Background: Long-term climate trends in mountain systems often vary strongly with elevation. Aims: To evaluate elevation dependence in long-term precipitation trends in subalpine forest and alpine tundra zones of a mid-continental, mid-latitude North American mountain system and to relate such dependence to atmospheric circulation patterns. Methods: We contrasted 59-year (1952–2010) precipitation records of two high-elevation climate stations on Niwot Ridge, Colorado Front Range, Rocky Mountains, USA. The sites, one in forest (3022 m a.s.l.) and the other in alpine tundra (3739 m), are closely located (within 7 km horizontally, ca. 700 m vertically), but differ with respect to proximity to the mountain-system crest (the Continental Divide). Results: The sites exhibited significant differences in annual and seasonal precipitation trends, which depended strongly on their elevation and distance from the Continental Divide. Annual precipitation increased by 60 mm (+6%) per decade at the alpine site, with no significant change at the subalpine site. Seasonally, trends at the alpine site were dominated by increases in winter, which we suggest resulted from an increase in orographically generated precipitation over the Divide, driven by upper-air (700 hPa) north-westerly flow. Such a change was not evident at the subalpine site, which is less affected by orographic precipitation on north-westerly flow. Conclusions: Elevation dependence in precipitation trends appears to have arisen from a change in upper-air flow from predominantly south-westerly to north-westerly. Dependence of precipitation trends on topographic position and season has complex implications for the ecology and hydrology of Niwot Ridge and adjacent watersheds, involving interactions among physical processes (e.g. snowpack dynamics) and biotic responses (e.g. in phenologies and ecosystem productivity).

Eddy covariance measurements of the surface energy balance and carbon dioxide exchange above high-elevation (3,480 m above sea level) alpine tundra located near Niwot Ridge, Colorado, were compared to simultaneous measurements made over an adjacent subalpine forest over two summers and one winter, from June 9, 2007 to July 3, 2008. The surface energy balance closure at the alpine site averaged 71 and 91%, winter and summer, respectively, due to the high wind speeds, short turbulent flux footprint, and relatively flat ridge-top location of the measurement site. Throughout the year, the alpine site was cooler with higher relative humidity, and had a higher horizontal wind speed, especially in winter, compared to the forest site. Wind direction was persistently downslope at the alpine site (summer and winter, day and night), whereas upslope winds were common at the forest site during summer daytime periods. The latent and sensible heat fluxes were consistently larger in magnitude at the forest site, with the largest differences during summer. The horizontal advective flux of CO2 at the alpine site averaged 6% of the net ecosystem exchange (NEE) during summer nights (5% during summer daytime), and was small in relation to the high wind speeds, relatively flat site, and weak sources of CO2 upwind of the site. The magnitudes and diurnal behavior of the alpine NEE calculated using three methods; eddy-covariance, friction velocity filter, and with advection and storage calculations, gave similar results. The period of net CO2 uptake (negative NEE) was 100 days at the alpine site with a net uptake of 16 g C m−2, compared to 208 days at the forest site with a net uptake of 108 g C m−2, with initiation of net uptake coinciding with air temperatures reaching +10°C. Winter respiration loss at the alpine site was 164 g C m−2 over 271 days, compared to 52 g C m−2 over 175 days at the forest site, with the initiation of net loss coinciding with air temperatures reaching −10°C at each site.

We describe the process by which the Long-Term Ecological Research (LTER) Network standardised their metadata through the adoption of the Ecological Metadata Language (EML). We describe the strategies developed to improve motivation and to complement the information technology resources available at the LTER sites. EML implementation is presented as a mapping process that was accomplished per site in stages, with metadata quality ranging from 'discovery level' to rich-content level over time. As of publication, over 6000 rich-content standardised records have been published using EML, potentially enabling the goal of machine-mediated, metadata-driven data synthesis.