A field study was made of the tiller dynamics of the New Zealand perennial ryegrass (Lolium perenne L.) cultivars Grasslands Ruanui, selected from a Hawke's Bay ecotype, and Ellett, selected from the Mangere ecotype. The work tested whether nitrogen (N) fertiliser or irrigation (Irr) could manipulate the persistence strategies of these morphologically distinct perennial ryegrass ecotypes, represented by these cultivars. Tiller births and deaths, and tiller survival, were monitored over 2 years, under rotational grazing by dairy cows. Patterns of cultivar tiller birth and death were similar. Much of the temporal change in tiller population density was shown to be a response to increasing or decreasing herbage mass associated with seasonal feed surplus or deficit. The probability of tiller survival between observation dates varied markedly with date, and was lowest in autumn. The cultivar effect on tiller survival was small but a significant (P < 0.05) cultivar × date interaction was detected, with survival probability for Ellett tending to be higher than for Grasslands Ruanui in March–April 1997 but lower from July 1997 until March 1998. A cohort effect on tiller survival was also identified, in that tillers formed in January, July, and August 1997 had lower survival probability than those formed at other times. N-treated plots produced 34% more tillers than did untreated plots, and irrigation had no effect on overall tiller density. N and Irr treatments had variable effects on tiller survival. Size density compensation theory and analysis of survival probability are introduced to aid the interpretation of tiller dynamics data.

Disturbance to wildlife from human recreational activities is increasing as remote areas become accessible to greater numbers of people. We used Global Positioning System tracking collars to monitor the movements of red deer (Cervus elaphus) stags (n=8) in a herd whose feeding grounds lie close to a popular walking track in the Highlands of Scotland. The track is used by around 20,000 walkers per year and is busiest in summer and at weekends. In a 2-year study, the locations of collared deer were recorded at 2-h intervals on typically busy days (Sundays: mean number of walkers=204) and quiet days (Wednesdays: mean number of walkers=49) during May and June. The deer were consistently further from the track on Sundays than Wednesdays (371 vs 286 m) and moved greater distances between fixes (365 vs 308 m). The amount of time spent (percentage of total fixes) in the small area of grassland closest to the track was lower on Sundays than Wednesdays (6% vs 13%). Although 97% of walkers use the track during the day (between 0800 and 2000 h), there was no evidence of compensatory use of grassland at night, when the deer moved to higher ground dominated by heather moorland. The results demonstrate that animals which appear to be habituated to regular disturbance within their home territory may nevertheless alter their behaviour and potentially diet composition, as a result of that disturbance.

Understanding how habitat selection changes with population density is a key concept in population regulation, community composition and managing impacts on biodiversity and ecosystem services. At low density, it is expected that individuals select habitats in terms of their preference, but as population density increases, the availability of resources per individual declines on preferred habitats, leading to competition which forces some individuals to exploit less preferred habitats. Using spatial information of Scottish red deer (Cervus elaphus) winter counts, carried out in 110 areas across Scotland between 1961 and 2004 (a total of 1,206,495 deer observations), we showed how winter habitat niche breadth in red deer has widened with increasing population density. Heather moorland and montane habitats were most and least preferred for deer, respectively. Increasing density favoured the selection of grassland, to the detriment of the selection of heather moorland. The selection of heather and grassland decreased when temperature increased, while the selection of montane and peatland habitats increased. These findings are important for understanding how habitat use, density and population are likely to be affected by weather, and allow us to predict habitat impacts by large mammal herbivory and climate.

Sources of variation in milk protein composition were determined after approximately 203 days-in-milk for spring-calving Friesian cows grazing ryegrass-white clover pastures in New Zealand. Three groups each of 15 multiparous cows were managed as separate herds. Each herd was subjected to one of three management strategies: pasture only, pasture supplemented with 2-3 kg cow 1 d -1 maize grain, and pasture plus 5-6 kg cow -1 d -1 maize grain and 3-4 kg cow -1 d -1 pasture silage. During a 3-week period beginning in February 1994, milk samples were collected twiceweekly from individual cows and the detailed protein composition determined. Variation was observed in milk yield and the concentrations of all protein components measured, with betweencow variation being greater than within-cow variation for all parameters except non-protein N and urea. Supplementing pasture with maize grain and/or silage increased milk yield but had only minor effects on protein composition. These data will enable increased efficiency of experimental design in this research area, and provide details of the natural variation in milk protein composition. This variation gives an indication of the extent that *Author for correspondence A98046 Received 9 September 1998; accepted 7 April 1999 milk protein composition could be manipulated by on-farm practices, or through the selection of cows with defined milk protein profiles.

No abstract