Part of a large male woolly mammoth (Mammuthus primigenius) was preserved in permafrost in northern Yakutia. It was radiocarbon dated to ca. 18,50014C yr BP (ca. 22,500 cal yr BP). Dung from the lower intestine was subjected to a multiproxy array of microscopic, chemical, and molecular techniques to reconstruct the diet, the season of death, and the paleoenvironment. Pollen and plant macro-remains showed that grasses and sedges were the main food, with considerable amounts of dwarf willow twigs and a variety of herbs and mosses. Analyses of 110-bp fragments of the plastid rbcL gene amplified from DNA and of organic compounds supplemented the microscopic identifications. Fruit-bodies of dung-inhabiting Ascomycete fungi which develop after at least one week of exposure to air were found inside the intestine. Therefore the mammoth had eaten dung. It was probably mammoth dung as no bile acids were detected among the fecal biomarkers analysed. The plant assemblage and the presence of the first spring vessels of terminal tree-rings of dwarf willows indicated that the animal died in early spring. The mammoth lived in extensive cold treeless grassland vegetation interspersed with wetter, more productive meadows. The study demonstrated the paleoecological potential of several biochemical analytical techniques.

The retinas of placental mammals appear to lack the large number and morphological diversity of cone subtypes found in diurnal reptiles. We have now studied the photoreceptor layer of a South American marsupial (Didelphis marsupialis aurita) by peanut agglutinin labeling of the cone sheath and by labeling of cone outer segments with monoclonal anti-visual pigment antibodies that have been proven to consistently label middle-to-long wavelength (COS-1) and short-wavelength (OS-2) cone subpopulations in placental mammals. Besides a dominant rod population (max. = 400,000/mm2) four subtypes of cones (max. = 3000/mm2) were identified. The outer segments of three cone subtypes were labeled by COS-1: a double cone with a principal cone containing a colorless oil droplet, a single cone with oil droplet, and another single cone. A second group of single cones lacking oil droplets was labeled by OS-2 antibody. The topography of these cone subtypes showed striking anisotropies. The COS-1 labeled single cones without oil droplets were found all over the retina and constituted the dominant population in the area centralis located in the temporal quadrant of the upper, tapetal hemisphere. The population of OS-2 labeled cones was also ubiquitous although slightly higher in the upper hemisphere (200/mm2). The COS-1 labeled cones bearing an oil droplet, including the principal member of double cones, were concentrated (800/mm2) in the inferior, non-tapetal half of the retina. The two spectral types of single cones resemble those of dichromatic photopic systems in most placental mammals. The additional set of COS-1 labeled cones is a distinct marsupial feature. The presence of oil droplets in this cone subpopulation, its absence in the area centralis, and the correlation with the non-tapetal inferior hemisphere suggest a functional specialization, possibly for mesopic conditions. Thus, sauropsid features have been retained but probably with a modified function.