Crop production and prevailing farming practices have greatly reduced biodiversity and nearly eliminated native prairie in the central USA. Restoring small areas of prairie on cropland may increase plant biodiversity and native species abundance while benefiting the cropland. In Iowa, we incorporated buffer strips composed of prairie vegetation within catchments (0.5 ha to 3.2 ha land areas in which precipitation drained to a collection point at the slope bottom) used for corn (Zea mays) and soybean (Glycine max) production. We planted prairie buffer strips in three designs, varying the proportion of the catchment converted to buffer and/or the continuity of the buffer. Within the catchments, we determined the identity and percent cover of buffer strip plant species during 2008–2011 and of weed species in cropped areas during 2009–2011. We found 380% more species in 6 m2 of buffer strip than in 6 m2 of crop, indicating that the presence of buffer strips greatly increased catchment diversity. Plant community composition did not differ among the three buffer designs. Despite being surrounded by cropland, the buffer vegetation was dominated by native perennial species—the targeted vegetation type for both ecohydrological functions (e.g., erosion control) and native species conservation— within four years of establishment. Furthermore, weed species richness and prevalence did not differ between cropped areas of catchments with buffer strips and cropped areas of catchments without buffer strips. These results indicate that converting 10–20% of cropland to prairie buffer strips successfully reintroduced perennial species characteristic of native prairie without increasing weeds in adjacent crops.

Monitoring is a key activity in management and restoration, yet practitioners' choices among methods may be limited by available resources. Coefficients of Conservatism (C values) have been widely used among practitioners because they are easy to apply. To determine C, expert botanists assign species in a flora to one of 11 categories, from 0 to 10, with “0” assigned to species most tolerant of human disturbance, and “10” to species of highest-quality native habitats. Although first proposed over 30 y ago, C values are criticized as subjective, creating a need for external validation using independently derived metrics. Our study corroborates earlier validations of C values for forest understory species. Our work was made possible by consistent collection of herbaceous layer data from 126 plots at 38 sites across Iowa, USA, spanning five types of forests common in the Midwest: secondary, grazed, urban, managed (timber), and preserved. We used PCA to develop an independent metric of human disturbance (LHi) based on five plant attributes (exotic, annual, biennial, closed-canopy specialist, mesic-site specialist, and fern) that are determined apart from species' C values. We found high correlation between mean C and LHi (r = 0.837) and an identical pattern of means for these indices between the five forest types, with secondary forests having lowest and preserved forests highest values. This evidence supports earlier assertions that mean C provides a valid, simple, and inexpensive means to assess qualitative differences in the forest herbaceous layer of upland mesic forests due to human disturbance.

Restoring the forest herbaceous layer in remnant forests throughout the Midwestern United States (U.S.) is limited by the lack of seed and propagules for many plant species. As a result, restorationists often have limited material to work with and must seek out plant material at a regional rather than a local scale, without knowing whether regional provenances are ecologically appropriate. We conducted greenhouse and field experiments to examine persistence, growth, and reproduction of three herbaceous perennials (wild ginger, Virginia waterleaf, and James’ sedge) that could be used for restoration. The greenhouse experiment represented a common garden and was conducted to identify whether there were genetic differences in morphological characters between local plants and non-local transplants from commercial nurseries. The two-year field study was conducted to determine whether any genetic differences noted in the greenhouse persisted in a natural setting, and also to determine what planting density (two or five individuals in a 0.25 m² plot) would be sufficient for the plants to establish. In the greenhouse, growth and reproductive measures for non-local plants were generally equal to or greater than those of local plants. However, we found the reverse for many traits, particularly related to reproduction, in the field during year two. In natural field conditions local plants had equal or greater vegetative growth and reproduction than non-local plants, although both had similar persistence. Further, similar persistence and growth in low- and high-density field plots suggested that a limited number of transplants would be adequate for successful establishment of non-local transplant stock.

The forest herbaceous layer provides important ecosystem services in the central United States. However, human impacts have caused declines of many of these species. Restoration of this layer is uncommon in temperate forests, so best practices are not yet established. There has been widespread concern about negative outcomes (for example, failure due to genetic swamping or outbreeding depression) when plant material is transferred beyond a local scale. Current practice is to use local sources under the assumption that they are optimal genotypes for the site. However, few local sources are available for many species. We examined genetic variability and phenotypic plasticity by comparing performance of local and non-local populations (from sites approximately 250 km apart) of six forest herbaceous species. We used a common garden study to test for genetic differences in plant traits, and a field study to test for phenotypic plasticity. Based on the common garden we found genetic differences between local and non-local populations for each species. Trait differences we observed in greenhouse trials we also detected in the field in the first year. However, these differences diminished in the second year of the field study and we did not detect them in the four species measured in the third year. This provided evidence that phenotypic plasticity was operating, as plant characters responded plastically to local conditions. We found no evidence that local plants consistently outperformed non-local plants. These results suggest less need for strict adherence to locally sourced seeds or transplants.

Stream pollution by nutrient loading is a chronic problem in the Midwest, United States, and greater impacts on water quality are expected as agricultural production and urban areas expand. Remnant riparian forests are critical for maintaining ecosystem functions in this landscape context, allowing water infiltration and capture of nutrients before they are lost from the system. Our objective was to identify linkages between riparian forest plant community composition and water quality in remnant forested headwater streams. We identified watersheds with embedded headwater streams in three land use categories: grazed, urban, and preserved. We assessed plant community composition and nutrient storage. We sampled the forest streams to monitor discharge rates and sediment, nitrogen (N), and phosphorus (P) loads. Herbaceous communities in preserved riparian forests had more native specialist species than urban or grazed sites. Plant N content was higher in preserved forests (17.6 kg ha⁻¹ [15.7 lb ac⁻¹]) than grazed (12.5 kg ha⁻¹ [11.2 lb ac⁻¹]) or urban forests (10.5 kg ha⁻¹ [9.4 lb ac⁻¹]). Conversely, stream water total N delivery was higher in urban watersheds (0.043 kg ha⁻¹d⁻¹[0.038 lb ac⁻¹day⁻¹]) than preserved (0.026 kg ha⁻¹d⁻¹ [0.023 lb ac⁻¹day⁻¹]) or grazed watersheds (0.02 kg ha⁻¹d⁻¹ [0.018 lb ac⁻¹day⁻¹]). Stream water nitrate (NO₃-N) concentration and total P delivery were highest for streams in urban areas. The most pronounced differences for plant composition and stream discharge and pollutant loads were between preserved and urban forests. Seasonal patterns were variable. We detected a weak negative but seasonally important relationship between plant N content and stream water N. We did not detect a similar relationship for P, which may indicate saturation of this nutrient in the watershed system. Detailed knowledge about relationships between land use, plant community composition, and water quality outcomes could be used to target forest restoration efforts in landscapes highly impacted by humans.