chased. We initially tried a number of dif- 

 ferent roundup methods, and one of the 

 simplest proved the most effective. On 

 rainy nights, the sheep take shelter in the 

 cleits; by moving very quietly, we were 

 able to block the entrance before any could 

 escape. Then one of our team would crawl 

 through the low entrance into the cleit with 

 a flashlight, grab a sheep, and drag it out to 

 the open, where it could be weighed, mea- 

 sured, and have its blood sampled. Some- 

 times after crawling down the low, muddy 

 entrance into pitch blackness and switch- 

 ing on his helmet lamp, a catcher would 

 confront a ram with its head down, ready 

 to charge directly at the Ught. 



We found that the least painful method 

 of capture was a large-scale netting opera- 

 tion. With volunteers from the Mammal 

 Conservation Trust, who are experienced 

 in netting deer, we learned to build corrals 



of netting, well hidden behind the dereUct 

 cottages of the village street; to erect hun- 

 dreds of feet of side nets around the mead- 

 ows where the sheep collect; and then to 

 slowly ease the sheep up the tunnels of 

 netting into the corral, where they could be 

 caught and weighed. This way, we eventu- 

 ally trapped enough sheep to allow us to 

 compare the weights of mothers that had 

 raised no lambs with those that had raised 

 singletons or twins. As predicted, all three 

 categories of mothers proved to be of sim- 

 ilar weight in August — two full months 

 before the rut — showing that mothers are 

 able to regain weight lost during lactation 

 in the two months following the weaning 

 of their lambs. This contrasts strongly 

 with red deer on Rum, where mothers that 

 have raised calves are still in poor condi- 

 tion in September. 

 So what does our understanding of 



sheep cycles on Saint Kilda tell us about 

 cycling in other ungulates? The features of 

 the Soay sheep population that create cy- 

 cles are the high rate of population in- 

 crease (caused by first-year breeding, low 

 juvenile mortality, and no dispersal) and 

 the absence of any strong effect of popula- 

 tion density on fecundity and lamb mortal- 

 ity (fostered by the superabundance of 

 food in early summer and by early wean- 

 ing). This combination of factors is not 

 common in ungulates. Most ungulate fe- 

 males do not conceive until their second, 

 third, or fourth year of hfe; twinning is 

 rare; and neonatal mortahty is high. As a 

 result, unlike Soay sheep, populations of 

 other ungulates carmot exceed by a large 

 margin the number of animals that the 

 winter food supply can support. 



Some wild ungulates do parallel the 

 sheep's situation, however. The Saiga an- 



33 



