Understanding how raptors select prey is important to determine taphonomic biases both in modern and paleo pellet assemblages. We tested whether pellets more closely represent raptor dietary specialization or local small mammal diversity by sampling pellets from seven raptor species across four study sites in Yellowstone National Park, Wyoming, USA. We identified small mammal craniodental elements from each pellet and tested for differences among small mammal assemblages for each raptor species and study site. We found that reconstructed avian predator diets clustered significantly by site but not by predator species. Bray-Curtis diet dissimilarities were also significantly lower when comparing different raptor species within a site than when comparing the same raptor species across different sites. Our results suggest that raptors choose to eat a diversity of small mammal species close to their roosts rather than fly long distances to specialize on a particular prey species. Neontologists and paleoecologists alike can therefore be confident that raptor pellets faithfully represent local small mammal diversity.

Performance, water intake and feeding behaviour of two groups of 30 (trial 1), 40 (trial 2) or 50 (trial 3) weaned pigs offered either pellets or meal of the same formulation were compared.

Average daily weight gains (ADG) were higher for pigs given pellets rather than meal in trials 2 (413 v. 363 g/day,

P < 0·001) and 3 (356 v. 324 g/day, P < 0·05). Mean daily water intake (DWI) was higher with meal than with pellets but only during trial 1 (2·31 v. 1·65 I/day, P < 0·01).

The occupation time (ОT) and the number of animals using the feeder simultaneously (N) were higher when pigs were given meal rather than pellets, whatever the animal density: trial 1: 82·6 v. 69·9% (P = 0·05) and 3·8 v. 2·3 (P < 0·01); trial 2: 90·9 v. 77·9% (P > 0·05) and 5·2 v. 3·1 (P < 0·01); trial 3: 96·2 v. 83·6% (P < 0·05) and 5·9 v. 3·8 (P < 0Ό1). When using pellets, ОT and N were always significantly lower during the night than during the day but when using meal in groups of 40 and 50, ОT during the night was almost as high as during the day.

The greater the group size, the lower were ADG (both diets) and DWI (only with meal) and the higher were ОT and N (both diets). Furthermore, significant linear and curvilinear regressions of DWI, ОT and N according to time were calculated.

In conclusion, pigs need more time to eat meal than to eat pellets. Thus the number of pigs per feeder has to be adapted to the food presentation. Too high a number of pigs per feeder impairs feeding behaviour and eventually welfare, by preventing preferential diurnal feeding activity and this may affect productivity.