Seeds and cuttings I collect around North Carolina, what they turned out to be, and how to grow them for a NC Piedmont native-plant nursery.
Cornus amomum · Cornaceae
Porcelain-blue berries on red stalks, hanging over the water. Ripe this week, and its winter sticks root in a bucket.
Community photos of Cornus amomum to study before I head out: leaf, flower or bract, ripe fruit, bark, whole plant.
"Common" in all three NC provinces (Mountains, Piedmont, Coastal Plain) on shores, streams and bottomlands. A default Piedmont riparian shrub. 18 iNaturalist records in Alamance County alone, several logged from the Haw River Trail itself
Best bet: Cuttings first. Come back Dec–Mar and cut dormant live stakes, pencil-to-thumb-thick, 18–24 in, stuck two-thirds deep into wet ground; streambank-restoration work reports essentially 100% take from dormant-season material, no hormone, no mist bench. Seed second, for genetic diversity: macerate the fruit by hand, float off pulp and floaters, then either fall-sow the cleaned stones ¼–½ in deep outdoors and let winter do the work, or cold-moist stratify 90–120 days at ~40°F. Never let cleaned seed dry; that alone pushes germination to the second spring.
A Piedmont native at the easy end of its range (USDA 5a–8b), and one of the few natives genuinely happy in the spots that defeat everything else: rain-garden bottoms, pond and ditch edges, the soggy back corner, heavy clay that stays wet. Tolerates full sun to deep shade and takes seasonal flooding without complaint. Two cautions: it wants moisture and will sulk on a baked droughty slope; and it sprawls, 6–12 ft high and as wide, layering where branches touch soil, so site it as a bank-holder or thicket. A bonus for a wild-collected nursery: the "silky dogwood" of the nursery trade is often actually C. obliqua or C. sericea, so locally-foraged Haw River material is the provenance-correct thing.
Each look-alike with the one character that settles it. Tap a photo to zoom.
Also common in NC wetlands, also opposite, also blue fruit on reddish stalks. Split the twig: C. foemina's pith is white and its young twigs are hairless and slick, against silky dogwood's brown pith and finely hairy twigs. Backup characters: foemina runs only 3–4 vein pairs per leaf side (amomum 5+), its leaves are narrower and hairless-feeling beneath, and it grows into a stiffly ascending 10–25 ft small tree rather than a 6–12 ft sprawl.
White fruit on brilliant coral-red pedicels that stay red and showy long after the birds strip the berries, against blue fruit on merely reddish stalks. Also rhizomatous, forming even-aged colonies out in old fields rather than hugging the bank. "Infrequent" in NC and Piedmont-only, but it is in range here.
White fruit and, in winter, brilliant coral-red glossy stems on a white-pithed twig, against silky's blue fruit, duller maroon-purple twigs and brown pith. In NC essentially a mountain plant plus landscape plantings; not expected wild on the Haw.
White fruit plus a leaf upper surface that feels like fine sandpaper, with erect rather than appressed hairs beneath. Mostly academic here: C. drummondii doesn't occur in NC, and C. asperifolia is very rare in the southeastern corner of the state.
Alternate leaves, clustered in false whorls at branch tips, the only alternate-leaved dogwood in the East. Its drupes are blue-black on vivid coral-red pedicels and ripen a month or two earlier. A cool mountain-cove species; on the Haw floodplain it would be the anomaly.
The real non-dogwood trap: opposite leaves, wet ground, blue-black fruit in flat clusters. Viburnum veins run straight out into the margin (and into the coarse teeth, on V. dentatum) instead of arcing forward to parallel it. Confirm with the tear test: pull a leaf gently in half and Cornus leaves stay joined by fine white elastic threads; viburnum's don't.
Don't walk the trail looking for a plant. Walk it looking for a color: chalky, opaque, slightly unreal porcelain blue, the blue of a cheap ceramic bead, hanging in loose drooping clusters on reddish stalks at chest height over the water. Nothing else on the Haw makes it. Blackgum and viburnum go blue-black and glossy, pokeweed goes purple-black and juicy. Silky dogwood's berries look chalked. Half of them will be mottled blue-and-white, caught mid-ripening, which is itself a giveaway.
The other half of it is the shape: a leaning, multi-stemmed, 6–12 ft shrub that never stands up straight, throwing arching maroon canes out over the current from the top of the bank. It grows exactly where the trail is wettest and brushiest, in backwater sloughs, the mouths of feeder ditches, seeps, gravel and sand bars, the tangle I'd normally push through without looking. So push through and look.
Go now. Of the 128 fruiting iNaturalist records for North Carolina, 73 fall in August and 23 in September, and the count collapses to 2 in October. That isn't the fruit rotting. It's fall-migrating birds, for whom silky dogwood is a preferred food and who clear a bush in a couple of days.
Standing at a candidate, the ID is three seconds of hand work: opposite leaves in pairs at every node, arcuate dogwood veins, and a thumbnail split into a second-year twig to find brown pith.
Silky dogwood gets written three ways. Weakley's Flora of the Southeastern US splits the old catch-all Cornus into several genera on phylogenetic grounds and files this plant as Swida amomum; most regional sources, including NC Extension, still say Cornus amomum. Same plant. Separately, the northern/midwestern C. obliqua (= C. amomum ssp. obliqua), with narrower leaves, cuneate rather than rounded leaf bases, and a whitish underside with uniformly rigid white appressed hairs, is now generally treated as its own species. It doesn't reach here: in Virginia it's already rare and at its range edge in the mountains and northern Piedmont, with intergrades only along the Potomac. On the Haw in Alamance County this is plain C. amomum.
Ripe means chalky porcelain blue, soft to a light squeeze, releasing from the pedicel with a gentle tug. A single cyme ripens unevenly, so pick berry by berry and leave the hard green and half-white ones; snipping whole heads just fills the bag with immature stones. Each drupe holds one hard round seed in a thin pulp.
Clean the same week. Soak briefly, then macerate the fruit by hand; the skin and pulp slip off easily and the stones are too hard to damage (a food processor is fine for volume). Then float: sound seed sinks, pulp and aborted seed pour off. Repeat with fresh water until only clean sinkers remain.
The one mistake that costs a year is letting cleaned seed dry out. Silky dogwood germinates well the first spring if the seed is cleaned and kept moist, and gets delayed to the second spring if it was left in the pulp or allowed to dry. Get cleaned seed into stratification or into the ground within one to two weeks.
Break dormancy by either route:
Then come back in winter, which is the real nursery route. Silky dogwood is one of the standard live-stake species for streambank restoration because it's almost impossible to fail with. Cut dormant wood December through March: straight, pencil- to thumb-thick, 18–24 in lengths, angle-cut at the bottom, buds pointing up. Drive them two-thirds of their length into wet ground and walk away. Restoration trials report essentially 100% survival from dormant-season stakes, with no hormone, no mist bench, no potting mix. Softwood cuttings in late spring with IBA under mist work too, and layering happens on its own wherever a low branch touches damp soil, so a sprawling wild plant usually has free rooted starts already attached at its skirt.
For quantity of plants, cut live stakes. The reason to collect seed anyway is genetic diversity: a stake bed is one clone, a seedbed is a population. Do both from the same mother plants, berries now and sticks in January.
Park at Saxapahaw Island Park (5550 Church Rd) and work the Haw River Trail's Saxapahaw section, roughly 2.5 mi of riverside path (full out-and-back about 6.3 mi), passing behind the Rivermill and following the river downstream. Silky dogwood is already documented on this exact trail: an iNaturalist record from the Saxapahaw segment logged 12 June 2026 at roughly 35.9452, −79.3193, and more from the same trail at Swepsonville (30 Sep), Graham, and Burlington (18 Sep), one of which the observer annotated simply "Brown pith." Eighteen records in Alamance County in all.
Work the water's edge specifically, not the upland side of the tread: the shrubby fringe on top of the bank, backwater sloughs and old channel scars, the mouths of feeder ditches and seeps, and any gravel or sand bar with woody growth on it. Where the trail pulls away from the river, so does the plant.
On a heavy-fruiting plant, flag it and note where it stands relative to normal water level. That's a mother plant twice over: berries this week, and a bundle of live stakes off the same stool in January.