Provide supplementary food to increase reproduction/survival
Overall effectiveness category Trade-off between benefit and harms
Number of studies: 24
Background information and definitions
Many mammals have long gained a proportion of their diet as a direct result of human activities (Oro et al. 2013). Many of these are cases where by-products of production, harvesting or consumption are exploited. However, in some cases, food is provided specifically for mammals. This is often to increase survival and condition of hunted animals, such as deer. In some other cases, food may be provided to aid the conservation status of rare species. Some studies are less directly conservation-motivated but are included here if the findings can help to inform conservation actions.
Studies that provide supplementary food as part of translocation or reintroduction programmes are discussed in: Provide supplementary food during/after release of translocated mammals and Provide supplementary food during/after release of captive-bred mammals.
Oro D., Genovart M., Tavecchia G., Fowler M.S. & Martínez‐Abraín A. (2013) Ecological and evolutionary implications of food subsidies from humans. Ecology Letters, 16, 1501–1514.
Supporting evidence from individual studies
A controlled study in 1975–1976 in a grassland site in Illinois, USA (Cole & Batzli 1978) found that where supplementary food was provided, prairie vole Microtus ochrogaster numbers were temporarily higher and litter size was larger than in an area with no supplementary food. Voles reached higher densities in the food supplemented area (135 voles/ha in April 1976) than in the area with no supplementary feeding (90 voles/ha in October 1975). However, 16–18 months after supplementary feeding commenced, vole numbers were similar in fed and unfed areas (<5/ha). Voles in the fed area had a longer life expectancy and were more likely to breed in winter than voles in the unfed area (data expressed as model results). Average litter size was larger in the fed area (5.1) than in the unfed area (4.3).A 1.5-ha abandoned pasture was divided into two live-trapping grids of 0.80 and 0.55 ha, separated by a 10-m-wide mown strip. On the 0.55-ha grid, 210 feeding stations (200-ml bottles, filled with rabbit pellets, replenished as required) were placed 5 m apart. No supplementary food was provided on the other grid. Voles were surveyed using 60 wooden traps in the supplementary feeding grid and 72 in the unfed grid. Every three weeks from May 1975 (supplementary food grid) and August 1975 (unfed grid) to November 1976, traps were set for three days and checked twice daily. Traps were baited for two days before setting.Study and other actions tested
A before-and-after study in 1975–1976 in a woodland in Illinois, USA (Hansen & Batzli 1979) found that supplementary feeding did not increase white-footed mouse Peromyscus leucopus densities. Monthly densities varied seasonally but were not higher in supplementary feeding plots than in plots without supplementary food provision. After supplementary feeding commenced, the highest numbers were 5–9 months later, with 20–26 mice/ha in supplementary feeding plots and 22–29 mice/ha in plots without supplementary food. Four plots, 0.36 ha each, were established within a 9-ha live trap grid, with 20-m trap intervals. Traps were operated across the grid over three days/month, from January 1975 to July 1976. Additional trapping took place fortnightly on grid points within and immediately surrounding plots, from March–December 1976. Supplementary feeding, using mouse chow at 10-m intervals, commenced in two plots in January 1976. No food was provided in the other two plots.Study and other actions tested
A controlled study in 1984 on three areas of a predominantly grassland site in Colorado, USA (Baker & Hobbs 1985) found that supplementary feeding of mule deer Odocoileus hemionus hemionus increased overwinter survival. Mortality was lowest for deer provided with as much supplementary food as they could consume (24%), intermediate for deer given fixed quantities of supplementary food (33%) and highest for deer not provided with supplementary food (53%). Three study areas (≥5 km apart, 660–1,000 ha extent) were monitored. Supplementary food consisted of wheat middlings, brewer’s dry grain, cottonseed hulls and alfalfa formed into wafers. It was provided daily, from 7 January to 10 April 1984, in equal or greater quantities than deer consumed in one study area and at 0.9 kg/deer/day in another study area. No supplementary food was provided in the third area. Biweekly aerial deer counts were conducted from 27 January 1984 and mortality was assessed by ground surveys for carcases, during 1–15 June 1984, of randomly selected sample plots from each study area.Study and other actions tested
A before-and-after study in 1966–1969 and 1973–1974 on a forested island in Lake Beldany, Poland (Banach 1986) found that when supplementary food was provided, the abundance, body weight and survival of bank voles Clethrionomys glareolus was higher. Annual peak vole abundance was higher in years when food was provided (835–1,068 individuals) than when no food was provided (157–368 individuals). The average body weight of young voles (3–9 weeks old) was higher in years when food was provided (17.2 g) than when they were not fed (13.9 g). The survival of individuals to autumn in the year they were born was higher in years when food was provided (49%) than when voles were not fed (8–42%). Voles were live-trapped every six weeks from spring to autumn 1966–1969 and 1973–1974, in five 10–14-day trapping sessions/year. Two to five traps baited with oats were set at each of 159 trapping locations and checked twice daily. From spring 1973 to autumn 1974, a total of 159 boxes with 3 kg of oats each were distributed 15 m apart across the 4-ha island, next to trapping sites. Boxes were replaced when half the oats had been consumed, but were removed during trapping.Study and other actions tested
A replicated, paired sites, controlled, before-and-after study in 1983–1986 in four mixed spruce and pine forest sites in British Columbia, Canada (Sullivan 1990) found that providing supplementary food increased the abundance of red squirrels Tamiasciurus hudsonicus. After two years, squirrel abundance in sites with supplementary food was higher (41–53 squirrels/site) than in unfed sites (9–15 squirrels/site). One year after supplementary feeding ceased, squirrel numbers declined in previously fed sites (23–31 squirrels/site) but not in unfed sites (11–12 squirrels/site). A 9-ha grid, with 100 stations at 30 m intervals, was established in each of four forest sites (two each in two forests). Sunflower seeds (83–90 kg/month) were provided in cans nailed to trees distributed across two sites (50 cans/site), from September 1983 to September 1985. No food was provided at the other two sites. From June 1983 to June 1986, squirrels were captured and measured using one Tomahawk live trap at alternate stations. Traps were set for two days, every 3–4 weeks in summer (April–September) and 4–10 weeks in winter (October–March). Cans were refilled after each trapping period.Study and other actions tested
A replicated, controlled study in 1979–1990 in four mountainous grassland areas in northern Sweden (Angerbjörn et al. 1991) found that providing supplementary food increased occupancy of Arctic fox Alopex lagopus dens and the number of fox litters born, but not the numbers of cubs in each litter. Where supplementary food was provided, a higher proportion of dens were occupied (35%) than where no supplementary food was supplied (6%). Over five years, 17 of 65 dens (26%) where food was provided contained a litter while only three of 103 dens (3%) where no food was provided contained a litter. However, there was no significant difference in average litter size (supplementary food: 5.2 cubs; no food: 5.7 cubs). During January–April of 1985–1989, reindeer Rangifer tarandus and moose Alces alces meat was placed 50–200 m from 168 dens which showed signs of Arctic fox activity. In some cases, meat was buried in the snow. About 50–100 kg of meat/den/year was provided. Dens were surveyed for presence of foxes and offspring in June–August of 1979–1990.Study and other actions tested
A replicated, site comparison study in 1988–1992 of forest and heathland across nine management areas in the Netherlands (Bruinderink et al. 1994) found that when supplementary feed was provided, wild boar Sus scrofa annual population recruitment rates were higher. No statistical analyses were performed. In seven areas, where boar were fed, annual recruitment (number of piglets >2 months old/ adult female) averaged 2.2–2.5, compared to 0.0–2.5 at a site where supplementary feeding ceased in the year before the study began. At a further site, where supplementary feeding ceased two years into the study, recruitment averaged 2.0–2.4 over those first two years and 1.5–1.7 in the subsequent three years. Recruitment data were obtained from nine boar management areas, based on spring counts at feeding locations.Study and other actions tested
A controlled study in 1986–1989 of a forested area in Wisconsin, USA (Lewis & Rongstad 1998) found that supplementary feeding of white-tailed deer Odocoileus virginianus did not increase their overall survival. The average annual survival of winter-fed deer (78%) or summer-fed deer (53%) did not differ significantly from that of unfed deer (64%). Summer- and winter-fed deer had higher over-winter survival during a single severe winter only (summer-fed: 96%; winter-fed: 100%; not fed: 79%), but not during other periods. From October 1986 to July 1989, deer were fed shelled corn or commercial deer food from mid-April to mid-December (summer-feeding – 53 deer), 1 December to 30 April (winter-feeding – 66 deer) or were not fed (48 deer). All deer, except 24 that were winter-fed, occupied a 15 × 30-km area. No deer was winter-fed and summer-fed in the same year. Survival was monitored through radio-tracking. Deer use of feeders was determined by direct observations.Study and other actions tested
A replicated, controlled study in 1995–1996 in a grassland in Middleveld, eSwatini (Monadjem & Perrin 1998) found that multimammate mice Mastomys natalensis provided with supplementary food had similar home range sizes and distance between captures to unfed mice. The average home ranges of 66 multimammate mice provided with supplementary food (600–923 m2) did not differ significantly from those of nine unfed mice (838–960 m2). Similarly, average distances between captures of mice provided with supplementary food (20–21 m) did not differ significantly from those of unfed mice (25–28 m). In May 1995, three 100 × 100-m plots were established in a natural grassland. Supplementary food (4 kg of rolled oats and 4 kg of rabbit pellets) was provided monthly, from July 1995 to May 1996, in two plots. No supplementary food was added to the third plot. From June 1995 to May 1996 mice were surveyed monthly using 100 Elliot and Sherman live traps/plot. Traps were set 10 m apart, on three consecutive nights/month. Mice were individually toe-clipped and weighed when captured. Only individuals captured at least five times were used to calculate home range sizes.Study and other actions tested
A controlled study in 1995 on a grassland in KwaZulu-Natal, South Africa (Perrin & Johnson 1999) found that providing supplementary food increased striped mouse Rhabdomys pumilio density and the proportion of the population that was breeding. Three to six months after feeding began, there were more striped mice in the plot with supplementary food (30) than in the plot with no supplementary food (21). Over the same time period, a higher proportion of adult mice were reproductively active in the plot with supplementary food (85%) than in the plot with no supplementary food (38%). In one of two plots (>60 m apart) 25 trays, each with 1 kg of oat seeds, were filled weekly. The second plot had no supplementary food. In each plot, mice were monitored at 49 stations, in a 7 × 7 grid, at 10-m intervals. Each station was surveyed for two consecutive nights/month with one baited and insulated Elliot or Sherman live trap, from January–June 1995.Study and other actions tested
A randomized, replicated, controlled study in 1991–1995 in two mountain ranges in Colorado, USA (Miller et al. 2000) found that supplementary winter feeding of Rocky Mountain bighorn sheep Ovis canadensis canadensis did not increase lamb survival. Average annual recruitment did not differ between herds provided with food (0.5–0.7 lambs/adult female) and herds where no food was provided (0.6–0.7 lambs/adult female). Adult bighorn females of four herds were captured in February–March 1991–1995 and were marked and radio-collared. Between 1991 and 1995 the herds were either fed from mid-December for 8–10 weeks with 2 kg/individual/day of alfalfa hay and 1 kg/individual/day of apple pulp, or not given any supplementary food. Each year, one herd under each feeding regime was additionally medicated for lungworm using fenbendazole, while the other was not medicated. Treatments were rotated annually under a predetermined, randomly selected scheme. Lamb survival for 11–18 marked adult females/herd was assessed every two weeks between May and October the following year.Study and other actions tested
A replicated, controlled study in 1990–1992 in a forest reserve in Kansas, USA (Eifler et al. 2003) found that cotton rats Sigmodon hispidus had different growth rates after the provision of supplementary food, depending on their size and sex, and the time of year. In winter, the growth rate of small cotton rats provided with supplementary food was significantly higher than that of small rats not provided with food, but the opposite was true for larger rats. In spring, males on supplemented grids grew faster than males on control grids, but the opposite was true in females. In summer, there was no difference in growth rates between supplemented and non-supplemented grids. In autumn, males were the same as in winter, but larger females grew faster with supplementary food (data presented as model results). Additionally, seven reproductive cotton rat females had a higher growth rate when provided with food (2.5 g/day) than did 14 non-supplemented females (2.0 g/day). Seven litters born to females on food supplemented grids had higher growth rates in their first month of life (1.4 g/day) than 23 litters born on non-supplemented grids (0.94 g/day). Between June 1990 and May 1992, supplementary food was distributed along two out of four trapping grids. Food (50 g each of sorghum seeds, millet seeds and commercial rabbit chow) was provided in cans that were refilled every two weeks. Grids contained 64–99 trapping stations, 15 m apart, each with two Sherman traps baited with scratch grain. Traps were set for three consecutive days/month, and checked twice daily. Rats were individually marked and weighed when captured. In June 1991, one of the food supplemented and one of the non-supplemented grids were switched.Study and other actions tested
A replicated, randomized, paired sites, controlled, before-and-after study in 1996–1999 in three forest sites in British Columbia, Canada (Ransome & Sullivan 2004) found that supplementary feeding did not alter the abundance and body mass of northern flying squirrels Glaucomys sabrinus and Douglas squirrels Tamiasciurus douglasi, but it did increase survival of northern flying squirrels. Between June 1997 and April 1999, the survival rate of northern flying squirrels was higher in plots with supplementary feeding (0.93) than without supplementary feeding (0.79). Survival did not significantly differ between plots before feeding began (plots to be fed = 0.84; control plots = 0.92). The survival of Douglas squirrels was similar between fed (0.72) and unfed (0.80) plots. The abundance and body mass of squirrels did not differ significantly between plots with supplementary food (northern flying squirrel abundance: 11.8/ha; body mass: 131 g; Douglas squirrel abundance: 14.2/ha; body mass: 200 g) and plots without supplementary food (northern flying squirrel abundance: 7.7/ha; body mass: 128 g; Douglas squirrel abundance: 20.1/ha; body mass: 207 g). From April 1997 to May 1998 and from September 1998 to April 1999, supplementary food was provided at 90 feeding stations, 60 m apart in a 9×10 grid, in each of three 30-ha forest plots. Stations were filled with 7 kg of sunflower seeds at 5–6-week intervals or when seed was depleted. Three other 30-ha plots had no feeding stations. In each plot, squirrels were trapped every 5–6 weeks (when snow-free), from June 1996 to March 1999, using 80 baited Tomahawk live traps, at 40-m intervals in an 8×10 grid.Study and other actions tested
A replicated, controlled study in 1985–2008 in two shrubland areas in southern Spain (López-Bao et al. 2010) found that supplementary feeding did not increase the breeding rate of Iberian lynx Lynx pardinus or survival of offspring. The proportion of female lynx that reproduced in areas where supplementary food was provided (66%) did not differ significantly from that in areas where it was not (83%). Similarly, survival of lynx offspring did not significantly differ (supplementary food: 100%; no supplementary food: 88%). In 2002–2008, six lynx breeding territories were each supplied, throughout the year, with live domestic rabbits at approximately three feeding stations. An unspecified number of other territories were not supplied with rabbits. Fifteen adult female lynx were fitted with radio-collars and were monitored in 1985–2007. Data on breeding were obtained in March–May of 1993–2008, by tracking females to locate dens. Lynx were also monitored by sightings, camera-trapping, and radio-tracking.Study and other actions tested
A replicated study in 2007–2009 in six agroforestry sites in Alentejo and Algarve, Portugal (Loureiro et al. 2011) found that European rabbits Oryctolagus cuniculus used most available artificial feeding stations. Rabbits used almost 70% of 48 feeding stations surveyed. Rabbit numbers were higher in areas where a higher proportion of feeding stations was used (data presented as a correlation). Over the course of the study, which included providing artificial shelters and waterholes, the number of rabbit latrines increased from 16 to 25 latrines/km (no statistical analysis conducted). Between July and September in 2008 and 2009, wheat, oat and alfalfa were made available through 120 artificial feeding stations in six agroforestry. Each station was protected by a fence, aimed at excluding large animals. However, 60% of feeding stations were destroyed by deer or wild boar, so data for 48 feeding stations were analysed. These were surveyed monthly and considered to be used if rabbit droppings were detected. Rabbit abundance was estimated based on the number of latrines/km counted along paths at each site.Study and other actions tested
A replicated, randomized, controlled study in 2001–2006 in eight forest, grassland and shrubland sites in Utah, USA (Peterson & Messmer 2011) found that providing supplementary food over winter did not increase mule deer Odocoileus hemionus survival or reproductive success. The average annual survival of deer with supplementary feeding (80%) did not differ significantly from that of deer without supplementary feeding (73%). Similarly, the average reproductive success of deer with supplementary feeding (0.58 fawns/female deer) did not differ significantly from that of deer without supplementary feeding (0.57 fawns/female deer). In 2001, eight sites known to host winter concentrations of mule deer were randomly selected. Supplementary food (corn, alfalfa and protein pellets, 0.9 kg/deer/day) was provided over winter (December–March 2001–2005) at four sites. No food was provided at the other four sites. Sites with and without supplementary food were >3 km apart. Fifty-two female mule deer receiving supplementary food and 38 that were not fed were radio-collared between January and March 2001–2005. They were monitored 2–3 times/week, from May 2002 to January 2006.Study and other actions tested
A replicated, randomized, controlled study in 2009–2010 in 23 mixed wetland, scrubland, and wasteland sites in New Hampshire, USA (Weidman & Litvaitis 2011) found that supplementary feeding increased survival of eastern cottontail rabbits Sylvilagus floridanus. After two months, rabbit survival in sites where supplementary food was provided was higher (9 of 15 animals; 60%) than in sites where no food was provided (5 of 13 animals; 38%). In November 2009–March 2010, twenty-eight rabbits were trapped and fitted with radio-collars and ear tags. Between December 2009 and March 2010, commercial rabbit food was provided every three days (450 g) at some sites and no food was provided at other sites. The number of sites where food was provided is unclear.Study and other actions tested
A replicated study in 1997–2003 in forest, meadows and farmland in a mountain range in central and southern Slovenia (Jerina 2012) found that in areas where supplementary food was provided, the home-range of red deer Cervus elaphus was smaller. Red deer had smaller home ranges in areas where more supplementary feeding occurred (data expressed as model results). Between 1997 and 2003, twenty-five adult female and 17 adult male red deer were caught across a 2,100 km2 study area. Deer were radio-collared and released, and were relocated at least once a week, during all daylight hours, for at least one year. Annual home range size was estimated for each individual for each full year that it was monitored (total = 73 deer-years from 42 animals). Information on the location of supplementary feeding sites, and the type and quantity of food provided, was collected from a national register of feeding sites and used to model deer home-ranges alongside other relevant variables.Study and other actions tested
A replicated, controlled study in 1999–2011 in 10 tundra sites in Norway and Sweden (Angerbjörn et al. 2013) found that the number of artic fox Vulpes lagopus litters increased after supplementary winter feeding at den sites, along with control of red foxes Vulpes vulpes. At two sites where an average of 11–13.5 dens were fed, both the number of active arctic fox dens in winter, and the number of litters produced in summer, increased more than at sites where no feeding or a low level of feeding was undertaken (data reported as statistical model results). During winter 1999–2011, commercial dog food or remains from slaughtered reindeer Rangifer tarandus was provided to a large number of arctic fox dens (11–13.5) at two sites, where red foxes were also intensively culled in winter. At four other sites, low numbers of arctic fox dens (1–3) were provided with food, and low numbers of red foxes were culled (0–7). At the remaining four sites, no food was provided and no red foxes were culled (3 sites) or intensive culling was conducted (92 animals, 1 site). The number of arctic fox litters was counted in known arctic fox dens during July and August 1999–2011.Study and other actions tested
Referenced paperAngerbjorn A., Eide N.E., Dalén L., Elmhagen B., Hellström P., Ims R.A., Killengreen S., Landa A., Meijer T., Mela M., Niemimaa J., Norén K., Tannerfeldt M., Yoccoz N.G. & Henttonen H. (2013) Carnivore conservation in practice: replicated management actions on a large spatial scale. Journal of Applied Ecology, 50, 59-67.
A replicated, controlled study in 2007–2013 in four forested mountain areas in Wyoming, USA (Jones et al. 2014) found that elk Cervus canadensis provided with supplementary food migrated shorter distances and spent less time on their summer feeding grounds than unfed elk. Elk provided with supplementary food in winter migrated shorter distances (35.4 km) than did unfed elk (54.6 km). Fed elk arrived at their summer range an average of five days later and left 10 days earlier than did unfed elk. More fed elk used stopover sites on spring (56% of elk) and autumn (49% of elk) migration than non-fed elk (48% and 42% of individuals). Two hundred and nineteen adult female elk were caught and fitted with GPS radio-collars between January and March 2007–2011 at 18 sites where supplementary food was provided and at four sites with no supplementary food. Sites were located in four mountain areas within elk winter ranges. Supplementary feeding began when elk started to congregate at feeding sites and ceased once most elk had departed. GPS locations were taken from the elk every 30–60 minutes, for 1–2 years. Fed and unfed elk were monitored for 164 and 116 elk-years, respectively. The precise number of fed and unfed elk monitored is not detailed.Study and other actions tested
A review of evidence within studies looking at effects of feeding wild ungulates in North America (48 studies), Fennoscandia (25 studies) and elsewhere in Europe (28 studies) (Milner et al. 2014) found that supplementary feeding increased ungulate survival, reproductive rates or condition in varying proportions of studies. Ungulate survival rates increased in four out of seven relevant studies. The reproductive rate increased in five of eight relevant studies. Birth mass increased in one of three relevant studies. Loss of mass in winter was reduced or winter condition improved in five of seven relevant studies. Autumn mass increased in three of 11 relevant studies. Autumn mass or condition of offspring was improved in four of six relevant studies. Carrying capacity was increased in all three relevant studies. The review reported evidence from 101 studies that met predefined criteria from an initial list of 232 papers and reports.Study and other actions tested
A before-and-after, site comparison study in 2007–2013 of a conservation park and a game park in South Africa (Yarnell et al. 2015) found that when carrion was provided at a vulture feeding station, there were more brown hyaena Hyaena brunnea and black-backed jackal Canis mesomelas scats in that area. At the vulture station site, there were more hyaena scats in the final year of carrion provision (5.0 scats/km) than before carrion provision (2.6 scats/km) and over the two years after carrion provision ceased (1.5–2.0 scats/km). Scat counts remained more stable over this period at a site without a vulture feeding station (3.2–4.3 scats/km). Similarly, there were more jackal scats at the vulture feeding station in the final year of carrion provision (3.3 scats/km) than before (0.5 scats/km) or over two years after (1.5–2.0 scats/km) carrion provision. Scat counts remained low (0.2–1.4 scats/km) at a site without a vulture feeding station. A vulture restaurant was operated at a conservation park from March 2008 to August 2011. Predator density at this park, and on a game park where carrion was not provided, was monitored by annual scat transects from 2007–2013.Study and other actions tested
A site-comparison study in 2011 along two rivers in Austria (Sittenthaler et al. 2015) found that on a river stocked with fish for angling, densities of resident adult Eurasian otters Lutra lutra were not higher than those on an unstocked river. No statistical analyses were performed. Resident adult otter density on the stocked river (0.23 otters/km) was similar to that on the unstocked river (0.22 otters/km). However, including juvenile and non-resident otters, a slightly higher density was found on the stocked river (0.37 otters/km) than on the unstocked river (0.33 otters/km). Two river stretches, with similar hydromorphology, were studied. One (21.5 km long) was stocked with fish from a hatchery in April–September each year. The other (18.3 km long) was not stocked. Otter spraints were collected daily for five days during three visits from February–April 2011. Individual otters were identified by genetic analysis of faeces. Forty-eight faeces were successfully used to genetically identify individuals from the stocked river and 33 from the unstocked river.Study and other actions tested
A replicated study in 2008–2015 in a mountain forest and grassland site in the northeast Carpathians, Poland (Selva et al. 2017) found that supplementary feeding of ungulates altered brown bear Ursus arctos behaviour. Bears encountered feeding sites more frequently (GPS-tracked bears: 0.15 sites/km; snow-tracked bears: 0.93 sites/km) than expected at random (0.05 sites/km). From 2008–2010, a complete inventory of 212 ungulate feeding sites in the 1,500 km2 study area was compiled through interviews with land managers and field inspections. Feeding occurred regularly, often year-round but especially in autumn and winter, and usually in the same location for decades. In spring and autumn 2008–2009 and 2014–2015, nine bears were captured and fitted with GPS collars. Bear locations were recorded every 30 minutes for five days at the start of each month, and used to create 49 GPS-tracks (average 34 km long). From December–March 2010–2012, 40 snow tracks of unmarked bears longer than 500 m were recorded (average 6 km long). To determine what would be expected if movements were at random, for each of the 49 GPS tracks recorded, 100 random tracks were created using the same start point and number of locations, and by randomly choosing the distance travelled and angle turned between points.Study and other actions tested