Drivers of change

6.2.1 Thaw lakes

Thaw lakes are common features in ice-rich permafrost regions, and their initiation is generally associated with a warm climatic period (e.g. MacDonald et al. 2006). Their drainage, however, can be linked to drivers that are not linearly connected to climate (Jones et al. 2011). In this thesis, observed changes in thaw lakes related to lake deepening (Chapter 3) and lake drainage (Chapters 4, 5). The deepening of Roland Lake recorded in Chapter 3 happened within a few years and its timing was linked to climatic warming. In ice-rich unconsolidated sediments increased thaw and ground subsidence may cause incipient shallow lakes to become deeper and the water-level to rise (e.g. Jorgenson & Shur 2007). Lake drainage, which has provided the basis for ice-wedge polygon development at all studied sites (Chapters 4, 5), is, however, not necessarily associated with climatic forcing. On Herschel Island, the timing and drivers of lake drainage are well constrained (description and discussion in Chapter 4). The lake drained at about 4000 cal. yrs BP as a consequence of gully incision triggered by erosion of nearby coastal bluffs. Similar geomorphic drivers of localized lake drainage have been identified along unconsolidated coasts around the Arctic (Romanovskii et al. 2004, Hinkel et al. 2007, Mars & Houseknecht 2007, Marsh et al. 2009).

One of the most frequent reasons for thaw lake drainage is melting of ice wedges on the lakeshores, which at some point provide drainage pathways (Marsh et al. 2009, Jones et al.

2011). This process likely caused at least two of the lakes that existed at the modern ice- wedge polygon sites studied in Chapter 5 to disappear. Thus, ice-wedge degradation and coastal erosion, both of which are rapid and climate-sensitive geomorphic processes, were the main drivers of lake drainage on the Yukon Coastal Plain during the Late Holocene. Both

processes are also climate-sensitive, however, and climatic warming may contribute significantly to increased thaw lake drainage.

6.2.2 Ice-wedge polygons

The results from this thesis indicate that the main drivers of change in ice-wedge polygons were alterations in drainage regime, some of them warming-induced. This may have been caused by landscape-scale geomorphic change such as stream incision or mass movements altering pathways of surface water flow (e.g. Rowland et al. 2010) or by relief inversion through ice wedge melt (Liljedahl et al. 2016) (for a comprehensive discussion of drivers of ice-wedge polygon initiation and development see Chapters 4, 5). It is also hypothesized that internal self-organization through lateral material displacement (Mackay 2000) or through gradual peat accumulation in polygon centres (Ellis & Rochefort 2006) may cause conversion into high-centred polygons. Although present, these two processes were not the main drivers for the conversions found in this study. All four polygons investigated in this thesis emerged from incipient shallow lake environments, developing first into low-centred ice-wedge polygons with wet to partly submerged conditions before experiencing improved drainage in the twentieth century, which led to intermediate- and high-centred polygons at two elevated sites (Chapter 5). High-centred polygons are known to occupy elevated sites with some drainage, while low-centred polygons are found in depressed low-lying sites with impeded drainage (Rampton 1982, Schirrmeister et al. 2011b). The findings of this thesis suggest that the intermediate-centred and high-centred polygons only experienced improved drainage conditions during the twentieth century (Chapter 5), indicating a recent shift in relative relief and landscape hydrological conditions during the period of modern warming. Both polygons are also currently situated on elevated sites close to lakeshores. The deepening of one of these lakes at the beginning of the twentieth century is documented in Chapter 3. Recent increased thermokarst activity accompanied by deepening of thaw lakes may have contributed to draining the polygons. Stream incision and coastal erosion are also contributing to changing drainage pathways and thus promoting the development of high-centred polygons. Such climate-induced geomorphic change is increasingly reported from the Arctic (Rachold et al.

2000, Hinzman et al. 2005, Mars & Houseknecht 2007, Lantuit & Pollard 2008, Günther et al.

2013) and is currently altering landscape water balance and flow paths, which in turn determine ice-wedge polygon type. Ice wedges and the surrounding permafrost responded rapidly even to low-amplitude climatic change on short time-scales in the studies conducted for this thesis (Chapters 2, 4, 5).

6.2.3 Vegetation

Vegetation reconstruction in ice-wedge polygons revealed a broad regional development from aquatic to wetland taxa on centennial to millennial time-scales. A change towards mesic vascular plant taxa then happened during the twentieth century (Chapters 4, 5). The former development was locally variable, gradual and slow, while the latter appeared synchronous and much more rapid, acting on decadal time-scales and coinciding with climatic warming.

While the direct cause of vegetation change was ice-wedge melt and the resulting changes to ice-wedge polygon morphology, this geomorphic change was indirectly warming-induced.

This thesis also showed that vegetation patterns were strongly influenced by microtopography. In ice-wedge polygon environments, even a few centimetres of elevation differences may provide different microhabitats in terms of water availability (Chapter 2).

These highly structured small-scale vegetation mosaics experienced drying during the twentieth century, which was reconstructed from vascular plant macrofossils and sediment parameters in peaty sediment cores (Chapters 4, 5). This signal was not reflected in the 300- year regional vegetation record from pollen and sediment parameters in a lake sediment core (Chapter 3). The core showed regional vegetation stability with only a slight indication of shrub increase during recent decades. This discrepancy highlights the diverging interpretative scope of different archives and proxies. Plant macrofossils in peat cores generally capture a highly localized signal, while pollen in lake sediment is more likely to reflect the regional vegetation. Arguably, pollen analysis has a limited taxonomic resolution and environmental reconstruction value in tundra environments (Birks & Birks 2000, Gajewski 2006). The combination of a set of local records with a regional one may allow a much more comprehensive environmental reconstruction than either approach on its own. The discrepancy also reflects upon differences in climate-sensitivity of local- and regional-scale change. Low-amplitude climatic fluctuations such as the Little Ice Age trigger rapid and localized geomorphic change by altering permafrost conditions. This may force a rapid vegetation response, while climatic change alone may be buffered within ecosystems, and the vegetation response may be strongly delayed (e.g. Davis 1989) and hard to predict. High- amplitude climate change as reported from the Pleistocene-Holocene transition (e.g. Andreev et al. 2002, Payette et al. 2002) may, however, cause a regional-scale vegetation response on long time-scales.