262 S. Lapointe et al. Agriculture, Ecosystems and Environment 78 2000 261–272
gen, 1984. Ducks now have to nest in the remaining small and fragmented habitats where predation is often
higher than in contiguous habitats Clark and Nudds, 1991; Pasitschniak-Arts and Messier, 1995. Even
though dabbling duck populations have less declined in eastern than in western North America, biologists
are concerned with the rate at which wetlands and their surroundings are being converted into farmlands and
industrial lands Anonymous, 1986. They are there- fore seeking ways to overcome loss of natural habi-
tats. Island construction is one of the most productive techniques because ducks can nest at higher densities
and with better success attributed to lower mammalian predation Giroux, 1981; Duebbert, 1982. However,
island construction is expensive and could be inappro- priate where duck populations are low Lokemoen,
1984; Bélanger and Tremblay, 1989. Managing ex- isting islands may therefore be a more cost-efficient
strategy Lokemoen and Woodward, 1992.
In Quebec, the most productive areas for nest- ing dabbling ducks are located on islands of the St.
Lawrence river. More than 200 islands ranging in size from 0.1 to 2000 ha and representing nearly
5000 ha of land are found between Montreal and Trois-Rivieres. However, the value of these islands
for ground-nesting birds is often limited by cattle grazing Bélanger and Lehoux, 1995; Bélanger and
Picard, 1999. This reduces the screening effect of vegetative cover, which can lower nest density and
success Lokemoen et al., 1990; Gilbert et al., 1996; Kruse and Bowen, 1996. Grazing can also affect
residual vegetation, an important component of nest- ing cover for early nesting ducks Duebbert, 1969;
Kirsh, 1969; Kirsh et al., 1978. It may provide ideal temperature and humidity for better egg hatchability
Francis, 1968; Duebbert, 1969. In addition, graz- ing and trampling of shoreline vegetation decrease
over-water nesting Krapu et al., 1979; Kirby et al., 1992. Finally, trampling of eggs by cattle directly
affects nest success Jensen et al., 1990.
Studies have shown that it is possible to reduce the adverse effects of cattle with rotational grazing
systems Gjersing, 1975; Barker et al., 1990. Pas- tures are divided into smaller range units and cattle
are periodically moved among these units Kie et al., 1994. Rotation of cattle throughout the growing sea-
son maintains plants at a vegetative stage that pro- vides the most digestible forage Conrad and Martz,
1985. Thus, higher stocking rates are possible and calf weight gain per hectare is better than with contin-
uous grazing Barker et al., 1990. Furthermore, this leaves more undisturbed cover for nesting waterfowl
because less area is required for cattle.
The establishment of dense nesting cover DNC has also been proposed to enhance duck productivity
because it generally supports higher nest densities and success Duebbert and Kantrud, 1974; Klett
et al., 1988; but see McKinnon and Duncan, 1999 for different results. Improved nesting cover is easy
and less expensive to establish and can be sown on islands where cover is inadequate Lokemoen, 1984;
Willms and Crawford, 1989. Several years, however, may be needed before cover becomes adequate for
nesting Livezey, 1981.
The effectiveness of rotational grazing systems and cover improvements on nesting waterfowl has been
generally demonstrated in the mid continent of North America. This area is characterized by productive wet-
lands called prairie potholes that support large num- bers of breeding ducks Bellrose, 1979. On the other
hand, little is known about the effects of these tech- niques in eastern North America where wetlands are
often located along rivers or coastal shores and where breeding populations of ducks are less numerous than
in the mid continent. Moreover, growing conditions are different between the dryer prairie region and east-
ern North America and this could affect the estab- lishment of seeded vegetation. The objective of this
study was therefore to evaluate the effectiveness of ro- tational grazing systems and cover improvements to
enhance nesting habitat for waterfowl in eastern North America. More specifically, the aims were to com-
pare different grassland management practices on 1 quality of nesting cover, 2 nest density of dabbling
ducks and 3 their nesting success on islands of the St. Lawrence river in southern Quebec.
2. Material and methods
2.1. Study area The study was conducted between 1992 and 1994
on four adjacent islands near Varennes, Quebec, Canada 45
◦
40
′
N, 73
◦
27
′
W, 16 km northeast of Montreal along the St. Lawrence river. The climate is
S. Lapointe et al. Agriculture, Ecosystems and Environment 78 2000 261–272 263
continental temperate, with a mean annual tempera- ture of 6.1
◦
C and a total annual rainfall of 768 mm. The islands ranged in size between 9.4 and 59.8 ha for
a total of 111.5 ha. No trees or shrubs were present. As water levels recede following spring run-off, the
islands are joined together forming small interior marshes with stands of cattail Typha angustifolia L.,
big burreed Sparganium eurycarpum Engelm and arrowhead Sagittaria spp..
Cattle were brought to the islands by boat in late May to early June and removed in November. A to-
tal of 114, 100 and 85 cows were present in 1992, 1993 and 1994, respectively. During the first year,
cattle grazed everywhere on the islands Fig. 1.
Fig. 1. Location of the treatments on islands at Varennes, Quebec, 1992–1994.
Ile-aux-Fermiers, Masta and St. Patrice islands were dominated by red-top Agrostis alba L., red
fescue-grass Festuca rubra L., Kentucky bluegrass Poa pratensis L. and cow vetch Vicia cracca
L.. Grande-Ile was covered with reed canary grass Phalaris arundinacea L. and Canada reed-grass
Calamagrostis canadensis [Michx.] Beauv..
In August 1992, Ducks Unlimited Canada seeded DNC, improved the quality and quantity of forage of
some pastures and set up a series of electric fences to establish a rotational grazing system. The forage
on Grande-Ile was improved by sowing 19.2 ha of timothy Phleum pratense L., sweet clover Melilo-
tus officinalis [L.] Desr., smooth brome Bromus
264 S. Lapointe et al. Agriculture, Ecosystems and Environment 78 2000 261–272
inermis Leyss. and clover Trifolium spp.. Mixtures of reed canary grass in association with timothy,
tall fescue-grass Festuca elatior L., Orchard grass Dactylis glomerata L. or tall wheat grass Agropy-
ron elongatum [Host] Beauv. were sown on 5 ha on Grande-Ile to establish the DNC. Non-selective
herbicide glyphosate was applied to treated fields before cultivation. Masta, St. Patrice as well as parts
of Grande-Ile and Ile-aux-Fermiers were left idle with no grazing. Interior marshes were also protected from
cattle by permanent or temporary fences.
In 1993, a rotational grazing system was estab- lished Fig. 1. From the end of May to mid July,
cattle were restricted to the improved pasture. After the duck nesting season, they were moved to the
unimproved pasture on Ile-aux-Fermiers until mid September. Another rotation between these two pas-
tures took place between mid September and the end of October. Finally, cattle were allowed to graze
freely on all islands including in DNC for 2 weeks until they were removed in mid November. During the
duck nesting period from May to July, there were four treatments: idle field 59.9 ha, improved pasture with
cattle 19.2 ha, unimproved pasture without cattle 27.4 ha and a 1-year old DNC DNC93: 5.0 ha.
In the fall of 1993, another 9.4 and 5.6 ha of DNC were established on St. Patrice and Grande-Ile, re-
spectively, by sowing Western wheat grass Agropyron smithii Rydb. and crested wheat grass Agropyron
cristatum[L.] Gaertn.. A portion of Ile-aux-Fermiers was ploughed so it could be converted to improved
pasture but was not seeded on time. This pasture was therefore a ploughed field 17.6 ha during the 1994
duck nesting season. The other treatments included: idle field 39.1 ha, improved pasture with cattle
19.2 ha, unimproved pasture without cattle 15.6 ha, a 1-year old DNC DNC94: 15.0 ha and a 2-year
old DNC DNC93: 5.0 ha. Rotation of cattle among pastures in 1994 was similar to the previous year.
2.2. Cover evaluation In 1992, the effect of cattle grazing on the vege-
tation was evaluated by establishing 50 0.5 m × 1.0 m exclosures that were randomly located on the four is-
lands and put in place before the arrival of cattle. In July, all the vegetation above 1 cm was clipped in a
10 cm × 10 cm randomly selected quadrat within each exclosure and in grazed sites, also located randomly
at more than 50 m away from the exclosures. Vegeta- tion was sorted into live biomass defined as any green
plants or plant parts, and residual vegetation that con- sisted of dead biomass from the previous and current
years. Vegetation was then dried to constant weight in a micro-wave oven Bilanski and Ghate, 1978 and
weighed to the nearest 0.1 g. Data from unfenced plots were also used to compare live and dead biomass of
vegetation among treatments before and after the man- agement. In 1993, the same 50 plots were used while
in 1994 an additional 39 plots were randomly located, for a total of 89. Sampling was conducted during the
first week of July each year, using the same procedure.
In 1994, visual obstruction readings were taken ev- ery 10 m along a 100 m NE–SW line in each plot using
a Robel et al. 1970 pole. Readings were taken at a distance of 4 m from a height of 1 m. These readings
give an index of cover quality by taking into account height and density of vegetation; it is highly corelated
to biomass Robel et al., 1970. Depth of residual veg- etation accumulated on the ground was measured at
the same time with a ruler ±0.5 cm. Visual obstruc- tion and litter depth readings were taken every 2 weeks
from early May to mid July.
2.3. Nest search Nest searches were conducted three times each year
at 3-week intervals starting at the end of May. All upland habitats on the islands were searched for wa-
terfowl nests. In 1994, newly established patches of emergent vegetation were also searched for nests by
walking through the vegetation. Searches were done between 06:00 and 12:00 hours when most females
are attending their nests Gloutney et al., 1993. Four to eight observers walked side-by-side along transects,
making noise and beating vegetation with 2.5 m bam- boo sticks in order to flush nesting females. We con-
sider that nest detection probability was equivalent among treatments characterized by herbaceous cover
without shrubs. Species was determined by identifying the flushing female or using egg and down characteris-
tics. Nest location was marked with 1 m long bamboo sticks in 1992 and 1993, and with reed-grass Phrag-
mites communis Trin. stalks in 1994. Markers were
S. Lapointe et al. Agriculture, Ecosystems and Environment 78 2000 261–272 265
placed 4 m NE of the nest. Semi-rigid plastic poles were used as nest markers in pastures in 1993 and
1994 because bamboo sticks proved to be inadequate markers when cattle were present most of them fell in
1992. Nest locations were also marked on 1 : 10 000 aerial color photographs. A nest was defined as a bowl
containing at least one egg. Number of eggs, amount of down and incubation stage evaluated by floatation
Westerskov, 1950 were recorded at each visit. Nests were revisited during the following nest search or after
the expected hatching date. A nest was considered suc- cessful if at least one egg hatched. Unsuccessful nests
were recorded as depredated if there was evidence of broken or missing eggs, abandoned if the nest was not
tended by a female and had cold eggs, and trampled if the clutch and the surrounding area were destroyed by
cattle hooves. Predators were identified when possible following the criteria of Rearden 1951 based on re-
maining eggshells andor characteristics of the nests.
2.4. Costs of habitat improvements For each treatment, costs of ground preparation
tillage, harrowing, seeding and spreading of herbi- cides and fertilizers, chemical products, seeds and
fencing were evaluated. Costs for ground prepara- tion were obtained from Le Comité des références
économiques en agriculture du Québec 1994 and do not represent the actual costs involved during the
study because some operations were done on an ex- perimental basis and were therefore more costly than
regular operations. Seed prices are from Labon. All prices are in 1997 CDN dollars.
2.5. Data analyses Paired t-tests were used to evaluate the effect of cat-
tle grazing by comparing biomass between ungrazed and grazed plots. Comparisons among treatments or
sampling periods for live and dead biomass, visual obstruction and litter depth were made using analy-
ses of variance ANOVAs. Homogeneity of variances and normality were verified and logarithmic transfor-
mations were applied when necessary. When a sig- nificant effect was found, Tukey multiple comparison
tests were performed to determine where differences occurred.
Annual differences in the species composition of the breeding population were compared with χ
2
tests as well as the proportion of early, intermediate and
late nests in each treatment. Comparisons of expected and observed distributions of nests among treatments
were made using χ
2
. The proportion of the nesting population expected numbers of nests in each treat-
ment was weighed proportional to the area covered by each habitat. For each year, contrast tests were used
to determine where differences occurred among treat- ments. In 1994, nest success was calculated for each
treatment using the Mayfield-40 method followed by pairwise comparisons of daily survival rates among
treatments Johnson, 1979. For all analyses, signifi- cance level was fixed at P = 0.05 and all means are
shown ±1 SE.
3. Results