220 C. Stoate et al. Agriculture, Ecosystems and Environment 77 2000 219–226
In these extensive arable systems, grazed fallow pro- vides a source of food in the form of seeds in winter
and invertebrates in summer. Cereal crops also sup- port an invertebrate community with relatively large
insects, such as caterpillars Lepidoptera larvae and grasshoppers Orthoptera, taken by corn buntings dur-
ing the breeding season in the north of their range Brickle and Harper, in press. This study assesses the
abundance of corn buntings in December and April in relation to three arable systems and other environ-
mental variables in an agricultural landscape of Baixo Alentejo, southern Portugal, considering, in particu-
lar, various levels of agricultural intensification, and the potential effects of fallow area and private game
interests.
2. Methods
2.1. Study area The study area included parts or all of five ad-
ministrative regions in Baixo Alentejo Ferreira do Alentejo, Aljustrel, Castro Verde, Ourique and
Almodôvar, with a total area of 155,000 ha. Within this region, three land-use systems were recog-
nised: intensive agriculture, extensive agriculture and ‘montado’ Table 1.
The intensive agriculture category is characterised by a greater frequency 55 of heavy soils, much
of the area being irrigated. Wheat Triticum aestivum and barley Hordeum distichum are the main cereal
crops, and silage grass Lolium sp., sunflower He- lianthus annuus, sugar beet Beta vulgaris and oilseed
rape Brassica napus are also grown. Wheat yields are 2.5–3.5 tonnes ha
− 1
without irrigation, but can be almost doubled with full irrigation P. Eden, pers.
comm., 1998. There are short rotations with little or no fallow e.g., sunflower-1st cereal-2nd cereal. This
system requires frequent use of fertiliser 130 units of N
2
per hectare P. Eden, pers. comm., 1998 and her- bicides relative to the other land-use categories. With
the exception of some olive Olea europea groves, there is little tree cover.
The extensive agriculture category is characterised by thin soils and the largest average farm size of
the three categories Table 1. There is no irriga-
Table 1 Agricultural statistics for the three land-use categories considered
in Alentejo, Portugal source: Cordovil, 1993. Intensive
Extensive Montado Mean farm size ha all farms
48 161
66 Small farms
10 24
17 Medium farms
69 151
154 Large farms
310 519
438 Crop area
Total annual crops 81
42 28
Winter cereals 45
40 21
Sunflower 20
0.3 Forage crops
6 2
7 Fallow
15 52
66 Perennial crops
11 0.8
2 Olive
10 0.8
2 Vines
1 Land area per tractor ha
58 125
194 Livestock
Sheep 68
78 83
Cattle 8
11 7
Pigs 22
8 5
Goats 2
3 5
tion, and fallow area is relatively high. A typical rotation takes the form: plough fallow-1st cereal-2nd
cereal-fallow-fallow, with fallow periods often last- ing five years or more Rio Carvalho et al., 1995.
Wheat yields are 1.5–2.5 tonnes ha
− 1
, with yields at the lower end of this range being more common P.
Eden, pers. comm., 1998. Triticale Triticum aestivum ×
Secale cereale and oats Avena sativa are frequently grown in the extensive category, and grazed or cut
for silage. The incorporation of a fallow period into the rotation, and the relatively low potential yields
are associated with considerably lower annual inputs than in the intensive category.
‘Montado’ equivalent to the Spanish ‘dehesa’ is characterised by thin soils and tree cover, dominated
by holm oak Quercus rotundifolia and cork oak Q. suber. Like the extensive category, there is no irriga-
tion and the fallow area is high. A typical rotation is similar to that of the extensive category, although the
fallow stage is often longer and forage lupins Lupinus luteus may be included.
Sheep Ovis aries, cattle Bos taurus and pigs Sus scrofa are kept in all three land-use categories. Zero
grazing is adopted on some farms in the intensive cat-
C. Stoate et al. Agriculture, Ecosystems and Environment 77 2000 219–226 221
egory, but livestock normally graze fallows. Table 1 lists the proportion of crops and livestock in each cat-
egory. Public hunting of wild game is open to all li- censed hunters over much of the area, with no control
on the game bags taken and without any management. However, some areas are managed as private or asso-
ciative game estates, implying a control on game bags and some management measures.
2.2. Field methods A total of 115–250 m transects, starting in 1 km
grid intersections and stratified by land-use categories, were walked along a random bearing. Transects were
walked in the first three hours after dawn, or in the two hours before dusk, in December and April, from De-
cember 1994 to April 1997. Perpendicular distances from the transect line to each detected bird were es-
timated visually to the nearest 10 m. The number of birds seen together at an observation was recorded.
Tree density estimates were based on counts within a belt 25 m each side of the transect line. After each
spring count, the crop type and vegetation structure was sampled over the first 50 m of each transect, by
plotting vegetation height on millimetric paper, and proportional cover was determined for each of seven
height classes Table 2. A Shannon diversity index of these cover classes was computed for each transect.
Table 2 gives sources of other environmental variables. In April 1997, broad-leaved weed cover within cereal
crops in extensive and intensive categories was esti- mated within a 4 m
2
quadrat at 50 m intervals along each transect. Lepidoptera larvae and Orthoptera were
sampled using one sample of ten sweeps of approxi- mately 3 m amplitude along the start of each transect,
using a 50 cm diameter net.
2.3. Analytical methods 2.3.1. Density estimates
Density estimates were calculated for each land-use category using line transect sampling and the com-
puter program DISTANCE Buckland et al., 1993; Laake et al., 1993. For the data gathered in Decem-
ber, when birds were in flocks, flock size mean 4.2, range 1–40 was accommodated within the analysis,
and a hazard rate model with cosine adjustments was selected by the program. In April, flocks were not
present and individuals or pairs seen together were recorded as single units. A half-normal model with
cosine adjustments was selected by the program and applied to the April data. Analyses used i clusters
≥1 individual as analytical units, ii ungrouped data, and iii untruncated perpendicular distance
data. A variety of recommended robust estimators im- plemented by DISTANCE was used, the final model
selected in each case being the one with the lowest Akaike’s Information Criterion value Buckland et
al., 1993. Differences in log-transformed density estimates weighed by 1variance between years
and land-use categories were tested using one-way ANOVA.
2.3.2. Environmental models In order to evaluate the main environmental factors
affecting corn bunting abundance in the study area, two models winter and spring of corn bunting rela-
tive abundance were computed using multivariate re- gression and the number of buntings detected in each
transect as the dependent variable. As this could have been affected by differential habitat visibility between
land-use categories, it was checked before analysis whether there was a significant correlation between
the visibility-corrected line transect density estimates and the mean number of buntings detected in each
surveyfarming system, using: i all cases n = 18, and ii only significantly different line transect es-
timates n = 6. These correlations were highly sig- nificant r
s
= 0.92, p 0.001, in the first case; r
s
= 1,
p = 0, in the second, indicating that the number of buntings detected per transect and the line transect es-
timates had similar value as indices of relative abun- dance. Before analysis, the dependent variable was
logarithmically transformed [A = logx + 1] to nor- malise the distribution of residuals and equalise the
variances Zar, 1984. Tables 2–4 list the independent variables considered and the sources of data. The data
were incorporated and manipulated in a vector-based Geographic Information System GIS-ArcCAD.
A two-stage analysis was followed to generate the models, with a stepwise procedure in each stage. At
the first stage, independent variables constant in each transect between years see Table 2 and the logarith-
mically transformed between-year averages A
′
of the number of corn buntings detected per transect were
222 C. Stoate et al. Agriculture, Ecosystems and Environment 77 2000 219–226
Table 2 Independent variables considered in the environmental models of corn bunting relative abundance
Variables Sources of data
Game management: managed or unmanaged site
a
Field work Percentage of vegetation cover on the first 50 m of
each transect, for the following height classes: 0–20 cm, 20–40 cm, 40–60 cm, 60–80 cm, 80–100 cm, 100–400 cm,
400–800 cm Field work
Structural diversity of vegetation cover: Shannon diversity index of above Field work
Proportion of each land-use class along the transects; the land uses considered were plough, fallow, wheat, oats,
rye, lupin, sunflower, and olive groves Field work
Tree density treesha
a
Field work Farming system: montado, extensive agriculture, and intensive agriculture
a
Field work, ERENA 1994 Distance to the nearest water line m
a
1 : 25,000 topographic maps, GIS Distance to the nearest dirt track m
a
1 : 25,000 topographic maps, GIS Distance to the nearest road m
a
1 : 25,000 topographic maps, GIS Average annual air humidity at 9 T.M.G.
a
Serviço Meteorol´ogico Nacional 1975a Average annual rainfall mm
a
Serviço Meteorol´ogico Nacional 1975b Average annual temperature
◦
C
a
Serviço Meteorol´ogico Nacional 1975c Total length of field edges within a 500 m buffer around
the starting point of the transect m
a
1 : 30,000 aerial photographs of 1993, GIS Diversity of land uses Shannon index within a 500 m
buffer around the starting point of the transect
a
1 : 30,000 aerial photographs of 1993, GIS
a
Variables with constant values within sites between years. Table 3
Means ± standard errors of the independent variables constant in each 250 m transect between years Variables
Montado Extensive agriculture
Intensive agriculture n = 42
n = 42 n = 31
Game managed sites 33.3
29.3 34.5
Tree density treesha 10.54 ± 0.70
0.36 ± 0.32 0.39 ± 0.15
Distance to water line m 88.48 ± 10.27
81.63 ± 9.80 102.21 ± 16.12
Distance to dirt track m 144.34 ± 17.19
161.50 ± 20.42 133.62 ± 22.24
Distance to road m 701.62 ± 88.98
870.02 ± 90.37 612.67 ± 86.38
Average air humidity 76.79 ± 0.37
75.90 ± 0.31 79.12 ± 0.26
Average rainfall mm 60.95 ± 0.46
59.74 ± 0.26 63.55 ± 0.87
Average temperature
◦
C 16.14 ± 0.08
16.67 ± 0.15 16.04 ± 0.05
Length of field edges within a 500 m buffer m 3455.74 ± 136.54
1344.61 ± 155.48 1303.70 ± 185.70
Diversity of land uses within a 500 m buffer Shannon index 0.39 ± 0.02
0.20 ± 0.02 0.18 ± 0.03
Table 4 Means ± standard errors of the independent variables not constant in each 250 m transect between years
Variables Montado
Extensive agriculture Intensive agriculture
Winter Spring
Winter Spring
Winter Spring
Cover diversity Shannon index –
0.67 ± 0.04 –
0.40 ± 0.04 –
0.33 ± 0.05 Proportion plough
0.04 ± 0.02 0.03 ± 0.02
0.04 ± 0.02 0.03 ± 0.01
0.49 ± 0.05 0.25 ± 0.05
Proportion fallow 0.76 ± 0.04
0.80 ± 0.03 0.70 ± 0.04
0.71 ± 0.04 0.05 ± 0.02
0.06 ± 0.02 Proportion cereal
0.22 ± 0.04 0.19 ± 0.03
0.27 ± 0.04 0.28 ± 0.04
0.39 ± 0.04 0.68 ± 0.05
Proportion lupin –
0.01 ± 0.01 –
– –
– Proportion sunflower
– –
– –
– 0.04 ± 0.02
Proportion olive groves –
– –
– 0.04 ± 0.02
0.04 ± 0.02
C. Stoate et al. Agriculture, Ecosystems and Environment 77 2000 219–226 223
considered exclusively. Variables entering the mod- els winter and spring were selected through forward
stepwise selection. In the second stage, 114 transects and two-year dummy variables were considered to
account for between-site and between-year variation, and the remaining independent variables that changed
between years within each transect. The dummy vari- ables were entered at step 0, and the stepwise selection
proceeded from there. Once the final models of this stage were determined, the dummy variables were re-
placed by the variables selected in stage 1, and ‘nested’ models with all the non-dummy variables selected in
the two stages were fitted. Finally, it was checked whether the nested models successfully explained the
between-transect and between-year variation by com- paring the models with the dummy variables model
1 with the nested models model 2, using an F test, calculated as: model 1 sum of squares − model 2 sum
of squaresmodel 1 df − model 2 dfmodel 1 resid- ual sum of squaresmodel 1 residual df, the degrees
of freedom being equal to model 1 df − model 2 df, model 1 residual df.
A correlation matrix of the selected variables was generated for the final winter and spring models.
Additionally, significant differences between land-use categories for the variables selected were determined
using one-way ANOVAs for the independent contin- uous variables and chi-square tests for the categorical
ones.
3. Results