196 F.D. Menalled et al. Agriculture, Ecosystems and Environment 77 2000 193–202
3. Results
3.1. Landscape characterization The selected simple and complex agricultural
landscapes differed in several key structural vari- ables. The simple landscape contained more farmland
71.4 versus 59.4 and slightly less decidu- ous habitat 11.2 versus 14.3 than the com-
plex landscape. In the 3.2 km
2
representative areas of the simple and complex landscape there were
a total of 61 and 139 crop fields, respectively. A random sample of 30 fields per landscape revealed
that fields in the complex landscape had statisti- cally smaller area x ± 1 SD, simple: 12.4 ± 10.3 ha;
complex: 3.4 ± 3.1 ha; t = 4.62, p 0.0001, less perimeter x ± 1 SD, simple: 1638 ± 720 m; com-
plex: 776 ± 303 m; t = 6.04, p 0.0001, shorter dis- tance to field edge x ± 1 SD, simple: 101 ± 54 m;
complex: 63 ± 39 m; t = 3.16, p = 0.003, and a longer perimeter of wooded field edge per unit of
field area x ± 1 SD, simple: 8.7 ± 19.1 m; complex: 23.5 ± 24.4 m; t = −2.61, p = 0.012, than those lo-
cated within the simple landscape. Also, whereas the complex landscape had fields surrounded primarily by
wide hedgerows mean of field perimeter ± 1 SE, simple: 3.5 ± 1.7; complex: 18.7 ± 5.2; t = 3.16,
p 0.01, those from the simple landscape were en-
Table 1 Results of the nested ANOVA for the first trial testing the effect of landscape structure, fields nested within landscape, treatment, and
species on the percentage post-dispersal seed removal
a
Source of Variation df
SS F
p Landscape
1 613271.8
6.81 0.0349
Species 3
102029.9 1.24
0.3252 Treatment
2 15526652.9
140.43 0.0001
Landscape x Species 3
159826.8 1.94
0.1595 Landscape x Treatment
2 206493.9
1.87 0.1967
Species × Treatment 6
648959.1 4.85
0.0010 Landscape × Species × Treatment
6 238625.2
1.79 0.1299
FieldLandscape 7
644987.4 3.51
0.0015 Treatment × FieldLandscape
12 663993.0
2.10 0.0185
Species × FieldLandscape 18
500278.9 1.06
0.3990 Species × Treatment × FieldLandscape
36 802049.7
0.85 0.7158
Error 184
4837075.5
a
Landscape was tested using FieldLandscape as the error term, Species and Landscape × Species were tested using Species × FieldLandscapes as error term, Treatment and Landscape × Treatment were tested using Treatment × FieldLandscapes as the error term,
and Species × Treatment and Landscape × Species × Treatment were tested using Species × Treatment × FieldLandscape as the error term. Remaining terms were tested using the residual error term.
compassed mainly by herbaceous roadsides mean of field perimeter ± 1 SE, simple: 9.4 ± 2.4; complex:
2.4 ± 1.5; t = −2.48, p 0.05, and crops mean of field perimeter ± 1 SE, simple: 28.4 ± 5.0; complex:
13.4 ± 3.9; t = −2.38, p 0.05.
3.2. Seed removal in crop fields In both the landscapes and for all seed species, there
was a significant difference among exclosure treat- ments Tables 1 and 2. This indicates that cages effec-
tively reduced seed removal and that there was consid- erable post-dispersal weed seed removal by both verte-
brate and invertebrate seed predators. Percentage seed removal was highest in the no exclosure treatment,
intermediate in the vertebrate exclosure treatment, and lowest in the vertebrate + invertebrate exclosure
treatment Fig. 1. Analysis of the first and second trial showed no differences in the rate of seed removal
among the four weed species Tables 1 and 2. Al- though it was possible to observe a tendency towards
higher removal rates in the complex landscape than in the simple landscape Fig. 2, only in the first trial
were these differences statistically significant Tables 1 and 2. For both the trials, it was possible to detected
significant differences among fields suggesting that within the simple and complex landscapes, fields may
support different numbers of beneficial vertebrates
F.D. Menalled et al. Agriculture, Ecosystems and Environment 77 2000 193–202 197
Table 2 Results of the nested ANOVA for the second trial testing the effect of landscape structure, fields nested within landscape, treatment, and
species on the percentage post-dispersal seed removal Source of Variation
a
df SS
F p
Landscape 1
447105.9 2.01
0.1988 Species
3 132286.1
1.44 0.2649
Treatment 2
97008933.1 94.05
0.0001 Landscape × Species
3 66988.9
0.73 0.5487
Landscape × Treatment 2
152616.3 1.48
0.2667 Species × Treatment
6 465955.3
2.30 0.0558
Landscape × Species × Treatment 6
149155.9 0.74
0.6243 FieldLandscape
7 1553891.3
6.75 0.0001
Treatment × FieldLandscape 12
619394.7 1.57
0.1033 Species × FieldLandscape
18 552312.4
0.93 0.5389
Species × Treatment × FieldLandscape 36
1216744.2 1.03
0.4336 Error
186 6112466.5
a
See Table 1 for explanation of the statistical tests.
and invertebrates Fig. 3. Moreover, in both anal- yses there were significant species × treatment and
treatment × field interactions Tables 1 and 2. An analysis of the differential seed loss in the exclosure
treatments suggests that invertebrates and vertebrates had no preferences for removing different species
Fig. 4. Fig. 4 also suggests that the significant species × treatment and treatment × field interactions
were the result of lower recovery rates of D. san- guinalis and A. retroflexus seeds from the verte-
Fig. 1. Percentage seed removed per day mean ± 1 SE per exclusion treatment averaged across field, species and landscape type.
brate + invertebrate exclosure treatment. The small size of these two species seeds may have affected
recovery efficiency.
4. Discussion