CARMEN M. THOMAS, U.S. Fish and Wildlife Service, 2800 Cottage Way, Room W2605, Sacramento, CA 95825, USA

JOHN G. MENSIK, U.S. Fish and Wildlife Service, Sacramento National Wildlife Refuge Complex, 752 County Road 99W, Willows, CA 95988, USA

CLIFF L. FELDHEIM, California Department of Fish and Game, 1416 Ninth Street, Sacramento, CA 95834, USA

Abstract: At Sacramento National Wildlife Refuge, California, we examined 2 types of deep tillage (disking and plowing) as possible management options for reducing lead pellet densities in wetlands. In addition, we examined the vegetation changes that resulted from tilling. Both disking and plowing moved lead pellets below the zone of availability for dabbling ducks (>10 cm). However, plowing moved a higher percentage of pellets into the 15--20-cm layer of sediment (P = 0.02). Similarly, plowing was more effective (P = 0.04) than disking or controls in redistributing pellets below the deeper zone of availability (>20 cm) for tundra swans (Cygnus columbianus). Maximum height of vegetation increased (P < 0.001) on tilled plots during the first and second year after treatment. Tillage initially reduced percent cover (P = 0.03) and density (P < 0.001) of swamp timothy (Crypsis schoenoides), but resulted in increased swamp timothy cover and stem density by the second year posttreatment. Percent cover by California loosestrife (Lythrum hyssopifolium) showed the opposite trend, with an initial increase (P < 0.001), followed by a decrease to levels similar to control plots in the second year. In certain managed wetlands, disking and plowing can be effective management tools for redistributing residual lead shot deeper into wetland sediments and potentailly reducing waterbird mortality due to lead poisoning.


Key words: lead poisoning, lead shot, Sacramento National Wildlife Refuge, tillage, waterfowl, wetlands.

Avian lead poisoning has been a recognized problem for >100 years (U.S. Geological Survey 1999). Shot ingestion usually occurs when birds feed in hunted areas and inadvertently ingest spent lead pellets (Sanderson and Bellrose 1986). Diving ducks typically have the highest rates of lead pellet ingestion, followed by dabbling and grazing waterfowl (Sanderson and Bellrose 1986, Pain 1990). In some wetlands with soft sediments, little residual lead shot (at depths available to waterfowl) has been found from 1 year to the next (Mudge 1984). In areas underlain by hardpan clay layers, pellets are prevented from settling beyond the level of availability to ducks and swans, and may result in continued lead poisoning in these areas (Mudge 1984, Pain 1991). One such area appears to be the Sacramento National Wildlife Refuge in northern California.

Although non-toxic shot has been required for waterfowl hunting on the Sacramento National Wildlife Refuge since 1986, lead poisoning of waterfowl continued to be diagnosed annually (Mauser et al. 1990, Rocke and Samuel 1991, Rocke et al. 1997). Snow geese (Chen caerulescens), mallards (Anas platyrhynchos), and pintails (A. acuta) are the species most at risk from lead poisoning on Sacramento National Wildlife Refuge. These 3 species constitute 60% of all waterfowl that visit the refuge. Historic lead shot ingestion rates in the Central Valley of California are approximately 16% for snow geese, 13% for mallards, and 12% for pintails (National Wildlife Health Lab, Madison, Wisconsin, USA, unpublished data). Site-specific lead exposure and mortality in sentinel mallards were investigated in 1986--90 on the Sacramento National Wildlife Refuge (Rocke and Samuel 1991, Rocke and Brand 1994, Rocke et al. 1997). One unit (Pool 8) was found to have lead pellet densities, >2,000,000 pellets/ha in the top 10 cm of sediment, among the highest ever recorded for a wetland. Lead shot ingestion rates by mallards did not differ between the historical values (1987--92) and the 1992--93 hunting season, suggesting that ingestion rates for toxic pellets have not decreased significantly since the refuge's ban on lead shot in 1986 (Ramsey 1993).

Management practices such as vegetation and water manipulation, grit provisioning, and tilling have been suggested for reducing the availability of lead pellets to waterbirds (Jordan and Bellrose 1951, Sanderson and Bellrose 1986). Encouraging the growth of submerged leafy aquatic plants may offset the toxicity of ingested lead pellets to waterfowl (Jordan and Bellrose 1951), but it will not reduce pellet availability. Maintaining a water depth sufficient to prevent most waterfowl from reaching the bottom may be an option in permanent wetlands, but will not work in areas that support diving ducks or where moist soil seed production is a primary goal. Grit provisioning is of limited use because not all species may partake and maintaining a constant supply requires continued staff effort. Rocke et al. (1997) suggested that seasonal dietary shifts may be partially responsible for reduced rates of grit (and accompanying accidental lead pellet) ingestion. For managed wetlands, tilling may be a cost effective technique for reducing the availability of lead pellets to waterfowl. In this study, we examined 2 types of deep tillage (plow and disk) as possible management options for areas known to have high pellet densities.


This study was conducted on the Sacramento National Wildlife Refuge, a 4,300-ha refuge in northern California located about 9.6 km south of the town of Willows. The refuge is divided into 66 habitat units managed primarily for seasonal wetlands and migratory birds. Most wetlands are flooded from September through March with water depths varying from 15 to 91 cm. Dominant vegetation includes hardstem bulrush (Scirpus acutus), tuberous bulrush (S. tuberosus), cattails (Typha latifolia and T. angustifolia), barnyard grass (Echinochloa crusgalli), sprangletop (Leptochloa fasicularis), smartweed (Polygonum spp.), and swamp timothy. Pond bottoms consist of a detrital layer of decomposing vegetation and mud, 20--30 cm deep, overlaying a hard clay pan of Willows series soils (Soil Conservation Service 1968). Sacramento National Wildlife Refuge has peak wintering populations exceeding 750,000 ducks and 200,000 geese annually.

Several habitat units on Sacramento National Wildlife Refuge were used for a botulism study (Rocke and Brand 1994) during which some of the highest lead pellet densities ever recorded in North American wetlands were observed. Lead poisoning was subsequently investigated using wing-clipped mallards, and toxicosis was found to be significantly more likely in Pool 8 and 2 other units with similarly elevated concentrations of lead pellets (Rocke et al. 1997). Pool 8 was chosen as our study site because of its history of hunting and high densities of lead pellets, a lack of historical tilling, and ease in managing water levels.


Experimental Design and Sampling

We randomly selected 5 rectangular plots (30 × 44 m) in Pool 8, which is the same area used for the botulism study in 1986--89 (Rocke et al. 1997). Each plot was subdivided into 3 strip transects (20 × 8 m), with each being surrounded by a 5-m buffer. In each plot, a transect was then randomly assigned to 1 of 3 treatments: (1) single passes with a Bush Hog 20-cm disk pulled by a John Deere 8640 4-wheel drive tractor, (2) single passes with a John Deere 965 7-blade moldboard plow pulled by a John Deere 8570 4-wheel drive tractor and followed by a light disking to smooth the surface, and (3) untilled control. Tillage depth was estimated for each treatment by placing a ruler in the furrow produced by the equipment and measuring to the surface of adjacent untilled sediment.

Prior to treatment (June 1997), we collected 20 randomly selected sediment cores (10 × 30 cm) from each transect at the time the wetland was being dewatered. All cores were collected with modified aluminum clam diggers (10 × 40 cm), which were fitted with plungers. As it was removed, each core was divided into 6 5-cm layers. Each layer was stored in a separate sealable plastic bag. Each sediment layer was washed through a series of sieves (4.75, 1.7, and 1.4 mm) to recover pellets. The diameter of the smallest sieve allowed recovery of shot as small as number 9 (2.03 mm diam). Pellets were identified as lead or steel using a magnet and the resistance of the pellet to compression by pliers.

After the wetland had dried sufficiently to allow the completion of the treatments (July 1997), Pool 8 was reflooded. In August, it was again dewatered to allow post-treatment sampling, which was conducted in September.

We conducted vegetation sampling in August 1998 and 1999, after the wetland had been dewatered and annual growth was complete. One-meter squares were placed at 5 random locations in each transect, and density and percent cover were measured for each species present (Hays et al. 1981). Average vegetation height was measured for each sample.

Statistical Analyses

Although both lead and steel pellets were recovered, steel pellets were recovered infrequently (Table 1). Thus, all analyses were conducted using data for lead pellets. We counted the number of pellets in each layer and used a square-root transformation to correct for non-normality of count data. Treatment effect was determined by computing the difference between the pre- and post-treatment counts. Results were back-transformed to the original units, and are presented with standard errors. All analyses were performed using STATISTICA© (StatSoft 1995).

We used a repeated measures ANOVA to examine differences in the distribution of pellets among plots, treatments, depths, and interaction terms. Insignificant terms (P > 0.10) were sequentially deleted. When a significant difference was detected (P < 0.05), Fisher's least significant difference (Fisher's LSD) was used as a multiple comparisons procedure (Ramsey and Schaeffer 1997). A 1-way ANOVA was used to examine differences among treatments in the number of pellets below 10 and 20 cm.

Plots that were plowed or disked were combined into 1 tilled category for analyses of vegetation data. Stem density was square-root transformed to normalize its distribution (Ramsey and Schaeffer 1997). We used a 2-way MANOVA to examine differences between treatments and years in vegetation density and percent cover. When a significant difference was detected (P < 0.05), Fisher's LSD was used as a multiple comparisons procedure. Results were back-transformed to the original units for presentation.


Pellet Density and Depth

We collected 1,164 pellets from 600 sediment cores; 1,074 (92%) were lead and 90 (8%) were steel (Table 1).

Prior to treatments, the estimated density of lead pellets in sediment cores (controls included) was 427,864 pellets/ha (2.02 ± 0.10 pellets/core sample). After tilling, the estimated density of lead pellets in sediment cores (controls excluded) was 215,700 pellets/ha (1.53 ± 0.11 pellets/core sample). In the top 10 cm of sediment prior to treatments, the estimated density was 342,999 pellets/ha (1.61 ± 0.07 pellets/core sample). After tilling, the estimated density was 81,330 pellets/ha (0.58 ± 0.04 pellets/core sample) in the top 10 cm of sediment.

Prior to treatments, 52--55% of all pellets were in the top 5 cm of sediment, and 78--84% of all pellets were in the top 10 cm of sediment. The vertical distribution of pellets remained relatively unchanged in the control plots. After tilling, the average number of lead pellets present in the upper 10 cm decreased by 59% in the disked plots and by 70% in the plowed plots. In contrast, the average number of lead pellets detected in the 11--20-cm layers increased by 195% in the disked plots and 260% in the plowed plots. The effective depth of tilling was estimated to be 25 cm for disking and 40 cm for plowing.

The pellet depth and treatment-by-pellet depth interaction terms were significant explanatory variables in the final model (P < 0.001 and P = 0.03, respectively). The number of pellets present at a given depth changed after tilling, and the amount of that change differed between disking and plowing (Table 1). Both disking and plowing resulted in a higher percentage of lead pellets below 10 and 20 cm relative to untilled controls. These tillage practices were effective in redistributing pellets below 10 cm (P < 0.001). There was no difference (P = 0.32) between disking and plowing below 10 cm.

Vegetation Alteration

Tillage altered the maximum height, percent cover, and stem density of wetland vegetation. In the first year after treatment, maximum height of vegetation increased (P < 0.001) from 194 ± 23 mm on control plots to 261 ± 17 mm on tilled plots. Maximum vegetation height decreased in the second year after treatment from 261 ± 17 mm in 1998 to 140 ± 11 mm in 1999 (P < 0.001). The composition of vegetation cover changed (P < 0.01) between treatments and between years (Table 2).

In addition, there were interactions between treatment and year (P < 0.01). In the first year post-treatment (1998), tillage reduced the swamp timothy cover (P = 0.03). In contrast, swamp timothy cover increased during the second year (1999) in both control and treated plots. California loosestrife showed the opposite trend, increasing by more than 3 times in disturbed soils during 1998 (P < 0.01). Percent cover by California loosestrife decreased in both control and treated plots between 1998 and 1999. Nine other species were present infrequently.

After accounting for differences between years, tillage altered the density of swamp timothy (P > 0.001). There was no interaction between year and treatment in plant density (P = 0.42). In the first year post-treatment, swamp timothy was almost 8 times more dense in control plots (95.6 stems/m2) than in treated plots (11.8 stems/m2). Between the first and second years post-treatment, the density of swamp timothy increased 96% in treated plots, from a mean of 11.8 ± 2.0 stems/m2 in 1998 to a mean of 312 ± 10 stems/m2 in 1999. Although loosestrife stem density did not change between years (P = 0.26), its initial stem density was lower (P = 0.02) in control plots than in tilled plots (5.1 versus 8.1 stems/m2).


Effects of Tillage on Lead Pellets

Although the density of lead pellets in Pool 8 during this study ( = 342,999 pellets/ha in the top 10 cm) was lower than the 1986--89 study (>2,000,000 pellets/ha; Rocke et al. 1997), pellet density remained high. Elevated blood lead levels (0.2 ppm) in caged mallards on Sacramento National Wildlife Refuge were linked to sediment pellet densities ranging from 15,750 to 2,299,700 pellets/ha in the top 10 cm of sediment (Rocke et al. 1997). Previous studies have indicated that shallow tillage (<5 cm) was effective in redistributing lead pellets below surface layers of wetland sediment (Fredrickson et al. 1977, Esslinger and Klimstra 1983, Castrale 1989). Deep tillage, however, was more effective than shallow tillage in reducing shot densities near the surface (Peters and Afton 1993).

The upper sediment layers on the Sacramento National Wildlife Refuge are relatively shallow and underlain by hardpan clay. The high density of lead pellets in Pool 8, an area in which lead shot has not been used legally since 1986, indicates that the hardpan clay layer may be preventing the pellets from moving deeper into the sediment profile. This is consistent with Pain (1991) and Mudge (1984), who reported that shot settlement rates depend on sediment composition and plant density, and found that denser sediments (those with a higher percentage of clay) retain more shot.

Both disking and plowing were effective in moving lead pellets deeper into the sediment profile, thereby reducing the amount of shot available to dabbling ducks (<10 cm deep). Only plowing reduced the amount of shot available at depths between 20 and 30 cm. Therefore, plowing was more effective than disking in redistributing lead shot below the zone of availability for most waterfowl, and hence reducing the risk of lead poisoning to waterfowl, especially swans.

We saw changes in the distribution of lead pellets throughout the sediment profile. Therefore, tilling a wetland may be a viable management option to reduce the availability of lead pellets to waterfowl. The outcome of repeated tilling in successive years is unknown. Lead pellets may be moved deeper into the sediment profile, or they may be recycled toward the surface. Because of this potential for recycling (thereby increasing availability of lead pellets), it is important to determine the vertical distribution of lead pellets in wetland sediments prior to tilling. Further research is needed to clarify the effect of repeated tilling on the distribution of lead pellets in wetland sediments.

Effects of Tillage on Vegetation

Tillage may affect the composition and structure of wetland vegetation. Swamp timothy is a short, dense plant that produces abundant seeds and provides good habitat for invertebrates (Fredrickson and Taylor 1982, Severson 1987). Waterfowl feed on both swamp timothy seeds and invertebrate inhabitants. Loosestrife is a taller, somewhat woody plant that produces little waterfowl food in terms of seeds or invertebrates (Fredrickson and Taylor 1982, Severson 1987). During the first year post-treatment in our study, vegetation in undisturbed areas consisted primarily of swamp timothy. In contrast, vegetation in tilled areas was dominated by loosestrife. Typically, this would not be viewed as a positive change in the quality of the wetland habitat for waterfowl. However, these changes were reversed by the second year, indicating that the negative impacts of tilling in this marsh were temporary.

Incidence of Steel Pellets

Steel pellets were recovered infrequently relative to lead pellets. While lead shot may still be deposited illegally, we suggest that the differences in occurrence are largely due to the length of time each shot has been used (legally) on Sacramento National Wildlife Refuge. Hunters on Sacramento National Wildlife Refuge have been required to use nontoxic shot (e.g., steel) since 1986. In contrast, hunters had been using lead shot in the Pool 8 area since the late 1800s (Hall 1975). Due to refuge management practices, hunting pressure and the resulting (nontoxic) shot deposition has been restricted since 1986. Sacramento National Wildlife Refuge has implemented spaced-blind hunting in Pool 8 and a 25 shot shell limit per hunter in the field. Given the nature of the soils underlying Pool 8 (e.g., hardpan) and the persistence of lead pellets in the upper sediment strata, lead and steel pellets may be more equally represented in the future.


In areas with a clay hardpan underlying wetland sediments, tilling may be an effective means of reducing lead shot availability to waterfowl, especially ducks. The temporary reduction in swamp timothy that we observed in the tilled areas can be minimized by tilling a small portion of the total land available for waterfowl each year. Tillage could be used on many wetlands, provided that they have water control that would allow seasonal drainage.


This study was funded by the U.S. Fish and Wildlife Service. We thank J. S. Edwards, C. S. Johnson, J. E. Isola, M. A. Ennis, N. D. Stanley, A. R. de Knijf, and all volunteers for their assistance. W. B. Flournoy and M. L. Stevenson managed and operated the heavy equipment. T. E. Rocke and C. J. Brand, National Wildlife Health Laboratory in Madison, Wisconsin, provided historical data on lead shot distribution and abundance in refuge sediments. M. S. Peters, M. A. Wolder, J. G. Silveira, T. C. Maurer, and S. E. Schwarzbach provided helpful comments on this manuscript.


Castrale, J. S. 1989. Availability of spent lead shot in fields managed for mourning dove hunting. Wildlife Society Bulletin 17:184-189.

Esslinger, C. G., and W. D. Klimstra. 1983. Lead shot incidence on a public goose hunting area in southern Illinois. Wildlife Society Bulletin 11:166-169.

Fredrickson, L. H., T. S. Baskett, G. K. Brakhage, and V. C. Cravens. 1977. Evaluating cultivation near duck blinds to reduce lead poisoning hazard. Journal of Wildlife Management 41:624-631.

-----, and T. S. Taylor. 1982. Management of seasonally flooded impoundments for wildlife. U.S. Fish and Wildlife Service Resource Publication 148.

Hall, F. A., Jr. 1975. An environmental history of the Sacramento National Wildlife Refuge. Thesis, California State University, Chico, California, USA.

Hays, R. L., C. Summers, and W. Seitz. 1981. Estimating wildlife habitat variables. U.S. Fish and Wildlife Service FWS/OBS-81/47.

Jordan, J. S., and F. C. Bellrose. 1951. Lead poisoning in wild waterfowl. Illinois Natural History Survey Biological Notes 26.

Mauser, D. M., T. E. Rocke, J. G. Mensik, and C. J. Brand. 1990. Blood lead concentrations in mallards from Delevan and Colusa National Wildlife Refuges. California Fish and Game 76:132-136.

Mudge, G. P. 1984. Densities and settlement rates of spent shotgun pellets in British wetland soils. Environmental Pollution (Series B):299-318.

Pain, D. J. 1990. Lead shot ingestion by waterbirds in the Camargue, France: an investigation of levels and interspecific differences. Environmental Pollution 66:273-285.

-----. 1991. Lead shot densities and settlement rates in Camargue Marshes, France. Biological Conservation 57:273-286.

Peters, M. S., and A. D. Afton. 1993. Effects of deep tillage on redistribution of lead shot and chufa flatsedge at Catahoula Lake, Louisiana. Wildlife Society Bulletin 21:471-479.

Ramsey, M. A. 1993. Shot ingestion study of drake mallards on Delevan National Wildlife Refuge 1992-1993. Report for U.S. Fish and Wildlife Service, Sacramento National Wildlife Refuge, Willows, California, USA.

Ramsey, F. L., and D. W. Schaffer. 1997. Statistical sleuth: a course in methods of data analysis. Duxbury Press, Belmont, California, USA.

Rocke, T. E., and C. J. Brand. 1994. Use of sentinel mallards for epizootiologic studies of avian botulism. Journal of Wildlife Diseases 30:514-522.

-----, -----, and J. G. Mensik. 1997. Site-specific lead exposure from lead pellet ingestion in sentinel mallards. Journal of Wildlife Management 61:228-234.

-----, and M. D. Samuel. 1991. Effects of lead shot ingestion on selected cells of the mallard immune system. Journal of Wildlife Diseases 27:1-9.

Sanderson, G. C., and F. C. Bellrose. 1986. A review of the problem of lead poisoning in waterfowl. Illinois Natural History Survey Special Publication 4.

Severson, D. J. 1987. Macroinvertebrate populations in seasonally flooded marshes in the northern San Joaquin Valley of California. Thesis, Humboldt State University, Arcata, California, USA.

Soil Conservation Service. 1968. Soil survey of Glenn County area, California. Soil Conservation Service, Davis, California, USA.

StatSoft. 1995. General Conventions and Statistics I. Statsoft. Tulsa, Oklahoma, USA.

U.S. Geological Survey. 1999. Field manual of wildlife diseases: general field procedures and diseases of birds. U.S. Geological Survey, Biological Resources Division. Information and Technology Report 1999-001.

Received 25 April 2000.

Accepted 14 July 2000.

Associate Editor: Rattner.

Table 1. Mean number of lead and steel pellets (± SE) present in sediment samples (2.39 L volume) collected from Sacramento National Wildlife Refuge, Pool 8, 1997.
Treatment Stratum (cm) Lead Steel
Pre-treatment Post-treatment Pre-treatment Post-treatment
Control 0--5 105 ± 0.13 97 ± 0.11 16 ± 0.04 16 ± 0.04
6--10 53 ± 0.09 33 ± 0.07 1 ± 0.01 0
11--15 10 ± 0.03 11 ± 0.04 0 0
16--20 4 ± 0.02 6 ± 0.03 1 ± 0.01 0
21--25 10 ± 0.05 13 ± 0.04 0 1 ± 0.01
26--30 6 ± 0.03 4 ± 0.02 0 0
188 ± 0.18 164 ± 0.14 18 ± 0.04 17 ± 0.04
Plow 0--5 119 ± 0.13 27 ± 0.05 19 ± 0.04 3 ± 0.02
6--10 50 ± 0.09 23 ± 0.06 0 1 ± 0.01
11--15 11 ± 0.04 34 ± 0.06 0 1 ± 0.01
16--20 9 ± 0.04 38 ± 0.07 1 ± 0.01 2 ± 0.02
21--25 17 ± 0.04 23 ± 0.05 1 ± 0.01 0
26--30 10 ± 0.039 12 ± 0.04 0 0
216 ± 0.17 157 ± 0.15 21 ± 0.04 7 ± 0.04
Disk 0--5 104 ± 0.12 36 ± 0.08 10 ± 0.03 3 ± 0.03
6--10 54 ± 0.11 29 ± 0.05 0 3 ± 0.03
11--15 10 ± 0.03 34 ± 0.07 0 1 ± 0.01
16--20 10 ± 0.04 25 ± 0.05 0 3 ± 0.03
21--25 13 ± 0.04 17 ± 0.05 2 ± 0.02 3 ± 0.03
26--30 10 ± 0.03 7 ± 0.03 0 2 ± 0.02
201 ± 0.17 148 ± 0.15 12 ± 0.05 15 ± 0.03
a Strata are 5-cm increments of a 30-cm sediment core, n = 100 for each stratum for each treatment.
Table 2. Effects of tillage on vegetation cover (%) in Pool 8, Sacramento National Wildlife Refuge, 1998--1999.
Species Untilled (n = 25) Tilled (n = 50) a
1998 1999 1998 1999
Swamp timothy 48.8 ± 12 60.5 ± 4 18.1 ± 3 68.8 ± 2
Loosestrife 12.0 ± 3 0.8 ± 0.3 39.9 ± 4 1.9 ± 0.6
Barnyard grass 0.0 0.2 ± 0.1 1.0 ± 0.4 0.4 ± 0.1
Sprangletop 0.0 0.0 1.4 ± 0.4 0.5 ± 0.1
Dwarf spikerush
Eleocharis parvula
18.2 ± 4.7 0.0 26.7 ± 3.4 0.1 ± 0.1
Bermuda grass Cynodon dactylon 11.3 ± 4.2 4.4 ± 2.6 4.9 ± 2.6 1.1 ± 0.3
Burhead Echinodorus berteroi 1.0 ± 0.4 0.0 2.5 ± 0.8 0.0
Bare ground 7.2 ± 2.8 28.1 ± 3.4 4.9 ± 1.2 13.6 ± 1.7
Otherb 0.8 ± 0.6 0.0 0.4 ± 0.1 0.4 ± 0.2
a Tillage occurred in 1997. b Redstem (Ammania spp.), Spikerush (Eleocharis macrostachya), Cudweed (Gnapifolium palustre), and Juncas (Juncas balticus).