S. Lapointe et al. Agriculture, Ecosystems and Environment 78 2000 261–272 269
Table 5 Mayfield nest success and fate of nests in six treatments on islands at Varennes, Quebec, 1994
Treatment n
a
Mayfield success 95 confidence intervals
Fate Hatched
Depredated Abandoned
Trampled Idle field
110, 105 69 a
b
58–81 81
12 7
Improved pasture 21, 18
15 b 5–46
39 28
33 Unimproved pasture
37, 34 68 a
51–89 76
6 18
DNC93
c
35, 29 82 a
67–100 86
10 4
DNC94
c
34, 32 53 a
35–81 72
16 12
Ploughed 15, 15
47 ab 21–99
73 27
Total 252, 233
63 55–71
76 14
7 3
a
Numbers of nests used to calculate Mayfield success and fate of nests, respectively. Nests with unknown fate are excluded for calculations of apparent nest success hatched.
b
Percentages followed by the same letter are not significantly different P 0.05.
c
DNC93 and DNC94 refer to the dense nesting cover treatments established in 1993 and 1994, respectively.
3.5. Costs of habitat improvements The improved pasture was the most expensive treat-
ment because of the higher price of seeds and fencing Table 6. Fencing was considered only for pastures as-
suming that cover was indirectly protected from graz- ing in DNC and idle field. When cows are present,
costs associated with unimproved pasture should be added when establishing idle fields or DNC.
4. Discussion
The study lacked spatial replication because it was impossible to find other islands similar in size,
cover and distance from shore. Moreover financial
Table 6 Cost per hectare 1997 CDN of habitat improvements on islands
at Varennes, Quebec Treatment
Ground Chemical Seeds Fencing
c
Total preparation
a
products
b
Improved pasture 33
180 192
d
474 879
Unimproved pasture – –
– 474
474 Dense nesting cover 33
180 135
e
– 348
a
Includes land tillage 14, spreading of herbicide 2, har- rowing 9, seeding 6 and spreading of fertilizers 2.
b
Includes fertilizers 94, lime 76 and herbicide 10.
c
Permanent fences for 1 ha 100 m × 100m = 400 m. Cost in- cludes labor and material.
d
Seeds: 20 kg ha
− 1
of a mixture of Bromus sp. 24, Phleum pratense 34, Melilotus officinalis 34 and Trifolium sp. 8.
e
Seeds: 16 kg ha
− 1
of a mixture of Phalaris arundinacea 56 and Festuca elatior 44.
constraints precluded the set up of a large scale repli- cated experiment. Nevertheless, temporal replication
before and after was utilized to circumvent this problem. Furthermore, the study results are consid-
ered applicable to the Varennes islands and to other nearby islands of the St. Lawrence river.
Results from this study, along with those of Barker et al. 1990, indicate that with appropriate manage-
ment, use of prairies by cattle and their improvements for nesting waterfowl is possible. Even though the ro-
tational grazing system and cover improvement did not result in a greater overall nest density in 1993, changes
in nest distribution occurred and a greater proportion of nests were found in the idle field. Two years of rest
was sufficient for the plants to recover and to estab- lish new production. In 1994, there was more live and
dead biomass in the idle field than in the improved pasture. The number of duck nests increased in the idle
field and was higher than expected based on the area covered by this treatment whereas it was lower than
expected in the improved pasture. Before the transfer of cattle in July, unimproved pasture had more vege-
tation than the improved pasture and this resulted in an increase of duck nest numbers in 1994. Restrict-
ing cattle to a smaller improved pasture was clearly beneficial for the overall waterfowl production on the
islands.
Seeding dense vegetation was an effective way of improving nesting cover on the St. Lawrence river is-
lands. Newly sown DNC did not have enough litter in May to be suitable for early nesters. Two years after
its establishment, however, the DNC93 had reached a density of 7.0 nests ha
− 1
with high nest success. This
270 S. Lapointe et al. Agriculture, Ecosystems and Environment 78 2000 261–272
contrasts with results in southern Saskatchewan where only 1.1–1.4 nests ha
− 1
with 8–26 nest success were found in DNC plots McKinnon and Duncan, 1999.
In dryer regions such as the mid-continent prairies, DNC reaches maximum growth after 3–5 years and
may become too dense for nesting when 7–8 years old Duebbert and Kantrud, 1974; Duebbert et al., 1983.
Maximum growth and a reduction in the value of DNC for nesting waterfowl could occur over a shorter time
interval in Quebec where growing conditions are bet- ter. In this case, grazing by cattle during short periods
as it was done in the fall 1993 on the Varennes islands, could be used as a tool to maintain cover quality of
DNC.
Nest densities on islands at Varennes were similar to those on other artificial or natural islands in southern
Quebec Bélanger and Tremblay, 1989; Bélanger and Lehoux, 1995 but lower than in the Prairie pothole
region, reflecting different abundance of waterfowl Giroux, 1981; Duebbert, 1982; Willms and Craw-
ford, 1989. Gadwall, mallard, and northern pintail were the most abundant species nesting at Varennes
consistent with the results reported by Bélanger and Lehoux 1995 for islands in southern Quebec. Gad-
wall and mallard are common island nesters but the use of islands by pintail is less frequent Giroux, 1981;
Duebbert et al., 1983. This may be attributed to re- gional variation in the abundance of this species and
to the low and scarce vegetation found on some parts of islands of the St. Lawrence river, providing suit-
able nesting cover for pintails Bélanger and Trem- blay, 1989; Bélanger and Lehoux, 1995.
About one third of the nests in the improved pasture were trampled. Cow density increases with reduction
of area allocated to grazing and therefore the proba- bility of nest trampling is higher Jensen et al., 1990.
Although the present study recorded a high percent- age of nests destroyed by cattle, the total proportion of
trampled nests for all the islands remained low 3 after fencing. Trampling of nests by cattle seems to
be of secondary importance compare to the effects of grazing since many studies have also showed a low
percentage of trampled nests with specialized grazing systems Koerth et al., 1983; Bareiss et al., 1986.
High nest success is typical of island nesting ducks because isolation from mainland reduces mammalian
predation Lokemoen and Woodward, 1992. The absence of trees at Varennes may have also reduced
nest predation by raccoons and by birds like Ameri- can crows Corvus brachyrhyncos Brehm, which are
more important predators when they have perches. Herring gull Larus argentatus Pont. and Great
Black-backed Gull Larus marinus L. are also known to prey upon duck nests but their numbers were low.
Before fencing, grazing and trampling of interior marshes by cattle was considerable and emergent veg-
etation was nearly absent. Excluding cattle from marsh edges allowed over-water nesting by mallard and red-
head. This record of breeding redheads at Varennes is one of the most northeastern one for that species Gau-
thier and Aubry, 1995. Emergent vegetation can pro- vide additional space for pairs as well as good brood
rearing habitat both as escape cover and support for invertebrates that serve as food for ducklings Whyte
et al., 1981. A larger number of broods was recorded in marshes on and around the islands after fencing
Bélanger, L., unpublished data. Brood survival may have also been better in 1993 than in 1992, contribut-
ing to the increase in the number of duck nests in 1994.
Improved pasture was the most expensive manage- ment and resulted in low nest production but these ex-
penditures were necessary to keep cattle in a more re- stricted area with good forage. The DNC system was
expensive compared to idle fields but had three times as many successful nests per hectare than idle fields.
Life expectancy of the seeded DNC is estimated to be more than 10 years Lokemoen, 1984 with approxi-
mately 1000 per year for maintenance and rotation of cattle. Long-term effects of improvements were not
evaluated and costs could be amortized over the years. Benefits may then be higher if duck numbers increase
with homing of successful females Lokemeon et al., 1990.
5. Conclusions
