Northern Prairie Wildlife Research Center
For the study, we defined 3 types of farming practices. Conventional farming included tillage in spring before seeding, fall tillage after harvest, and use of synthetic chemical herbicides and tillage to control weeds in crops and in fallow. Minimum-tillage farming included the placement of seed into plant residue that was untilled or moderately tilled. Fallow fields in minimum-tillage farming systems usually were small grain stubble, where weeds were controlled by herbicides. Organic farming excluded synthetic chemicals for weed and pest control, instead using cultivation and crop rotation. Yellow sweet clover (Melilotus officinalis), a legume, was tilled into the soil on organic fallow fields to add nitrogen.
Conventionally farmed fields were identified from information given in county offices of the U. S. Department of Agriculture. We obtained nearly complete lists of fields farmed by minimum-tillage and organic methods from the alternative farming organizations. We visited potential cooperators, and those who had one or more of the focal crops and who granted access, were included in the study. Each year we attempted to survey 32 ha of each crop on 10 fields of each type. Fields, not farms, were used as the experimental units because most North Dakota farms include land that is spatially separated. Fields were selected randomly within farms and new fields were selected each year.
We used the term "hatching success" to indicate the percentage of nests that hatched at least 1 egg and the term "fledging success" to indicate the percentage of nests with young that produced at least 1 fledgling. We used the term "nest success" for passerines to indicate the percentage of nests with eggs that produced at least 1 fledgling.
Bird Counting Techniques
Birds were counted on fields once or twice each season, beginning in April 1991 and ending in July 1993. We selected the dates of the bird surveys to portray the spring and fall migration, breeding, and wintering periods in North Dakota. Spring surveys extended from 1 March to 20 April, summer surveys from 21 April to 2 July, fall surveys from 15 August through 30 November, and winter surveys from 2 December to 2 February. Each field survey was separated by a period of at least 30 days.
Birds were surveyed by individuals walking truncated line-transects (Burnham et al. 1980). Crop rows or furrows were used as transect lines. Birds were counted within 50 m on each side of the transect line. Surveys were conducted from sunrise to 1300 hours (Dawson 1981). Summer surveys were not conducted when it was raining, the temperature was less than 0 C or higher than 32 C, or the wind speed exceeded 30 km/hour (Dawson 1981). Spring, fall, and winter surveys were not conducted if it was raining or snowing, the temperature was less than -12 C or more than 32 C, or the wind speed exceeded 30 km/hour.
Only birds observed on the ground in the field were counted. Birds flying over fields or those using other habitats, such as wetlands or fencerows, were not counted. Also, we estimated snow depth on the fields during the bird surveys.
Estimating Bird Densities
We attempted to use techniques described by Buckland et al. (1993) to estimate the undetected proportion of birds in fields. Due to small sample sizes and large flock sizes we were not able to employ distance sampling methods. Instead we used the observed bird density to test for differences among the years, crops, and field types.
Nest Searching Techniques
Each field was searched for bird nests 3 times during the nesting season. Two complete searches were made with a 40-m chain flushing device pulled between 2 all-terrain vehicles (Klett et al. 1986). The first nest search was initiated mainly before plant growth in late April and early May. The second search was conducted mainly after plants were growing from mid-May to early July. A third visual, walking search was made in late July to locate nests of late-nesting red-winged blackbird (Agelaius phoeniceus). The location of each nest was plotted on a map, and a marker flag was set 4 m north of the nest.
The number of host eggs, number of parasitic eggs, and the incubation or nestling stage was recorded during each nest visit. To determine incubation stage, we developed a technique to estimate days of incubation by examining the eggs with candlers (Lokemoen and Koford 1996). Fledging stage (days since hatching) was estimated from information published on the development of mourning doves (Zenaida macroura, Hanson and Kossack 1963) and chestnut-collared longspurs (Calcarius ornatus, Bent 1968). Nests were monitored every 7 days and hatching success was calculated by the Mayfield method (Mayfield 1975), with modifications described by Johnson (1979).
Daily Survival Rates
We used daily survival rate (DSR) calculations to estimate hatching success, fledging success, and nest success for passerines, and hatching success for shorebirds and waterfowl. To increase the number of exposure days, nests were pooled for all years to form a single DSR for all combinations of crop and field type. Comparisons among the field types, crops, and the crop-field type interaction were made with the Chi-square test described by Sauer and Williams (1989) and implemented by the Interactive Matrix Language (IML) procedure (SAS Inst. Inc. 1989b). Testing was done sequentially. First, we tested the interaction between the field type and crop. If the interaction was not significant, we tested separately for crop and field type effects. A non-significant field type-crop interaction implies that the effect of one main effect (crop or field type) on DSR is the same regardless of the level of the other main effect (field type or crop).
Plant Cover and Field Treatment Measurements
Two assessments of vegetative cover were made at each field in the spring. Visual obstruction readings of plant density were estimated from 4 readings of a calibrated pole made at 25 stations distributed across each field (Kirsch et al. 1978). Plant canopy cover, percent dead litter, and plant height were also evaluated at each station as the means of 4 estimates from each quarter of the station with a Daubenmire (1959) frame.
We recorded the number and time of tillages performed on each field each year because these treatments physically altered vegetative cover used by birds. We subsequently examined the relation between the number of tillage treatments on fields with nest densities and nest success using Pearson product-moment correlations.
We used 3-way analysis of variance (ANOVA) techniques (Milliken and Johnson 1984) to assess the effects of years, crops, field types, and all interactions on bird densities (no./ha) by season, nest densities (no./ha), plant cover, and litter cover. Sample sizes varied because several fields had missing data. Unless noted, all analyses were conducted with General Linear Models (GLM) procedure (SAS Inst. Inc. 1989a). The population marginal means (or least squares means, LSM) are reported for all analyses. We used α ≤ 0.05 to indicate significance.
There were no conventional fallow, minimum-tillage fallow, or minimum-tillage sunflower fields surveyed in the 1991 field season. Therefore, the year-crop-field type (3-way) interaction could not be tested for all 3 years. The 3-way interaction was partially tested by analyzing the data as an expanded 1-way model (Milliken and Johnson 1984:177) and with a contrast statement to test the interaction in 1992 and 1993. If this 3-way interaction was not significant, we tested for main effects (field type, crop, and YR) and 2-way interactions (YR-crop, YR-field type, and field type-crop) using all 3 years of data with ANOVA. The non-significance of 2-way and 3-way interactions indicates that the differences in a main effect are consistent among all levels of the other main effects. If interactions are significant, differences in a main effect must be interpreted within the levels of the other effect. If significant effects were found, we used Fisher's least significant differences (LSD) procedure to isolate the differences (Milliken and Johnson 1984).
The number of species observed in summer and the number of nesting species were analyzed with analysis of covariance (ANCOVA) (Milliken 1984). The natural logarithm of the size of the field was used as the covariate. Because there were 3 missing combinations of crop and field type in 1991 and only 1 fallow field was surveyed in 1991, only 1992 and 1993 data were used in these analyses. The LSM are reported as the average value of the covariate (27 ha).
For the analysis of the bird densities in fall, winter, and spring we had data for only 2 years. We tested for crop, field type, and the crop-field type interaction with an expanded 1-way model (Milliken and Johnson 1984:177) and used contrast statements to test for interactions and main effects in 1992 (1993 for spring). To stabilize variances bird densities were logarithmically transformed with ln(y + l).
Birds have different preferences for the amount of vegetative cover (Stewart 1975). Recognizing these species differences, we separated the birds into 2 groups for correlations between density (bird or nest) and cover (plant or litter). One group included those species that prefer sparse cover, and the other included those that prefer dense cover.