Eighteen years of ecological monitoring reveals multiple lines of evidence for tundra vegetation change

The Arctic tundra is warming rapidly, yet the exact mechanisms linking warming and observed ecological changes are often unclear. Understanding mechanisms of change requires long‐term monitoring of multiple ecological parameters. Here, we present the findings of a collaboration between government scientists, local people, park rangers, and academic researchers that provide insights into changes in plant composition, phenology, and growth over 18 yr on Qikiqtaruk‐Herschel Island, Canada. Qikiqtaruk is an important focal research site located at the latitudinal tall shrub line in the western Arctic. This unique ecological monitoring program indicates the following findings: (1) nine days per decade advance of spring phenology, (2) a doubling of average plant canopy height per decade, but no directional change in shrub radial growth, and (3) a doubling of shrub and graminoid abundance and a decrease by one‐half in bare ground cover per decade. Ecological changes are concurrent with satellite‐observed greening and, when integrated, suggest that indirect warming from increased growing season length and active layer depths, rather than warming summer air temperatures alone, could be important drivers of the observed tundra vegetation change. Our results highlight the vital role that long‐term and multi‐parameter ecological monitoring plays in both the detection and attribution of global change. Arctic tundra is responding rapidly to climate change (IPCC 2014). Annual temperatures in the Arctic have already increased by more than 2°C since 1978 (IPCC 2013) and there is growing evidence that this warming is leading to changes in vegetation communities at sites around the tundra biome (Myers‐Smith et al. 2011a, Elmendorf et al. 2012b, IPCC 2014). Although the literature reports prominent ecological transformations, including changes in vegetation composition (Elmendorf et al. 2012b), plant phenology, and satellite‐observed greening (Post et al. 2009), recent studies also indicate high unexplained heterogeneity in vegetation responses at sites around the Arctic (Elmendorf et al. 2012b, Oberbauer et al. 2013, Guay et al. 2014, Myers‐Smith et al. 2015a, Prevéy et al. 2017). One possible reason that tundra biome‐scale data syntheses indicate high levels of heterogeneity is that different ecological parameters are measured in different places or at different points in time. Rarely are single sites monitored to a degree at which we can observe patterns of change over time from multiple lines of evidence (Hobbie et al. 2017). This omission also often precludes