Conservation Evidence strives to be as useful to conservationists as possible. Please take our survey to help the team improve our resource.

Providing evidence to improve practice

Action: Provide artificial nesting sites for wildfowl Bird Conservation

Key messages

Read our guidance on Key messages before continuing

  • Six studies from North America and Europe found that wildfowl populations increased with the provision of artificial nests, although one study from Finland found that there was no increase in the number of broods or chicks in areas with nest boxes.
  • Twelve studies from North America investigated the success of nests in artificial nests with nine finding that success and productivity was high, sometimes higher than or similar to natural nests. Two studies found that success for some species in nest boxes was lower than for natural nests. Two studies investigated the impact of nest box location, finding that hidden nests had higher success and that nests over water were more successful than those in trees over land.
  • Nineteen studies from across the world investigated occupancy rates of artificial nests, finding that rates varied from no use of 25 nest boxes in a single site in Indonesia to 100% occupancy across 20 sites in the USA with one study finding that nest boxes were used more than natural cavities. Two studies found that occupancy rates increased over time, whilst four studies found that occupancy rates appeared to be affected by design or positioning.
  • Three studies from North America found that nest boxes could have other impacts on reproduction and behaviour, with common starlings Sturna vulgaris (a nest site competitor) avoiding some nest box designs; hidden nest boxes having lower intra-specific nest parasitism than easily visible boxes and female common eiders Somateria mollissima losing less weight over incubation if they were nesting in shelters, compared to birds nesting in the open, although they lost weight quicker after nesting.


Supporting evidence from individual studies


A replicated study in 1941-6 at a wetland site in Connecticut, USA (Frank 1948), found that wood ducks Aix sponsa nested in only 12 of 274 nest boxes (4%) erected between 3.6 m and 7.3 m off the ground. In contrast, up to 67% of boxes were occupied by grey squirrels Sciurus carolinensis at once, with other rodents and insects also occupying boxes.



A replicated study in wetland habitats in Washington State, USA, in 1944-53 (Yocom 1952), found that 12 of 18 artificial nests made from willow were used by Canada geese Branta canadensis in 1944. Over the next four years, a further 13 nests of various types were placed in trees and by 1951 there were 53 available nests in the area. Twelve of these were inspected in 1951 and found to contain eggs. The author also states that placing logs on river islands increases their attractiveness to nesting geese, but no data was provided to support this.



A replicated trial in 20 woodland sites on Rhode Island, USA, in 1955-56 (Cronan 1957) found that wood ducks Aix sponsa used 36-100% of nest boxes installed in study areas. A total of 102 nest boxes were observed in 1955 and 85 in 1956. This study is discussed in more detail in ‘Use artificial nests that discourage predation’.



A before-and-after study on a marshland site in Montana, USA (Craighead & Stockstad 1961), found that the breeding population of Canada geese Branta canadensis did not increase consistently following the installation of raised nesting platforms in 1954 (200 pairs in 1953, increasing to 285 in 1955, falling to 154 in 1956-8). The authors note that the population decrease after 1956 appeared to be due to overhunting. Platform use increased from 5% of the breeding population in 1954 to 18% in 1958. Clutch size and nest success were similar on and off platforms, but significantly more goslings hatched in platform nests (average of 3.6 goslings/clutch for 49 nests on platforms vs. population average of 3.1 goslings/clutch for 1,113 nests). Platforms consisted of a wooden tray 76 cm x 66 cm, 15 cm deep and filled with soil and decaying vegetation. They were placed at heights of 1.2-13.7 m in trees on islands in a lake and on the lake’s shoreline. This study is also discussed in ‘Use artificial nests that discourage predation’.



A replicated, controlled study from 1958-1961 in multiple woodlots containing nestboxes (114 ha in total) and 1 woodlot containing all natural cavities (9.3 ha) in Illinois, USA (Bellrose et al. 1964), found that wood duck Aix sponsa breeding pair density increased from 10-15 to over 90 pairs during the study period. Ducks exhibited higher nest success in nestboxes (71% success for 574 metal nest boxes vs. 37% for 116 natural cavities), although a smaller proportion were occupied (48% compared 23% occupation), probably due to lower racoon Procyon lotor predation (see ‘Use artificial nests that discourage predation’). Female wood ducks usually returned to the nesting areas where they last bred successfully, so the authors suggest that nest boxes should be grouped into units (2-3 per ha in high-quality habitat were recommended). Most nestboxes were metal cylinders with elliptical entrances.



A replicated study on a total of 11 marshland sites in Iowa, USA, in 1964-9 (Bishop & Barratt 1970) found that mallards Anas platyrhynchos used 33% of 705 artificial nests over the study period. The percentage of the mallard population using artificial nests increased from 31% in 1966 to 46% in 1969. Four other species (blue-winged teal A. discors, gadwall A. strepera, redheads Arytha americana and Canada geese Branta canadensis) used a total of 12 nests over the study period. Nesting success in artificial nests was 87%, far higher than previous records of mallard nest success (normally 27-52%). Nests were cone-shaped, hardware-cloth baskets, 18 cm deep and erected on poles sunk into marshland, dry land and in vegetation. Not all 11 sites were used every year.



A small replicated, controlled study from 1963-1970 in 15 sites of marshy or wooded duck habitat in Maryland, USA (McGilvrey & Uhler 1971) found that wood ducks Aix sponsa had no significant preference for next box design but that common starlings Sturna vulgaris avoided horizontal nest boxes with large entrances. Starlings showed greater preference for vertical boxes with small entrances (3 x 4 inches) and avoided horizontal boxes with large entrances (semicircular, 11 inches in diameter). In 1965, when all vertical boxes were removed and all horizontal boxes had larger openings, there was an abrupt decrease in starling nest box use, despite no change in the starling population size. Starlings preferred boxes in open sites than those in wooded sites, whereas wood ducks showed no preference. The basic experimental nest box was a horizontal cylinder (24 inches long and 12 inches in diameter) made of metal or wire netting covered with roofing paper and had wooden ends.



A before-and-after trial in a mixed forest and wetland site in Mississippi, USA (Strange et al. 1971), found that the population of wood ducks Aix sponsa in the study site increased dramatically following the installation of 253 nest boxes between 1966 and 1969 (average of 30-35 pairs in 1960-5 vs. 231 nest boxes used in 1969). Between 1966 and 1969, 15,273 eggs were laid in nest boxes, with 6,036 ducklings leaving nest boxes. In contrast, ten belt transects (20 m x 1,128-5,486 m) detected 27 natural nesting cavities, none of which were used by ducks. Hatching success was 67% in nest boxes, with 10% of eggs being predated in 1969. A further study at a wetland site in Louisiana, USA, found that, in 1969, 53% of 30 nest boxes erected were used by wood ducks, with 243 eggs laid, of which 47% hatched (34% were destroyed by predators).



A replicated study in Australia (Norman & Riggert 1977), found that wildfowl in Western Australia used only 1% of 1,999 artificial nests in 1974, whereas 36% of 2,440 artificial nests in Victoria were used in 1975-6. The majority of records from Victoria were of chestnut teal Anas castanea, which are uncommon in Western Australia. All Western Australian nests were made from plastic drums and erected at 23 wetland sites in 1969-74; Victorian nests were made from various wooden and metal boxes and plastic drums and were erected at 26 wetland sites between 1975-6. All nests were attached to poles and trees at heights of up to 3.6 m. Nests below 50 cm off the ground were mostly avoided by birds.



A replicated study between 1964 and 1975 in six wetland sites in Texas, USA (McCamant & Bolen 1979), found that black-bellied whistling ducks Dendrocygna autumnalis used an average of 81% of nest boxes erected in trees. On average, 52 nest boxes were available each year and were monitored an average of 14 times a year. A total of 778 clutches were laid over the study period, with 40% incubated and 75% of these hatching at least one egg successfully (210 nests, 28% of all nests). Sixty three percent of eggs in successful nests hatched, compared with a population average of 20%.



A series of replicated studies at river and lake sites in northern Ontario, Canada in 1974-9 (Lumsden et al. 1980), found that cavity-nesting ducks (mainly common goldeneyes Bucephala clangula) preferentially used nest boxes with large (13 x 10 cm) entrance holes high (33 cm) above the floor of nest boxes with dark interiors. Nest boxes with large entrance holes were used more than those with medium (10.5 × 8 cm) holes; boxes with small (7.5 × 6 cm) holes were not used by goldeneyes or hooded mergansers Mergus cucullatus (318 sets of boxes tested). Boxes with entrance holes 18 or 25.5 cm above the base of the box were not used by goldeneyes (201 sets) and boxes with dark-stained interiors were used more than those with unstained interiors (39 breeding attempts vs. 13 attempts, 73-5 sets for each of six years). Differences between tree species were minimal and orientation had no impact.



A replicated study in 1978 in a forested marshland site in South Carolina, USA (Griffith & Fendley 1981), found that wood ducks Aix sponsa used 89% of 55 nest boxes erected between 1974 and 1978. Five-gallon plastic buckets were used slightly more often than wooden nest boxes and ‘fiber cylinders’ (95% of 20 buckets used, compared with 86% of 35 boxes and cylinders). Hatching rates did not vary between nest types, with 28% of the 847 eggs laid being predated or deserted and 79% of the remaining 608 eggs hatching. Bucket nests were five-gallon buckets with a 7.6 cm diameter entrance hole and a secured lid. All nests were placed at a variety of heights and in a variety of vegetation types.



A replicated controlled study 1977-9 in riverine forests in Louisiana, USA (McComb & Noble 1981), found that wood ducks Aix sponsa used nest boxes more frequently than natural cavities (0.4% of 5,374 nest boxes inspected contained nests vs. 0.03% of 3,993 natural cavities). The most frequently used nest boxes were large (60 X 60 X 30 cm), with a circular or oval entrance of less than 140 cm2. This study also examined nest box use by other birds (owls, woodpeckers and songbirds).



A before-and-after study in northern Scotland (Dennis & Dow 1984) found that a breeding population of common goldeneyes Bucephala clangula established itself in a forested landscape following the installation of a total of 83 nest boxes between 1961 and 1982. Goldeneye numbers were monitored from 1960, with a single female nesting in a natural cavity in 1970. Nest boxes were first used in 1974 (two breeding attempts), with 41 breeding attempts in 1982 and the percentage of occupied boxes increasing from 6% to 49% over the same period. Occupancy rates and nesting success were highest for boxes close to rivers (72% occupation for 13 boxes, 78% success for 36 attempts), compared with those by small lakes (30% occupancy, 57% success for 37 boxes and 58 attempts), large lakes (26% occupancy, 50% success for 25 boxes and 44 attempts) or marshes (4% occupancy and 50% success for nine boxes and two attempts).



A small single-site study from March-May in 1981-1982 in an island (1.2 ha) within Kentucky Lake in Tennessee, USA (Spero & Pitts 1984) found that wood ducks Aix sponsa nested in 44-63% of the nest boxes provided although in 1982, 11 of 44 wood duck nests were destroyed, probably by common grackles Quiscalus quiscula and nestlings in one nest were preyed upon by grackles. According to the authors, the presence of vacant nest boxes in both years suggests that grackles and wood ducks were not competing for nest sites. This study also discusses the use of nest boxes by grackles.



A series of replicated studies in 1977-84 at two lakes in eastern Ontario, Canada (Lumsden et al. 1986), found that the lining nest boxes with wood shavings significantly increased their use by four duck species, whilst entrance hole size and height above the ground had uncertain effects. All 18 ducks nesting in 100 pairs of nest boxes over two years chose boxes lined with wood shavings over those without. Entrance hole size did not significantly influence box choice by goldeneyes Bucephala clagnula but hooded mergansers Lophodytes cucullatus and wood ducks Aix sponsa used small entrances more (ten and four boxes used, compared with three and zero boxes with large holes), whilst common mergansers Mergus merganser only nested in four boxes with large entrances. Boxes 6 m above the ground were used more often by goldeneyes than those at 4.5 m or 3 m (14 breeding attempts vs. nine and three attempts, 20 sets of boxes in each of eight years) but the authors argue that occupancy rates would not change in the absence of choice.



A replicated study over 12 years between 1976 and 1987 in deciduous woodlands and wetlands in northeast Illinois, USA (Semel et al. 1988) found that successful wood duck Aix sponsa clutches in well-hidden nest boxes had significantly higher hatching success than those in conspicuous boxes (82% hatching success for 28 successful well-hidden nests vs. 74% for 150 successful conspicuous nests), probably due to lower levels of intraspecific brood parasitism (30% of 47 hidden clutches parasitized vs. 50% for 198 conspicuous ones). However, visible nests were more likely to raise at least one duckling (60% of 47 hidden nests successful vs. 76% of 198 conspicuous nests), and hatched more ducklings (7.1 ducklings/successful nest for hidden nests vs. 9.3-9.9 ducklings/nest for conspicuous nests) possibly due to larger clutch sizes caused by brood parasitism (12.4 eggs/clutch for hidden nests vs. 15.7-16.3 eggs/clutch for conspicuous nests).



A replicated study in 1987 at four ranches in Tamaulipas, Mexico (Markum & Baldassarre 1989), found that Muscovy ducks Cairina moschata only used 13 of 407 nest boxes (3%), with 77% of these successfully hatching eggs and fledging 96 ducklings. Overall hatching success was 54% of 177 eggs. Three black-bellied whistling ducks also fledged from the ten successful nests. Nest boxes were 42 x 42 x 62 cm with a 21 cm diameter entrance hole and were erected on metal or wooden poles or trees either on islands in a lake (168 boxes) or close to ponds and waterways.



A replicated study over four breeding seasons in 1985-8 at two lakes in Veracruz, Mexico (Feekes 1991), found that black-bellied whistling ducks Dendrocygna autumnalis using nest boxes had very low reproductive success (11.1% success for nine attempts in 1986, 6.6% for 30 attempts in 1987-8), mainly because of predation by opossums Didelphis spp., raccoons Procyon lotor and humans. Occupancy rates varied, with one of 16 pairs using boxes in 1985; 17-30% of 30 pairs in 1986 and 40-75% of 20 pairs in both 1987 and 1988. Nest boxes were made from either liana baskets or hollowed palm trunks, with the latter being the only nests used (except for a single basket in 1985). Thirteen baskets were provided in 1985 and ten in 1986; ten trunks were provided in 1986, 16 in 1987 and 17 in 1988. Boxes were placed in positions similar to naturally occurring nests and checked every two weeks during the breeding seasons. Opossums also frequently occupied nest boxes.



A replicated study at 16 areas in an open pine forest in South Carolina, USA (Hepp & Kennamer 1992), found that over nine breeding seasons (1982-90), between 19 and 44 female wood ducks Aix sponsa used nest boxes, with 120 (1982-3) or 150 (1984-90) boxes provided each year.



A small trial on dunes on an island in northern Alaska, USA, in 1992-3 (Divoky & Suydam 1995), found that 16 out of 20 nesting structures provided for common eiders Somateria mollissima were destroyed the year after installation, but that the remaining four provided seven potential nest sites, of which three (in two structures) were used, hatching at least one egg successfully. The structures consisted of a wooden cross providing four uncovered, semi-sheltered nesting quadrants protected on two sides by 20 x 61 cm boards. All three nests were in the south-facing quadrants.



A replicated study at three sites in southern Finland over four breeding seasons in 1993-7 (Pöysä et al. 1999) found that breeding common goldeneyes Bucephala clangula showed a significant preference for nest boxes erected on the shoreline of lakes compared to those 14–140 m into the surrounding forest (shore boxes occupied before forest boxes for 73-95% of 50 pairs of boxes, with 8% of pairs occupied in the same season). Female goldeneye inspected shore and forest boxes equally and therefore appeared to actively choose shore boxes.



A replicated one-year study in 2001 in Sumatra, Indonesia (Drilling 2001), found that no white-winged ducks Cairina scutulata were observed entering any of 25 nest boxes erected in trees in a swamp forest site. Boxes were 53 x 43 x 41-48 cm, with a 20 x 17cm entrance hole and four drainage holes, and were erected 1.5-4.0 m above the ground (mainly in durian trees Durio zibethinus). The author argues that nest boxes may take several years to be accepted and so may be used in the future.



A replicated, controlled study in a deciduous forest in British Columbia, Canada, in 1997-9 (Evans et al. 2002), found that Barrow’s goldeneyes Bucephala islandica laid larger clutches but had lower nesting success in nest boxes, compared to natural nest cavities (10.5 eggs/clutch and 45-50% success for 174 clutches in nest boxes vs. 7.5 eggs/clutch and 54-86% for 41 clutches in natural cavities). There were no differences for buffleheads B. albeola (8.4 eggs/clutch and 75-90% success for 46 clutches in boxes vs. 8.5 eggs/clutch and 55-90% success for 100 clutches in natural cavities). Hatching dates did not differ for either species between nest types. Goldeneye nests in boxes were predated mainly by black bears Ursus americanus compared with small mammals and common starlings Sturnus vulgaris in natural nests. Predation of all bufflehead nests was low and mainly by American red squirrel Tamiasciurus hudsonicus and American pine marten Martes americana. The authors suggest that differences in goldeneye nests were due to nest boxes being concentrated in highly visible locations, whilst natural nests were widely dispersed. Natural bufflehead nests, however, were positioned similarly to nest boxes.



A replicated and controlled before-and-after study in southern Finland in 1988-99 (Pöysä & Pöysä 2002) found that the number of common goldeneye Bucephala clangula breeding pairs increased on 35 lakes following the provision of 50 nest boxes (average of 0.8 pairs/lake in 1988-91, before nest box provision vs. 1.1 pairs/lake in 1995-99, afterwards). There were no increases on 17 lakes without nest boxes provided (average of 0.8 pairs/lake in 1988-91 vs. 0.9 pairs/lake in 1995-9). However, there was no increase in the number of broods at experimental lakes (0.17 broods/lake vs. 0.19 broods/lake) and the increase in the number of chicks fledging was not significant (0.59 young fledged/lake in 1988-91 vs. 0.86 young/lake in 1995-9). Nest boxes were 25 x 26 x 70 cm with a 9 cm diameter entrance hole. The authors suggest nest site availability may limit the number of breeding goldeneyes, but that other factors appear to limit reproductive output.



A replicated, randomised and controlled paired study in tundra on Mitivik Island, Hudson Bay, Nunavut, Canada, in 2001 and 2003 (Fast et al. 2007), found that female common eider Somateria mollissima provided with shelters whilst nesting lost less weight over the incubation period, compared to control females without shelters (average weight at end of incubation of 1,312 g for 34 sheltered birds vs. 1,266 g for 31 controls). Sheltered birds, however, appeared to lose weight more rapidly at the end of incubation. Sheltered nests had more stable temperatures than controls (average temperature range was 6.1oC lower for sheltered nest). Paired nests were less than 10 m apart (to ensure similar microclimates) and were at similar stages of incubation when a shelter consisting of a 46 x 46 cm roof and two 25 x 46 cm walls (with 12, 2.5 cm holes in) was placed over one of the nests. The walls were positioned facing east-west.



A replicated before-and after study in 1999-2004 in boreal forests surrounding 60 lakes in Québec, Canada (Savard & Robert 2007), found that the number of breeding pairs of common goldeneyes Bucephala clangula and Barrow’s goldeneyes B. islandica increased from ten and 28 pairs in 1999 to 46 and 43 pairs in 2003 following the provision of 105-133 nest boxes each year from 1998-9. The number of broods increased in 2000, but not subsequently. Goldeneyes used 23-43% of nest boxes, with 37-67% hatching success for 261 nests. Three nests were erected at each lake: those above water or on trees on the shore had higher success rates than those in clearcuts 25-160 m away from shore (50% success for 56 nests above water, 56% for 63 nests on the shore and 40% for 86 nests in clear cuts). Boxes were 24 × 22 × 60 cm with a 10 x 13 cm entrance hole. American kestrels Flaco sparverius also used nest boxes, although their use declined over time.


Referenced papers

Please cite as:

Williams, D.R., Child, M.F., Dicks, L.V., Ockendon, N., Pople, R.G., Showler, D.A., Walsh, J.C., zu Ermgassen, E.K.H.J. & Sutherland, W.J. (2019) Bird Conservation. Pages 141-290 in: W.J. Sutherland, L.V. Dicks, N. Ockendon, S.O. Petrovan & R.K. Smith (eds) What Works in Conservation 2019. Open Book Publishers, Cambridge, UK.