Abstract. In vegetated landscapes, rain must pass through plant canopies and litter to enter soils. As a result, some rainwater is returned to the atmosphere (i.e., interception, I) and the remainder is partitioned into a canopy (and gap) drip flux (i.e., throughfall) or drained down the stem (i.e., stemflow). Current theoretical and numerical modeling frameworks for this process are almost exclusively based on data from woody overstory plants. However, herbaceous plants often populate the understory and are the primary cover for important ecosystems (e.g., grasslands and croplands). This study investigates how overstory throughfall (PT,o) is
partitioned into understory I, throughfall (PT) and stemflow (PS) by a dominant forb in disturbed urban forests (as well as grasslands and pasturelands), Eupatorium capillifolium (Lam., dogfennel). Dogfennel density at the site was 56 770 stems ha−1, enabling water storage capacities for leaves and stems of 0.90±0.04 and 0.43±0.02 mm, respectively. As direct measurement of PT,o (using methods such as tipping buckets or bottles) would remove PT,o or disturb the understory partitioning of PT,o, overstory throughfall was modeled (PT,o′) using on-site observations of PT,o from a previous field campaign. Relying on modeled PT,o′, rather than on observations of PT,o directly above individual plants means that significant uncertainty remains with respect to (i) small-scale relative values of PT and PS and (ii) factors driving PS variability among individual dogfennel plants. Indeed, PS data from individual plants were highly skewed, where the mean PS:PT,o′ per plant was 36.8 %, but the median was 7.6 % (2.8 %–27.2 % interquartile range) and the total over the study period was 7.9 %. PS variability
(n=30 plants) was high (CV > 200 %) and may hypothetically be explained by fine-scale spatiotemporal patterns in actual overstory throughfall (as no plant structural factors explained the variability). The total PT:PT,o′ was 71 % (median PT:PT,o′ per gauge was 72 %, with a 59 %–91 % interquartile range). Occult precipitation (mixed dew and light rain events) occurred during the study period, revealing that dogfennel can capture and drain dew to their stem base as PS. Dew-induced PS may help explain dogfennel's improved invasion efficacy during droughts (as it tends to be one of the most problematic weeds in the improved grazing systems in the southeastern US). Overall, dogfennel's precipitation partitioning differed markedly from the site's overstory trees (Pinus palustris), and a discussion of the limited literature suggests that these differences may exist across vegetated ecosystems. Thus, more research on herbaceous plant canopy interactions with precipitation is merited.