This study examined relationships between invasive species distributions and LCPI flow-return intervals and soil drainage classes using species cover data compiled from four sources. First, data were acquired from a 2002 to 2004 survey of bird habitat along the Missouri River (Bezzerides et al., 2003; Thogmartin et al., 2006). Data were from plots classified as wet prairies, early-successional forests, and mature forests, and included species cover estimates in five vegetative strata (emergent canopy, canopy, subcanopy, shrub, and groundflora). Data were collected on federal lands at Fort Leavenworth Military Reservation and two U.S. Fish and Wildlife refuges (Squaw Creek and Big Muddy), and at 2 Missouri state conservation areas (Eagle Bluffs and Howell Island) (Figure 1). Data were collected on 4 units of Big Muddy: Lisbon Bottom, Jameson Island, Overton Bottoms, and St. Aubert Island (Figure 1). Second, from 2007 to 2009, new data were collected at 48 of the “wet prairie” plots sampled during the earlier study (Bezzerides et al., 2003; Thogmartin et al., 2006) and at 19 new plots in predominantly herbaceous communities at the Boone’s Crossing unit of Big Muddy. New cover data were collected using the Beyond NAWMA plot design (Stohlgren et al., 2008), with herbaceous species cover estimates in 3, 1 m2 subplots. The analysis presented here uses the maximum cover value recorded for each species at each of the above plots during either of the two studies noted above. Additionally, during the latter study, stem counts and cover of exotic species were estimated as encountered between plots. Third, between 2007 and 2009, a forest age gradient was sampled in the upper reaches of the Lower Missouri River (Dixon et al., 2010). Cover of herbaceous species was estimated at 24, 1 m2 quadrats distributed along transects within forest stands exceeding 4 hectares in size and distributed as follows: 59 in the Gavins section, 35 in the Platte section, 4 at Squaw Creek, and 9 at Fort Leavenworth (Figure 1). Data were previously used to describe the status of and trends in cottonwood forests (Dixon et al., 2010). The analyses presented here use stand-level estimates of cover derived from these data. Finally, during an expansion of the vegetation community map at Big Muddy, data were collected in 2012 on the previously unmapped western units of the refuge (Jackass Bend, Baltimore Bottom, and Cranberry Bend) and at the north unit at Overton Bottoms (Struckhoff, 2013; Figure 1). Percent cover of 22 species identified as invasive species of concern by biologists throughout the study area was estimated in 8 cover classes (Table 1) at map classification points. Cover of the same species was estimated between sampling locations along transects approximately 10 meter wide, the lengths of which were defined by LCPI polygon and vegetation community boundaries. Analyses focused on the six most common invasive species among the above studies: garlic mustard (Alliaria petiolata), Japanese hops (Humulus japonicus), Johnson grass (Sorghum halapense), reed canary grass (Phalaris arundinacea), sweet clover (Melilotus officinalis), and velevetleaf (Abutilon theophrasti). Each sample location from the four studies was attributed with the flow-return interval and drainage class by spatially joining sample point (or transect center point) data with the LCPI model within a geographic information system.
