%; cation exchange capacity 16.6 meq; all micronutrients in standard ranges. Particle size analysis indicated soil texture was 17.5% sand, 77.5% silt, and 5% clay. On 30 Apr. 1997, 280.2 kg·ha-1 (250 lb/acre) 18–46–0 (18N–20.1P–0K) and 224.2 kg·ha-1 (200 lb/acre) 0–0–60 (0N–0P–49.8K) were broadcast. The site was then chisel plowed in two directions and disced twice. Ladino clover (Trifolium repens) was sown at 4.5 kg·ha-1 (4 lb/acre), rolled, and allowed to grow through 1997. Final preparation for planting began in early May 1998. To ensure adequate soil drainage, raised beds were formed by alternate-direction mold-board plowing, which incorporated the ladino cover crop into the soil. Ridges were smoothed by discing, resulting in finished beds 20.3 to 30.5 cm × 1.5 m wide (8 to 12 inches × 5 ft) on 7.6-m (25-ft) centers.

adequate soil drainage, raised beds were formed by alternate-direction mold-board plowing, which incorporated the ladino cover crop into the soil. Ridges were smoothed by discing, resulting in finished beds 20.3 to 30.5 cm × 1.5 m wide (8 to 12 inches × 5 ft) on 7.6-m (25-ft) centers.

Seven named pawpaw varieties ('Davis', 'Taylor', 'Rebeccas Gold', 'Sunflower', 'Overleese', 'Prolific' and 'Wells' were commercially grafted (Northwoods Nursery, Mollala, Ore.) onto PRVT rootstock using whip-grafting at a height of 7.6 to 15.2 cm (3 to 6 inches). Non-grafted seedlings of the PRVT rootstock were also included for comparative purposes. Trees were grown in deep 5.7 × 30.5-cm (21/4 × 12 inches) pots in a greenhouse. Trees were shipped in full leaf; trunk diameter ranged in size from 3.2-6.4 mm (1/8-1/4 inch) and tree height was 30.5 to 61.0 cm (12 to 24 inches). Trees were field-planted 28 May 1998, with 2.1 m (7 ft) between trees and 7.6 m between rows, in a randomized complete block design with eight replications for the grafted varieties and four replicates for the seedlings. At planting, a soil ridge basin was formed around each tree to hold water. Tree shelters [0.9-m-tall (3-ft) × 15.2-cm-diameter fiberglass composite tubes; Tubex Treeshelters, Treesessentials Inc., Mendota Heights, Minn.] were placed around each tree after planting for the duration of the first season and through the 1998-99 winter. Pawpaws were described as an understory tree (Sargent, 1890). Although they grow and fruit well in full-sun culture, tree shelters were recommended for protection from direct sunlight the first season of growth (D. Layne, personal communication). Tree shelters also provide a windbreak and protection from browsing animals and herbicide applications. Trees were watered thoroughly by hand at planting and again on 15 July and 11 Sept. to supplement natural rainfall, which was generally adequate through 1 Aug. A transplant solution [Peters 20-20-20 (20N-8.7P-16.6K) Scotts Co., Marysville, Ohio] at 500 mg·L-1 (ppm), 0.95 L (1 qt) per tree; 1.5 g (0.05 oz.) nitrogen (N) per tree)] was applied 12 June. Trees were fertilized again 13 May 1999 with soluble fertilizer [Peters 20-20-20 at 850 mg·L-1, 0.95 L per tree; 2.5 g (0.09 oz) N per plant]. On 14 Apr. 2000, 10 g (0.35 oz) N per tree as ammonium sulfate (21%

N) was surface broadcast in a 0.76 m (2.5 ft) square around each tree.

Hardwood chip mulch was applied in a 1.2-m-diameter (4-ft) area, 10.2 cm (4 inches) deep around each tree on 15 July 1998. To help conserve soil moisture, on 10 Sept. mulch was applied between trees in a 0.9-m band. Hardwood chips controlled weeds and conserved water in plantings at KSU (Layne, 1996). Hardwood chip mulch [5.1 to 7.6 cm (2 to 3 inches)] was applied twice in 1999 to cover the entire ridge surface. Soil within 15.2 to 20.3 cm of the tree trunks was not mulched to prevent rodent and fungal problems. No mulch was added in 2000.

A 2% glyphosate solution was spot sprayed around the trees and on mulched and unmulched row surface areas to control weeds twice in each year. Weeds close to trees were removed by hand. The tree shelters proved useful for protecting the trees from herbicide drift in the establishment year. In early spring 2000, simazine at 4.7 L·ha-1 (2 qt/acre) (experimental use only) was applied on the ridge to within 30.5 cm (1 ft) of the trunks for preemerge weed control.

To evaluate natural growth habit, trees were not pruned except for dead tissue removal. Tree height, number of shoots/tree, and average shoot length were recorded in 1998, 1999, and 2000. Early bud break number from the original grafted whip and resulting primary scaffold development (no. three tiers) were also noted. Length of the longest shoot (LSL) and the most distal axillary shoot (DASL) were recorded in 1999 as an indication of relative apical dominance following the loss of the terminal shoot to cold damage. Overwinter damage was quantified by the length of dieback in the first-year whip (FYWD) in Spring 1999 and third-year terminal dieback (TYTD) in Spring 2001. Damage was expressed as the length of dieback and the percentage of the shoot length with dieback. TCSA at 20.3 cm (10.2 to 15.2 cm above the graft), maximum tree spread, and number and percentage of shoots longer than 30.4 cm were recorded for each tree in Spring 2001. The number of floral buds present were counted on 18 Apr., and at anthesis (no. of open blooms) on 27 Apr. and 9 May 2001.

All data were analyzed using analysis of variance with CoStat 6.1 (CoHort Software, Monterey, Calif.). With unequal sample sizes, a conser-