Assessing the Agricultural Conservation Planning Framework toolbox in a Southern Piedmont landscape of the United States

Results and Discussion

Southern Piedmont Landscape and Conservation Concerns. Originally hardwood forest, settlers cleared the Southern Piedmont to produce corn (Zea mays L.), cotton (Gossypium hirsutum L.), and tobacco, which led to topsoil (13 to 25 cm) being eroded (Schomberg et al. 2020). Conservation starting in the 1930s led to land use changes so that the current landscapes consist of forest and pastures. Pasture is the dominant agricultural land use in both watersheds, comprising over 85% of the agricultural land area (table 1), which reflects the Southern Piedmont NRCS Major Land Resource Area (166,251 km² of which forest is 57%, cropland 2%, and pasture/hay is 15%). Conservation tillage of crop lands began in the Southern Piedmont in the 1970s and is the predominant tillage (O’Connell and Osmond 2018; Schomberg et al. 2020). The Southern Piedmont area is distinctly different from the land resource area where ACPF was developed, intended, and tested—the Central Feed Grains Region (Iowa, Illinois, Wisconsin, Indiana, Minnesota, Kansas, Nebraska, Missouri, and Ohio) that includes a total of 737,656 km² of land of which 57% is cropland, 14% pasture/hay, and 13% forest (figure 2; USGS 2019). The substantial differences in land use between these regions supports the objective of this work as there are significant differences in land uses, topography, climate, and nutrient loss pathways.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Primary land use in the Southern Piedmont land resource region compared to the Central Feed Grains and Livestock Region. Agricultural land use in the two study watersheds is representative of the Southern Piedmont.

Land uses in Terrell’s and Dry Creeks have changed over time from cropland, primarily tobacco, to mostly pasture due to the tobacco buy-out program (Schomberg et al. 2020; USDA FSA 2021), with current conservation focused on pasture management, including exclusion fencing (ponds and streams), as well as stream crossings and filter strips. Exclusion fencing is the most currently accepted structural practice; research has shown nitrogen (N), phosphorus (P), and sediment reductions 20%, 47%, and 60%, respectively, when implemented in these watersheds (Line et al. 2016, 2019).

ACPF Predicted Conservation Practices. The number of conservation practices sited by ACPF was similar between the two watersheds when scaled against the relative difference in agricultural land area. The frequency that ACPF sited grassed waterways and contour buffer strips was also reflective of the amount of row crop agriculture in each watershed (table 2). Opportunities to site grassed waterways are limited to agricultural fields in row crop production (table 2) and thus exclude pasture, which covers 86% of agricultural land in both watersheds. Of the grassed waterways sited, all were located in convergent areas in fields (i.e., concave areas where overland flow concentrates and erosion potential is high). Visual confirmation demonstrated good spatial agreement between existing grassed waterways and those placed by the ACPF (figure S1). Similar to grassed waterways, contour buffer strips are only predicted in fields under row crop production. Although there were no instances of existing contour buffer strips in the watershed, ACPF identified a majority of the row crop fields as suitable for contour buffers. In Terrell’s Creek, seven sites identified for contour buffer strips also had mean field slopes greater than 10%, indicating that these areas are also suitable for terraces based on NRCS conservation standards and ACPF documentation (Porter et al. 2018).

Unlike contour buffer strips and grassed waterways, WASCOBs were identified by ACPF for both crop and pasture and with a wide range in the number and placement based on the user-specified bank height. Bank heights between 1.0 to 1.5 m sited the greatest number of WASCOBs with varied placement, while a bank height around 4.5 m was identified as a maximum upper limit for siting WASCOBs in this area. The placement of WASCOBs generally followed the landscape incision within a field. Specifying a 1.0 m bank height would generate WASCOBs at locations within a field with the lowest amount of incision, while specifying 4.5 m bank height identified high incision areas where grade stabilization (NRCS Code 410) is likely suitable. In some scenarios, multiple WASCOBs were identified in the same field. The difference in the number and placement of WASCOBs as determined by a single user-specified bank height illustrates the importance of parameter selection and local knowledge, but also suggests the need for evaluating multiple model runs within a range of values to assess the appropriateness of results and sensitivity of outputs.

There were 84 ponds generated by the farm ponds tool, but a majority (71) did not drain agricultural land. For the 13 ACPF pond opportunities that drained agricultural land, most were in close proximity to an existing pond or impoundments. We interpret this to mean two things. First, in this case ponds are likely of low priority due to the high level of implementation that already exists in this area. Existing impoundments in both watersheds are positioned in the midlands draining to nearly all low-order stream reaches. Second, the identification of ponds by ACPF in similar landscape positions to those that currently exist suggests that ACPF is capable of identifying relevant and meaningful locations for farm ponds draining agricultural lands. However, ACPF does not detect the presence or absence of existing structural practices and thus requires visual confirmation or prior knowledge to assess their need. No NRWs were sited due to the constraints discussed previously.

Comparison of ACPF with Expert Knowledge. Results from the double-blind discussion illustrate important differences between conservation concerns and practices identified by the local experts compared to the outputs generated by ACPF. Local experts listed conservation concerns in a broad context spanning the range of land uses across the watershed, but also provided conservation practices by field or fields (table 3). As designed, ACPF identified conservation practices and their placement in the landscape primarily in row crop agriculture. To compare ACPF results with tallies from the expert discussion, instances where multiple ACPF practices of the same type were located in the same field were counted once (e.g., a field with three grassed waterways). Over 80% (81/96) of the conservation practices identified by the experts were outside the ACPF control and trap practices, as they were primarily nutrient management or soil health practices. In fields under row crop agriculture, ACPF identified three times the number of conservation practices as those identified by local experts (48 versus 15). In most instances the crop-related practices suggested by ACPF would result in a significant loss of field acreage. As such, these practices would have little landowner support and very limited adoption. In general, the experts identified conservation practices proportional to the area of farmland identified as either pasture, crop, or poultry related. As such, and with over 85% of the fields identified as pasture in these two representative Southern Piedmont watersheds, the importance of using ACPF as a conservation planning tool in areas with limited row crop agriculture is underscored and highlighted by this comparison.

The main conservation practice identified by the local experts within row crop agriculture was over the use of cover crops. Of the 17 sites discussed, 12 included a recommendation for cover crops; a filter strip and riparian buffer were also recommended for one field. The cover crop recommendations were based on known production practices, familiarity of management within those fields, and the use of practices that avoid sediment and nutrient losses when possible. This is consistent with “avoid practices,” such as cover crops, being usually recommended as a baseline before structural practices are identified in ACPF.

The need for filter strips and riparian buffers was identified by local expects based on visual inspection of imagery within Google Maps combined with personal knowledge about conservation practices at that site. In comparison, ACPF placed approximately three times the number of conservation practices than that of the local experts in each watershed respectively. The use of contour buffer strips as an in-field practice was the most identified conservation practice by ACPF and placed in 75% of cropped fields; this is not a common practice in North Carolina partially because conservation tillage has been widely and successfully adopted with concomitant low sediment and nutrient loading (Line et al. 2016, 2019; Schomberg et al. 2020), but also because of the small field size (<5 ha). Between 2005 and 2019, only two contour buffer strips were installed in North Carolina (USDA 2021). Considering practices similar to contour buffer strips during this time period, there were 179 contracts for terraces (NRCS Code 600), 4,212 contracts for contour farming (NRCS Code 330), and 137 contracts for strip cropping (NRCS Code 585) statewide. Knowing these three practices are used more frequently in the state and take less land area out of production, there may be an opportunity to assess the suitability for these three practices where contour buffer strips were produced by ACPF.

Within each watershed, pasture land had the largest number of conservation practices identified by the local experts. Almost exclusively, exclusion fencing around streams and ponds was the conservation practice most discussed. On a few occasions, however, a stream crossing was recommended. The judgment about conservation practices at a site was primarily based on personal knowledge about the landowner and history of conservation at that location. In comparison, the practices identified by ACPF were similar in number but different in practice than those identified by the local experts. ACPF identified locations of concentrated flow and depressions in the underlying digital elevation model (DEM) outputs that may be useful in locating areas for exclusion fencing, and could possibly improve the effectiveness of the fencing investment; however, because ACPF is not intended for pasture, it does not include exclusion fencing as a conservation practice (Porter et al. 2018).

The ACPF was able to site impoundments, rate runoff risk, and determine buffer needs on pasture land using the same toolbox operations designed for crop land. These outputs were pertinent directly or as a proxy to help prioritize and identify the need for riparian buffers and filter strips in and surrounding pastures. By extension, any identified riparian buffer or filter strip in a pasture will, by necessity, need exclusion fencing.

Overall, the local experts relied heavily on knowledge acquired over 20 years working in, and living next to, these watersheds. Information and understanding gained from working directly with landowners and state conservation programs helped inform their approach to resource conservation planning and to the specific practices identified within this discussion. Their knowledge about the transition from tobacco to pasture in this landscape and changes of priorities and conservation practices over time made their recommendations adapted to local needs and applicability. Although not explicitly mentioned, knowledge about state conservation programs and local incentives influenced the conservation practices identified by the experts. For instance, local working groups, such as the Jordan Lake Watershed Oversight Committee in North Carolina, have identified conservation practices of higher importance to be prioritized in the area (e.g., exclusion fencing). There was no mention about the use of WASCOBs in these watersheds by the local experts, highlighting the importance of understanding local priorities and history of conservation in a watershed.

Utilizing ACPF in the Southern Piedmont. Implementing ACPF in the Southern Piedmont requires some efforts that are different than, and extended beyond that typically required in a Midwest landscape. In particular, the development of field boundaries and the hydro-enforcement of the elevation models required region-specific attention. In the Southern Piedmont, agricultural fields are comparatively smaller, irregularly shaped, and more sparsely distributed. As a result, additional time is required when using larger map scales to locate field boundaries precisely and accurately. Additional steps were also required to correct erroneous agricultural land cover classifications (row crop, pasture, or other) resulting from the “update edited field boundaries” utility script in ACPF. This script uses the USDA Crop Land data layer to assign a land use class. Because of the data’s relatively coarse raster resolution (30 m), this tool is increasingly prone to error when fields are smaller and highly irregular in shape. In these watersheds, a current higher resolution (<1 m) orthorectified aerial image was required to visually identify the location of field boundaries and to verify or correct the corresponding land use. However, in areas without current basemap imagery, the ability to resolve and identify existing landscape characteristics may be limited.

A comparatively different workflow was also required to produce a hydrologically representative DEM. Stream paths in the Southern Piedmont are highly dissected by man-made structures such as culverts, storm drains, and small ponds. Due to the relatively large amount of forested land in the region, visually locating these structures in aerial imagery is difficult and can require in-person, on-ground verification. Additionally, ponds were distributed across the study area at a density of nearly 2 km⁻², each requiring manual enforcement. Given their small size, ponds did not interrupt flow paths as severely as road crossings. In total, the modifications required for small ponds outnumbered the corrections made for culverts at road crossings. Consequently, the time required to properly classify land use, digitize agricultural fields, and hydro-enforce stream networks in the study landscape can be higher than in the Midwest where the Daily Erosion Project has aided in the development of these datasets. Overall, the effort required to preprocess data and set up ACPF for use in the study watersheds was approximately two to three times the number of person-hours required in other landscapes with larger fields, fewer impoundments, and sparser road networks.

Further, it is also important to consider the sources and magnitude of nutrient losses in the Southern Piedmont compared to the Midwest where ACPF was developed as soils, cropping system, climate, and many other factors affect runoff, leaching, and erosion, and the appropriate conservation practices. Total N losses in the North Carolina watersheds near to the HUC-12 watersheds used in this study were approximately three times lower (8 kg ha⁻¹ y⁻¹) than Midwestern watersheds (Cedar River in Iowa-Minnesota or Lake Erie watershed in Ohio), which lost approximately 25 kg ha⁻¹ y⁻¹ (Hanrahan et al. 2021; Kalkhoff 2018). Compared to the Iowa and Ohio watersheds, annual P losses in the nearby North Carolina watersheds were three times greater in the pasture watershed (3.3 versus ~1 kg ha⁻¹) but three times lower in the crop fields (0.32 versus ~1 kg ha⁻¹). To maximize the utility of ACPF in the Southern Piedmont, the conservation practice(s) selected for output should account for the water quality benefits, appropriateness, and intended use of that practice. Particular to the Southern Piedmont, it is likely users of ACPF will benefit more from the riparian characterization aspects of the ACPF toolbox than from the tools intended for Midwestern row crop agriculture.

Although not intended for use in a pasture-dominant landscape, ACPF was able to provide data outputs, analysis, and map-based information relevant to watershed planners in the Southern Piedmont. A strong understanding of ACPF data and functionality is required to produce value for planners focused on conservation in forage production and grazing operations. In the case of the Southern Piedmont, where these operations are common, novel adaptations to the way ACPF is used may be beneficial, for example, using outputs from ACPF to identify and predict closely related conservation practices (e.g., proxies). Proxies may be used to predict and/or site conservation practices outside of the intended scope of ACPF. Specific to this study, ACPF proxies were identified and matched to practices identified by the local experts and with known value and use in local conservation efforts. The conservation practices identified by the local experts that map to ACPF proxies are illustrated in table 4.

The feasibility of using ACPF conservation proxies in the Southern Piedmont will vary based on local conservation priorities and the ability of the ACPF analyst to identify and create such proxies. This is done by either identifying local conservation practices with similar function or by extracting and interpreting ACPF outputs for a use that is different than intended. An example in this study is using ACPF outputs to locate areas in need of exclusion fencing. ACPF does not directly site exclusion fencing; however exclusion fencing is expected where streams intersect fields classified as pasture. When used in this manner, ACPF lends itself as a scouting tool, helping guide local experts to locations with likely conservation concerns. Additionally, the use of ACPF proxies may bring focus to overlooked areas where a more locally accepted practice may be considered. When paired with on-site visits and an understanding of the appropriateness of the conservation proxy to the actual conservation practice, a proxy may offer planners additional value outside its intended scope. Although a greater understanding of ACPF is required to identify and use conservation proxies in watershed planning, their value should be considered against the additional time, effort, and knowledge required to generate and assess these outputs.

Proxy conservation practices can also be considered by local watershed planners when ACPF sites practices that are not regionally recommended or without local preference, but there is need at that location for a practice with similar function. Examples include grassed waterways and contour buffer strips. ACPF predicts the use of contour buffer strips at locations in need for reduced runoff; however local practices with similar function, including strip cropping, diversions, or contour farming, may be viable alternatives. This approach broadens the applicability of ACPF and provides flexibility to help meet the needs and resources available in a watershed or region. A conservation planner could exercise their best judgment while leveraging the spatially precise identification of conservation concerns from ACPF.

The data developed for ACPF and the landscape characterization contain rich spatial and attribute data that could be of interest to planners in the Southern Piedmont. Datasets developed include hydrologically corrected elevation, high resolution streams networks, field boundaries, and a by-field crop history. Attribute information that accompanies these datasets may add value to traditional conservation planning by facilitating watershed-scale spatial analysis, queries, and visualizations. At minimum, access to these datasets in a GIS environment grants planners more capabilities than visually inspecting aerial imagery alone. Additionally, stream and field boundary datasets offer improvements on most nationally available streamline and land use data. For instance, streamlines identified by ACPF can provide a more accurate depiction of jurisdictional streams as well as pertinent descriptors about the stream and the surrounding environment. In addition to being topographically enforced, the ACPF stream layer provided attributes for stream drop, upslope drainage area, bank height, and height of low-lying land. Both improved spatial precision and the ability to compare across a common set of attributes may help extend the use of streams and land use datasets produced by ACPF in planning scenarios in the Southern Piedmont.

As previously mentioned, utilizing this tool is an intensive process even though technical support is available in the form of data development tools, training resources, and quality control guidelines. Conservation practice identification by ACPF overlapped in some instances with feedback from discussion respondents, but there are many capabilities of the tool that were not appropriate for this landscape, such as the entire suite of tools that identify opportunities for tile drainage treatment. Although buffer design recommendations were produced as well, the streams in our study watersheds were forested on both banks in most locations (Osmond and Neas 2007). Notably, the tool does not recognize existing buffers, and thus, the results of riparian analysis do not account for the existing buffer coverage in the watershed. Such limitations are more significant in the Southern Piedmont where most stream reaches consist of forested riparian buffers.