Delaware Inland Bays Watershed
Nutrient Management Project
John Mackenzie, John H. Martin, Lilian Pintea, Boonchauy Boonmee, Negede Gedamu
and Theresa C. Thomas

Contents


Introduction

The Inland Bays of Sussex County, Delaware, include Rehoboth Bay, Indian River Bay, Little Assawoman Bay and the northernmost tip of Assawoman Bay. These shallow bays formed behind Atlantic barrier beaches along the Eastern edge of the Delmarva coastal plain. The Inland Bays region is defined as the land area draining into these bays plus the Lewes-Rehoboth Canal/Coast area immediately to the north. This region supports intensive corn, soybean and poultry production as well as a large coastal tourism industry. Agriculture in the watershed appears to be a significant contributor of nitrogen leachate and phosphorus runoff into the Inland Bays. Poultry litter is the principal source of N and P from agriculture.

The overall objective of this research is to identify economically efficient methods of minimizing agricultural runoff into the Inland Bays. We are using GIS tools to analyze terrain and land-use patterns in order to clarify spatial linkages between non-point pollutants and land uses at the sub-basin level, and to help formulate appropriate non-point pollution management strategies. Policy implications of these research results are discussed in several sections of this monograph.

This research project addresses five questions:

  1. Is there a correlation between intensity of agricultural nutrient use and water quality degradation in the Inland Bays?  We delineate drainage sub-basins within the Inland Bays, construct indices of nutrient generation from poultry production by sub-basin, and compare the spatial distribution of nutrient indices to benthic quality indices.
  2. To what extent can phosphorus runoff be reduced by nutrient trading between individual sub-basins of the Inland Bays?  In general, poultry litter is disproportionately high in P relative to crop needs; consequently most soils in the watershed have excessive P content. A nutrient budget analysis by sub-basin may indicate P-deficit areas which could absorb excess litter from P-surplus areas.
  3. Does the watershed have significant vegetative stream buffers to impede P runoff into streams?  Due to its low water-solubility, P tends to move overland into stream channels. The Conservation Reserve Program offers incentives to farmers to create and maintain vegetative riparian buffers to reduce P runoff. Vegetation density indices derived from satellite imagery can be used to gauge vegetation densities within stream corridors.
  4. If Delaware were to implement a complete riparian buffer program for all cropland within the Inland Bays watershed, how much acreage would be involved, and how might such a program be implemented?  Buffers of varying width (20, 40, 60, 80 and 100 meters) are delineated around all hydrography features in the watershed, and cropland acreages within these buffers are quantified.
  5. Is a riparian buffer program economically viable for Inland Bays agriculture?  This report outlines a few policy options for converting cropland in riparian zones to vegetative buffers, and discusses likely economic effects of such policies.
This research project has been supported in part by the Center for the Inland Bays, and is, in part, a complement to the Horsley-Whitten study "Assessment of Nitrogen Loading to the Delaware Inland Bays" (1998) commissioned by the Center.

Watershed Re-Delineation from Updated 1:24,000-scale Hypsography

A preliminary task in this research was to obtain accurate delineations of the Inland Bays Watershed and the individual sub-basins within it. 1992-93 USGS hypsography data (elevation contour lines at 5-foot elevation intervals) were used to create 30-meter resolution Digital Elevation Models for all 7.5-minute quads in Sussex County.  The source hypsography data were obtained in DLG optional format from the Delaware Geological Survey.  These files were imported to GRASS GIS as vector line features, labeled with elevation values, and rasterized at 30-meter resolution.  The raster hypsography files covering the study area were patched together, the GRASS r.surf.contour module was used to interpolate a full 30-meter resolution DEM, and the GRASS r.watershed module was used to determine inferred drainage patterns and re-delineate the principal watersheds in the county at higher resolution that current hydrologic unit area maps.

Sussex Co. Watershed Delineation (raster 30M-res, UTM18--NAD83)

The image on the left shows an analytically hill-shaded version of the DEM for the Inland Bays Watershed. The image was created by compositing hue from the source DEM and intensity from an aspect map generated from the DEM. The image on the right shows the initial drainage basin delineation from the r.watershed module, with black lines indicating groupings of sub-basins into the five principal basins.  Portions of Sussex county outside these basins are masked out of both images.

These sub-basin delineations were then corrected via on-screen digitizing for general consistency with 1995 TIGER hydrography data. Basin boundaries were adjusted so that natural streams do not cross basin divides.  The area below the Indian River Bay contains numerous ditches, including ditching across natural basin boundaries; no attempt was made to adjust these boundaries. It should be noted that the TIGER hydrography data for Sussex County are largely derived from old USGS map sources digitized at 1:100,00 scale, and their spatial accuracy is arguably lower than the accuracy of the new hypsography data from which the DEM's and drainage delineations were derived.

The sub-basins were then aggregated to the equivalent of 14-digit hydrologic unit areas.  This mapping is reasonably consistent with NRCS's new draft HUA14 watershed mapping for Sussex County.

Under contract with DNREC, Photo Science, Inc. of Gaithersburg, MD, (now known as Earth Data, Inc.), developed a 1992 mapping of Land-Use/Land Cover for Delaware from its 1992 digital ortho-photo series. The LULC mapping is based on digitized polygons, with a 4-acre minimum mapping unit. Digital vector data for each county were reprojected in Arc/INFO to UTM (NAD 1983), exported to GRASS and rasterized at 30x30-meter resolution. The following table presents a generalized 1992 LULC categorization of each basin.  Pixels classed as wetland in 1992 and forest in 1984 were reclassed to forest.

TABLE 1: DELAWARE INLAND BAYS: GENERALIZED LAND-USE/LAND COVER, 1992

LULC CATEGORY   Coast/Canal   Rehoboth   Indian R  L.Assawmn  Assawoman      TOTAL
--------------  -----------  ---------  ---------  ---------  ---------  ---------
residential         1785.27    5152.33    8562.76    1688.76    1064.95   18254.07 
comml/indl/infr      925.22     782.15    4142.76     431.73     431.81    6713.66 
recreation           127.06     559.31     463.83      33.43      62.71    1246.35 
agriculture         2027.90   14298.71   44155.46    9431.04    4836.24   74749.36 
brushland            249.08    1035.89    2760.01     146.41      92.00    4283.38 
deciduous forest      50.04     545.89    5249.44    1736.35     830.03    8411.75 
coniferous forest   1143.97    1328.04    8605.16     994.23     760.28   12831.68 
mixed forest        1040.34   10741.81   24385.36    2137.54    1008.32   39313.36 
water                293.33     948.57    1488.16     474.65     212.09    3416.80 
wetlands            1988.02    3244.89    6222.10    2013.89     780.14   14249.05 
beach/barren        1112.02    1107.43    2147.47     450.49      94.89    4912.30
other/outside DE      62.79       6.97       0.00      22.91    2013.96    2106.63
----------------   --------   --------  ---------  ---------   --------  ---------
TOTAL ACRES        10805.04   39751.99  108182.51   19561.42   12187.42  190488.38
Spatial Distribution of Poultry Production

The first phase of this research involved calculation of nutrient budgets for the sub-watersheds within the Inland Bays. Poultry litter represents the primary source of nutrients in the watershed; additional nutrients, particularly N, are applied in commercial fertilizers.

Current poultry farm locations were mapped by Gedamu from Sussex County tax maps. These point data include numbers of poultry houses and aggregate flock capacity. These data update a comprehensive mapping of individual poultry houses developed by Sparco (1995) from USGS quad maps, and eliminate a large number of now-defunct houses. The farm capacity data are used to calculate aggregate N, P and K values in land applications of poultry litter, following nutrient management calculations developed by Sims (1998).

The following table summarizes poultry farm capacity per acre of cropland (birds/acre) by sub-basin in order to identify problem sub-basins. This analysis indicates particularly high concentrations of poultry litter per acre of cropland in the sub-basins draining into the southern Indian River Bay.

TABLE 2: CROPLAND ACREAGE AND POULTRY PRODUCTION DENSITIES, BY SUB-BASIN
          
 sub-basin                                  total  aggr flock  cropland  birds per
                                            acres    capacity     acres  crop acre
Coast/Canal
 Lewes & Rehoboth Canal . . . . . . . .    10,805           0     2,028          0
Rehoboth Bay
 Western Rehoboth & Bethany Beach . . .     6,911           0     1,112          0
 Love Creek . . . . . . . . . . . . . .    11,970     257,787     5,564       46.3
 Herring Creek. . . . . . . . . . . . .    20,871     361,094     7,623       47.4
Indian River Bay
 Swan Creek . . . . . . . . . . . . . .    14,998     752,383     5,696      132.1
 Morris Mill Pond . . . . . . . . . . .    13,459     387,403     4,598       84.3
 Cowbridge-Millsboro Pond . . . . . . .    15,075   1,363,925     7,464      182.7
 Betts Pond . . . . . . . . . . . . . .    10,987     950,993     4,790      198.5
 Iron-Whatons Branch. . . . . . . . . .    14,573   1,160,539     6,342      183.0
 Peppers Creek. . . . . . . . . . . . .     8,293   1,278,722     4,219      303.1
 Vine Creek . . . . . . . . . . . . . .    13,843   1,498,914     5,410      277.1
 Blackwater Creek . . . . . . . . . . .    16,955     740,123     5,635      131.3
Little Assawoman Bay
 Little Assawoman . . . . . . . . . . .     9,091     570,695     3,367      169.5
 Dirickson-Little Assawoman . . . . . .    10,470   1,476,658     6,065      243.5
Assawoman Bay
 Assawoman-Buntings Branch. . . . . . .    12,187   1,062,601     4,836      219.7
 ---------------------------------------------------------------------------------
 TOTAL                                    190,488  11,861,837    74,752      158.7
Boonmee's nutrient budget analysis is based on soil test results recorded by the University of Delaware Soil Testing Lab for zip codes within the Inland Bays watershed. Over 70% of commercial soil tests performed were excessive in P; another 21% were high in P. Over 9% of tests were excessive in K; another 67% were high in K.

The details of the nutrient budget calculations are reported in Boonmee (1998). The are based on the following very conservative assumptions:

  1. cropping patterns (wheat, corn and soybeans) are uniform across all sub-basins, and are proportional to 1996 crop acreages reported for these three crops for all of Sussex County.
  2. poultry litter generated within a sub-basin is applied on cropland within that basin based on crop nutrient requirements; there is no transfer of litter between sub-basins.
  3. commercial fertilizers are applied only as needed to remedy particular deficiencies in N, P or K for a
These assumptions imply that our reported nutrient surpluses are very conservative lower-bound estimates of true surpluses. Nevertheless, these numbers indicate substantial nutrient surpluses on the majority of cropland in the watershed, and confirm that the principal problem is excess P.
TABLE 3: ESTIMATED N, P AND K SURPLUS/DEFICIT, BY SUB-BASIN
                              -BASINWIDE TONS-  CROPLAND   --POUNDS/CROPLAND ACRE--
sub-basin                       N     P     K     ACRES      N        P        K

Coast/Canal
 Lewes/Rehoboth Canal         -52   -16   -11     2,028   -51.28   -15.78   -10.85
Rehoboth Bay
 W.Rehoboth/Bethany           -29    -8    -6     1,112   -52.16   -14.39   -10.79
 Love Creek                   -96    12     5     5,564   -34.51     4.31     1.80
 Herring Creek               -129    18     8     7,623   -33.84     4.72     2.10
Indian River Bay
 Swan Creek                    -2   116    71     5,696    -0.70    40.73    24.93
 Morris Mill Pond             -45    47    27     4,598   -19.57    20.44    11.74
 Cowbridge-Millsboro Pond      70   232   143     7,464    18.76    62.17    38.32
 Betts Pond                    59   165   102     4,970    23.74    66.40    41.05
 Iron-Whatons Branch           87   201   125     6,342    27.44    63.39    39.42
 Peppers Creek                138   239   149     4,219    65.42   113.30    70.63
 Vine Creek                   149   277   172     5,410    55.08   102.40    63.59
 Blackwater Creek              -3   114    69     5,635    -1.06    40.46    24.49
Little Assawoman Bay
 Little Assawoman              23    95    59     3,367    13.66    56.43    35.05
 Dirickson-Little Assawoman   128   267   166     6,065    42.21    88.05    54.74
Assawoman Bay
 Assawoman-Buntings Branch     79   188   117     4,836    32.67    77.75    48.39

Correlation Between Nutrient Surpluses and Benthic Quality

Data provided by Pintea (University of Delaware College of Marine Studies) indicates a clear spatial association between agricultural nutrient surpluses and benthic quality in the Inland Bays.  Although the Indian River Bay has faster flushing than Rehoboth Bay, the lowest benthic quality indices are found at the outlets of the Iron-Whatons Branch, Vine Creek and Peppers Creek drainages.  The first graphic shows interpolations of five selected water quality indices; the second and third graphics show alternative interpolations of overall benthic quality indices.

Potential for Nutrient Runoff Reduction Via Intra-Watershed Redistribution of Poultry Litter

Reducing non-point agricultural runoff by transferring poultry litter between sub-basins within the Inland Bays watershed does not appear to be a feasible strategy. The study area has large overall surpluses of N, P and K. The few nutrient-deficit sub-basins have relatively small deficits.  Furthermore, the deficit sub-basins are clustered in the northern portion of the watershed, so that surplus litter might typically be transported 20 or more miles to these deficit sub-basins within the study area.

Since areas adjacent to the Inland Bays (e.g. the Nanticoke watershed) have similar poultry production densities and cropland characteristics, growers in the study area will have difficulty finding sufficient outlets for excess litter within any feasible distance.

The next two phases of this research analyze current and potential use of riparian vegetative buffers to control non-point runoff.

Vegetation Density and Texture

The second phase of this research compares vegetation densities in riparian zones against vegetation densities throughout the Inland Bays Watershed. SPOT satellite imagery for August 30, 1994, November 30, 1995, and July 5, 1996 were co-registered to UTM coordinates and used to map vegetative biomass density indices and vegetative texture within the watershed.  The color-infrared images are from RGB composites of 20-meter resolution multispectral data. Simple histogram-equalization was applied to each band file's color table prior to compositing. The greyscale image is from 10-meter resolution panchromatic data. (All SPOT images @CNES 1994-96, Licensed by Spot Image Corp, Reston VA)


Three Normalized Difference Vegetation Index (NDVI) maps were calculated from the 3 sets of multispectral band files using the formula

max(0,255*(IR-red)/(IR+red)).

NDVI values should typically exhibit higher seasonal variation for cropland pixels than for most other land-use/land cover categories.

Three vegetation texture maps were created by calculating NDVI standard deviations in a 5-by-5-cell neighborhood. Cropland typically exhibits low spatial variation in vegetative densities relative to most other land-use/land cover categories.


Riparian vegetative buffers should be identifiable by distinct spectral signatures and NDVI values. To compare NDVI values of cells adjacent to streams versus all other cells in the watershed, we delineated 20-, 40-, 60-, 80- and 100-meter buffer zones around all water features in the watershed at a 10x10-meter cell resolution. We then extracted NDVI values of cells within the 20-, 60- and 100-meter buffers from each scene, and compared the histograms of these NDVI values against histograms of NDVI values for all cells outside the 100-meter buffer. The histograms show significant seasonal variation, but there are no discernible differences between any riparian buffer NDVI and non-buffer NDVI distributions in any scene. A specific comparison of all cropland NDVI's versus cropland NDVI's within 60-meter riparian buffers yields the same result. These analyses fail to detect any substantial vegetative buffering of stream features in the Inland Bays Watershed at present.

Identification of Cropland Suitable for Conversion to Riparian Buffers

The third phase of this analysis involves identification of cropland appropriate for a vegetative riparian buffer program, discussion of general management principles for such buffers, and a rough estimate of likely costs of implementing a comprehensive riparian buffer program throughout the watershed.

The following table quantifies cropland acreage within the 5 successive buffer zones.

TABLE 4: CROPLAND ACREAGE, BY DISTANCE FROM WATER,
DELAWARE INLAND BAYS WATERSHED

                                                      cum.  cum. % of
                                          acres      acres   cropland

 coincident with water features. . . . . .  668        668      0.89%
 0-20 meters . . . . . . . . . . . . . .  2,629      3,297      4.41%
 21-40 meters. . . . . . . . . . . . . .  2,798      6,095      8.15%
 41-60 meters. . . . . . . . . . . . . .  2,901      8,996     12.03%
 61-80 meters. . . . . . . . . . . . . .  2,661     11,657     15.59%
 81-100 meters . . . . . . . . . . . . .  2,898     14,555     19.47%
We can quantify cropland acreage within any buffer distance by sub-basin straightforwardly. There are an estimated 8,996 acres of cropland in the Inland Bays Watershed located within 60 meters of water features (which we consider to be a fairly wide buffer). This includes a substantial amount of ditched cropland in the Iron-Whatons Branch, Blackwater Creek and Little Assawoman sub-basins.
TABLE 5: CROPLAND ACREAGE WITHIN 60M RIPARIAN BUFFER ZONE, BY SUB-BASIN

 sub-basin                                                      acres
------------------------                                        -----
Coast/Canal
 Lewes & Rehoboth Canal . . . . . . . . . . . . . . . . . . .      44
Rehoboth Bay
 Western Rehoboth & Bethany Beach . . . . . . . . . . . . . .      55
 Love Creek . . . . . . . . . . . . . . . . . . . . . . . . .     103
 Herring Creek. . . . . . . . . . . . . . . . . . . . . . . .     209
Indian River Bay
 Swan Creek . . . . . . . . . . . . . . . . . . . . . . . . .     210
 Morris Mill Pond . . . . . . . . . . . . . . . . . . . . . .     419
 Cowbridge-Millsboro Pond . . . . . . . . . . . . . . . . . .     515
 Betts Pond . . . . . . . . . . . . . . . . . . . . . . . . .     601
 Iron-Whatons Branch. . . . . . . . . . . . . . . . . . . . .   1,350
 Peppers Creek. . . . . . . . . . . . . . . . . . . . . . . .     997
 Vine Creek . . . . . . . . . . . . . . . . . . . . . . . . .     885
 Blackwater Creek . . . . . . . . . . . . . . . . . . . . . .   1,231
Little Assawoman Bay
 Little Assawoman . . . . . . . . . . . . . . . . . . . . . .   1,008
 Dirickson-Little Assawoman . . . . . . . . . . . . . . . . .     837
Assawoman Bay
 Assawoman-Buntings Branch. . . . . . . . . . . . . . . . . .     534
 --------------------------------------------------------------------
 TOTAL                                                          8,996

 Total cropland acreage in Inland Bays . . . . . . . . . . . . 74,752

Policy Discussion

Although conversion of approximately 9,000 cropland acres within the Inland Bays Watershed to 60-meter wide riparian buffers appears to be an ambitious policy objective, the likely cost of such a program is probably quite reasonable. Fair-market rental payments on this land at $50 per acre (consistent with average cropland rental figures for non-irrigated land in the county) would only total about $450,000 per year for the entire watershed, excluding administrative costs. Long-term management of these buffers should focus on minimizing nutrient applications, planting to maximize retention of P runoff, and removal of P in hay, forest products or some other P-absorbing crop.

Annual broiler production in the Inland Bays can be roughly estimated by multiplying aggregate flock capacity (conservatively estimated at 12 million birds) times 6 flocks per year to obtain 72 million birds/year -- about 28 percent of Delaware's total 1997 broiler production (Delaware Agricultural Statistics Service 1999).  Based on an average liveweight of 5.5 pounds per bird, the Inland Bays is producing roughly 400 million pounds of broilers annually.  Multiplied by an average 1997 value of 37.5 cents per pound, the total value of poultry production in the Inland Bays is roughly $150 million annually.

The Environmental Protection Agency is currently proposing a tax on poultry processors of one-half cent per pound to fund agricultural non-point source pollution control programs.  Under this tax, Inland Bays broiler production would generate an annual tax revenue of $750,000 -- easily enough to fund a comprehensive riparian buffer program in the watershed.

We sketch out a refinement of this tax concept and discuss the probable economic implications of a riparian buffer program funded by this kind of tax:

The local poultry industry is comprised of hundreds of contract growers and five regionally-dominant integrated companies which provide flocks and feeds to contracted growers; transport, slaughter and process chickens; and control intermediate marketing of chicken products.  There is a high degree of buyer-side concentration in the intermediate market for live birds.  Some contract growers are also involved in feed crop production, which serves as a hedge against poultry price risk, since grain prices move more or less inversely with prices received by growers for live birds.  A majority of contract growers do not farm significant cropland acreage, however (Michel, 1996).

Although the tax would be levied on the integrators at the processing stage, economic theory indicates that some portion of this tax burden will be passed backward in reduced live bird prices paid to poultry growers, and another portion will be passed forward to retailers.  Prices paid to growers currently range as low as four cents per pound--evidence that that the five regional integrator companies do exert some oligopsony power over contract growers.  Thus any significant transfer of tax burden back to growers will squeeze grower revenues.

A riparian buffer program would probably have relatively minor effects on poultry feed costs.  Like the rest of the Delmarva Peninsula, the Inland Bays is a feed-deficit region, requiring significant imports of feeds from other parts of the US to support its poultry production.  For growers who also produce feed crops, possible increases in local feed crop prices are unlikely to offset reductions in finished bird prices due to pass-through of processors' higher feed costs.

Current political resistance by integrators and growers to this proposed tax suggests both parties are aware of the likely tax shift effects.  Recent anecdotal evidence indicates that the mere anticipation of future controls has made cropland owners more reluctant to accept litter, and left some growers' with no outlet for their litter.  As noted above, most of the watershed (and most of Delmarva) is in nutrient surplus, so that shipping manure to nutrient-deficit areas will involve substantial transportation costs.

Riparian buffers alone will not solve the nutrient management problems in the study area.  Current research indicates that phytase additives in poultry feeds can improve bird metabolism of phosphorus and thus reduce phosphorus content in poultry litter by 25% or more.  Other promising management techniques includes alum treatments, better composting, pelletizing, etc.  An economically efficient management will probably involve some combination of these practices.

We note that a riparian buffer program could provide additional environmental benefits: wildlife habitat, stormwater and erosion control, etc.  Mackenzie, Heckscher and Broaddus (1998) analyzed Delaware Natural Heritage Program data on occurrences of rare and endangered species, and demonstrated that forested stream corridors support more rare species in Delaware than any other land-use/land cover category.

In separate research, one of us (Mackenzie, 1999) has conceptualized a graded tax system, with poultry production in each sub-basin taxed according to water quality monitoring data obtained from each basin outlet.  Poultry produced in high-pollution sub-basins could be taxed at a higher rate than poultry produced in low-pollution sub-basins.  The tax would be collected from integrators via a flock manifest system, and adjusted annually to reflect changes in water quality measures at sub-basin outlets.  Each sub-basin would have a water quality management cooperative comprised of farmers, which would use tax revenues from its own sub-basin to finance least-cost nutrient management practices within the basin.  These cooperatives would be somewhat analogous to tax ditch associations, receiving technical assistance from county conservation districts, NRCS and other agencies.  Each cooperative would thus have a competitive incentive to identify and implement cost-efficient water quality improvements in order to improve its members' competitive position relative to growers in other sub-basins.  The efficiency of this tax program would depend on good scientific, Extension and GIS support.

Future Research Directions: Incorporation of Soils Data, Benefit-Cost Analysis of Riparian Buffer Program, Analysis of Other Non-Point Pollution Sources, ...

The buffers we have delineated in this study do not account for varying soil properties  in the watershed.  Optimal buffer widths will vary with slope, soil permeability, erodability, etc.  We have obtained updated SSURGO soils data under cooperative agreement from NRCS for the Inland Bays. This soils mapping covers the Inland Bays area as delineated by DNREC, which is not entirely consistent with our DEM-based delineation.

Soil drainage characteristics in the Inland Bays vary widely. In general, the soils to the west and north of the Rehoboth and Indian River Bays are coarser-textured and well drained, while the soils between Indian River Bay and Little Assawoman Bay are finer-textured and poorly-drained (hence the higher incidence of drainage ditches). These data will be used to derive differential soil leaching and runoff coefficients to support more precise nutrient budgeting in the study area. The soil differences are likely to reinforce the above findings: riparian buffer protection efforts should be targeted to areas with poorly-drained soils.

The rough estimate of costs of a riparian buffer program in the Inland Bays Watershed do not appear particularly prohibitive. However, we propose a follow-up benefit-cost analysis of such a buffer program in order to provide a firmer policy justification for it. The establishment of riparian buffers would provide a range of environmental benefits: improved wildlife habitat, hunting, fishing, boating, swimming, etc. This analysis would include a landowner survey to determine actual willingness to participate in a buffer program (to estimate actual economic costs of the program under voluntary versus compulsory participation) as well as a more general contingent valuation survey to evaluate the economic benefits of likely water quality improvements in the Inland Bays resulting from such a program.

This study has focused on agriculture exclusively.  While agriculture is clearly the major source of nutrients  in the Inland Bays (as confirmed by Horsley-Whitten, 1998), it is not the only source.  Horsley-Whitten analyzed nitrogen loadings from the entire spectrum of land uses in the watershed.  Their report notes that unsewered residential developments probably contribute more N per acre to the Inland Bays than cropland. On the other hand, the watershed contains about 20 times as much cropland as unsewered residential acreage.
 

References:

Delaware Agricultural Statistics Service.  1999.  “Poultry highlights.”  NASS/USDA (http://www.nass.usda.gov/de/p2097.htm)

Horsley & Whitten, Inc.  1998.  Assessment of Nitrogen Loading to the Delaware Inland Bays.  Prepared for the Center for the Inland Bays, Nassau, DE. (May, 1998)

Mackenzie, John, Kitt Heckscher and Lynn Broaddus.  1998.  Mapping wildlife habitat protection priorities in Delaware.  Spatial Analysis Lab, University of Delaware, Newark, DE.  (http://www.udel.edu/FREC/spatlab/habitat/habitat1.htm)

Mackenzie, John.  1999.  “Grow a local solution.”  Perspective column, Wilmington Sunday News Journal, February 14, 1999, pp. E1-4.

Martin, J.H., J.G. Farrell and J. Mackenzie.  1998.  An Analysis of Nutrient Utilization Efficiency by Agriculture in Delaware’s Inland Bays Drainage Basin.  Final project report submitted to the Center for the Inland Bays, Nassau, Delaware.  (October 1998)

Michel, K., et al. 1996.  Nutient management by Delmarva poultry growers: a survey of attitudes and practices. Food & Resource Economics Research Report 96-01, University of Delaware, Newark, DE.

Sims, J.T. 1999. Soil fertility evaluation. p. D113-D151. In M.E. Sumner (ed.) Handbook of soil
science. CRC Press, Boca Raton, FL.

Sparco, John.  1995.  A Benefit-Cost Methodology for Analysis of Nitrate Abatement in Sussex County, Delaware, Ground Water.  Ph.D. dissertation, University of Delaware Operations Research program, Newark, DE.  (May, 1995)