Volume 12, 2022 Journal of the North Atlantic No. 43 The Environmental History of Skeiðarársandur Outwash Plain, Iceland Thóra Ellen Thórhallsdóttir and Kristín Svavarsdóttir Journal of the North Atlantic The Journal of the North Atlantic (Online ISSN #1935-1933, Print ISSN #1935-1984), with an international editorial board, is a collaborative publishing effort of the Eagle Hill Institute, PO Box 9, 59 Eagle Hill Road, Steuben, ME 04680- 0009 USA. Phone 207-546-2821, FAX 207-546-3042. E-mail: office@eaglehill.us. Website: www.eaglehill.us/jona. Copyright © 2021, all rights reserved. On-line secure subscription ordering: rate per year is $40 for individuals, $32 for students, $250 for organizations. Authors: Instructions for authors are available at www.eaglehill.us/jona. The Eagle Hill Institute (Federal ID # 010379899) is a tax exempt 501(c)(3) nonprofit corporation of the State o f Maine, USA. The Journal of the North Atlantic (JONA) is a multi-disciplinary, peerreviewed and edited archaeology and environmental history journal focusing on the peoples of the North Atlantic, their expansion into the region over time, and their interactions with their changing environments. The journal—published online in the BioOne.org database and on the JONA website, and indexed in a full range of journal databases—serves as a forum for researchers, and as an information resource for instructors, students, and the intellectually curious who would like to learn about the latest research and study opportunities within the region. The journal publishes a wide diversity of research papers, as well as research summaries and general interest articles in closely related disciplines, which, when considered together, help contribute to a comprehensive multidisciplinary understanding of the historical interplay between cultural and environmental changes in the North Atlantic world. Specifically, the journal’s focus includes paleo-environmental reconstruction and modelling, historical ecology, archaeology, ecology of organisms important to humans, anthropology, human/environment/climate interactions, climate history, ethnography, ethnohistory, historical analyses, discussions of cultural heritage, and placename studies. The journal publishes individual papers on an article-by-article basis. Whenever a manuscript has completed its peer review process and the article galley has been approved by the author, it will be immediately published online in the BioOne database and on the JONA website. This publishing model is also available for special volumes such as conference and symposium proceedings or other collections of papers. In effect, this means that articles are grouped online over time, i.e., the table of contents of volumes will grow as articles are posted online, which has the advantage of rewarding prompt authors, while enabling tardier authors to retain the option of being included in a special volume without delaying its publication. The Journal of the North Atlantic’s publishing format is versatile enough that authors can include supplementary files with their articles. These supplements may include dataset, figure, and table files (e.g., files requiring a larger than normal journal page size, such as large maps), as well as text and protocol files, audio and video files (e.g., for ethnographic studies), and even Powerpoint files. The Journal of the North Atlantic is indexed in the Web of Science (clarivate. com), EBSCO.com, and by way of author entries in Google Scholar and Researchgate. It is included in full-text in BioOne.org and JSTOR.org. Board of Editors Jette Arneborg, Denmark Gerald F. Bigelow, Scotland, UK Steven A. Birch, Scotland, UK Colin Breen, Northern Ireland Mike J. Church, England, UK Christyann Darwent, USA Jane Downes, Scotland, UK Andrew J. Dugmore, Scotland, UK Mark Gardiner, England, UK Erika Guttmann-Bond, The Netherlands Agnar Helgason, Iceland Joerg-Henner Lotze, USA, Publisher Niels Lynnerup, Denmark Thomas H. McGovern, USA Helgi D. Michelsen, Faroe Islands Jacqui A. Mulville, Wales, UK Anthony Newton, Editor Georg Nyegaard, Greenland Ulla Odgaard, Denmark Astrid E.J. Ogilvie, USA Tadhg O'Keeffe, Ireland Bjørnar Olsen, Norway Richard D. Oram, Scotland, UK Michael Parker-Pearson, England, UK Else Roesdahl, Denmark Alexandra Sanmark, England, UK Niall Sharples, Wales, UK Ian A. Simpson, Durham, UK, Przemyslaw Urbanczyk, Poland Orri Vésteinsson, Iceland Alex Woolf, Scotland, UK James Woollett, Canada Cover Image: View from Sel farm in Skaftafell westwards across Skeiðarársandur. From right to left in the background are Skeiðarárjökull glacier, mount Lómagnúpur and in the far distance the mountains of Fljótshverfi and Síða. Photograph © Þóra Ellen Þórhallsdottir, 23rd June 2012. “ Skálholt Map” courtesy of The Royal Library, Copenhagen, Denmark Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 1 Introduction Regime shifts, multiple or alternative stable states, and thresholds are concepts that relate to the transformation of an ecosystem from a previous to a radically different state (Scheffer et al. 2001, Scheffer and Carpenter 2003). Regime or state shifts are most commonly used to describe the conversion of a productive ecosystem to a less productive or barren state, e.g., the relatively abrupt shift of the Sahara from an early-mid Holocene vegetated state to dry desert (deMenocal et al. 2000). Sometimes, a gradual change in an important variable, such as precipitation in the case of the Sahara, may suddenly push the system over a critical threshold. The shift may also be due to a single stochastic event, for example insect outbreaks that, together with climatic fluctuations, are believed to have lowered the treeline in the Fennoscandian mountains (Holtemeier and Broll 2006), with attendant changes in the understory vegetation (Jepsen et al. 2013). Internal ecosystem dynamics may trigger shifts, such as trophic cascades brought about by the removal or addition of a top predator (Beschta and Ripple 2012) or megafaunal extinctions (Barnosky et al. 2016). At present, most regime shifts are caused directly or indirectly by humans (Hughes et al. 2013), often through overexploitation of biological resources (D‘Odorico et al. 2013). Regardless of the cause, ecosystem state shifts are significant, spatially-extensive events with a long-lasting imprint and often far-reaching consequences for human societies, for example, affecting the production of food (Rocha et al. 2015). The new and often undesirable state is stable, and reversal back to the original state is slow, if it happens at all. For recovery to occur, it may be necessary to move the important external variables well beyond the state that triggered the transition in the first place (hysteresis), often requiring direct human intervention through ecological restoration. Few regions of the globe rival Iceland when it comes to diversity of environmental disturbances and their intensity, frequency, and spatial dimensions. Within Iceland, the southeast has the most extreme disturbance regime. Here is the most active part of the volcanic zone, the outlet glaciers most responsive to climatic fluctuations, and large glacial rivers that meander in shifting courses across wide outwash plains, sandur in Icelandic. Here the lowlands are also at greatest risk from outburst floods and the farming communities that suffered most from subglacial eruptions and that were most directly impacted by glacier advance during the Little Ice Age. The history of the 1000 km² Skeiðarársandur outwash plain and its adjacent farming community of Öræfi is a dramatic example of a marginal environment where deteriorating climate in conjunction with intensive, recurrent disturbances drive large scale ecological destruction. At present, parts of the plain are undergoing rapid vegetation succession and also serve to illustrate how fast ecosystems may re-establish once conditions have been ameliorated. The Environmental History of Skeiðarársandur Outwash Plain, Iceland Thóra Ellen Thórhallsdóttir1* and Kristín Svavarsdóttir2 Abstract - We sketch the Holocene history of Skeiðarársandur outwash plain, southeast Iceland, but concentrate on postlandnam changes. The dramatic human history of the Öræfi farming community is well known, but for the first time, medieval cartularia and late 16th to early 20th century sources are combined to reconstruct the plain’s environmental history. We identify trends and agents that have allowed recent ecosystem recovery and decribe the zonation and characteristics of the present major ecosystems. Skeiðarársandur’s history represents a state shift in an extreme disturbance regime, but it is also set to become a rare example of subsequent recovery through natural processes, albeit indirectly caused by global warming. The plain’s eastern flank at least carried extensive birch forests and riparian meadows in the first centuries after settlement. The first documented catastrope was the A.D. 1362 Öræfajökull eruption, and from then on, increasingly desctructive glacial floods swept across Skeiðarársandur, some covering almost the entire 1000 km² plain. At least 11 farms were abandoned by 1500, and by the 18th century, the farming community west of Öræfajökull had been reduced from ≥20 to four farmsteads. By the late Little Ice Age, Skeiðarársandur was an exceptionally barren wasteland. Over the past 80 years, fewer and less destructive outburst floods, warming climate, and enhanced seed rain with greater species diversity have facilitated plant establishment and rapid vegetation succession in parts of the plain. In the absence of major disturbances, one of the largest natural birch forest in Iceland may develop on Skeiðarársandur. Journal of the North Atlantic 1Institute of Life and Environmental Sciences, University of Iceland, Sturlugata 7, IS-101 Reykjavík, Iceland, 354-525- 4607. 2Soil Conservation Service of Iceland, Árleynir 22, IS-112 Reykjavík, Iceland, 354-488-3094. *Corresponding author: theth@hi.is. Associate Editor: Andrew Dugmore, Institute of Geography, University of Edinburgh. Vol. 12, 2022 43:1–21 Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 2 Skeiðarársandur also constitutes a rare 21st century example of the large-scale recovery of a terrestrial ecosystem mostly through natural forces with minimal direct anthropogenic intervention. Our objective is to reconstruct the environmental history of Skeiðarársandur, distinguishing six major time periods (Fig. 1). We briefly sketch the first two: 1) from the Last Glacial Maximum (LGM) to the early Holocene and 2) from the Holocene Thermal Maximum (HTM) through Neoglaciation to the human settlement (Landnám) of Iceland. We concentrate on the post-Landnám period, i.e., 3) the early settlement history ~ A.D. 900–1362, through 4) the Little Ice Age (LIA, ca. A.D. 1250–1890, e.g., Larsen et al. 2011, Hannesdóttir et al. 2015), to 5) 20th century changes, and finally 6) the present environment. The early settlement history of Öræfi was described by Thórarinsson (1958, see also Ives 2007). Thórarinsson (1974) summarized historical records of subglacial eruptions and outburst floods on Skeiðarársandur, and Björnsson (2003) traced documentary evidence of changes in the rivers, notably Skeiðará. Ours is the first attempt at reconstructing the vegetation and environmental history of Skeiðarársandur. Reconstructions of past vegetation and environments are most commonly based on pollen analyses, occasionally on macrofossils (e.g., Mitchell 2011). Such an approach is not possible for Skeiðarársandur, where plant remains are very unlikely to have been preserved and are, at best, buried deep under sand and gravel. Instead, we rely on historical records for reconstruction of the post- Landnám to the 20th century environment. Regional Setting Southeast Iceland is dominated by the 7800 km² Vatnajökull glacier. Beneath it lies Iceland’s Figure 1. Schematic illustration of six major phases in the environmental history of Skeiðarársandur from the late glacial to the present day (phase 1 only shown back to 10 ka). A rough approximation of the relative size of Vatnajökull glacier is shown in blue, vegetation cover on Skeiðarársandur in green and the size of the human settlement on Skeiðarársandur and in Öræfi in orange. Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 3 most powerful geothermal area and its most active volcano. From the main ice cap, dozens of outlet glaciers descend onto the lowlands, giving rise to glacial rivers that cross wide outwash plains on their way into the Atlantic. Inhabited areas are mostly a narrow strip between the glacier and the sea. Cut off by a harbourless sandy coast, the ice cap to the north and the hazardous glacial rivers of Breiðamerkursandur to the east and Skeiðarársandur to the west, the farming community of Öræfi was among the most isolated in Iceland. Skeiðarársandur lies south of Skeiðarárjökull, Vatnajökull’s largest outlet glacier, and west of Öræfajökull stratovolcano (2010 m a.s.l.) (Fig. 2). The central margin of Skeiðarárjökull is now ~24 km from the sea. The upper part of the plain is ~30 km wide, but >40 km by the coast. Skeiðarársandur is flat and homogeneous (Fig. 3a–d). Excepting the 76 m high headland Ingólfshöfði by the coast, topographical features are limited to kettleholes, shallow dry floodbeds, and Leymus arenarius (L.) Hochst. (Lymegrass) dunes. On the northwest corner, the promontory Lómagnúpur rises 670 m above the plain (Fig. 3b). Beyond Lómagnúpur is Núpsstaður, the first farm west of Skeiðarársandur. To the east are the farms of Öræfi district on lower mountain slopes or at the base of Öræfajökull. Counting from north to south, the 20th century farmsteads west of Öræfajökull are Skaftafell, Svínafell, Sandfell, and Hof. South of Öræfajökull are Hofsnes, Fagurhólsmýri, and Hnappavellir, and finally Kvísker further east (Fig. 2). Five of these (Skaftafell, Svínafell, Hof, Fagurhólsmýri, and Hnappavellir) encompassed two to several independent farms. Skaftafell is now within Vatnajökull National Park, and Sandfell and Kvísker are not inhabited. Traditional farming is still practiced in Svínafell, Hof, Fagurhólsmýri, and Hnappavellir. Presently, three large rivers flow from Skeiðarárjökull. Skeiðará, originating at the eastern corner of Figure 2. Sentinel 2 satellite image of Skeiðarársandur and vicinity from September 6th 2017 with names of major landscape features. Names of 20th century farms are abbreviated, from left to right: M = Maríubakki, R = Rauðaberg, N = Núpsstaður, Sk = Skaftafell, Sv = Svínafell, Sa = Sandfell, H = Hof, Hn = Hofsnes, F = Fagurhólsmýri, Hn = Hnappavellir and K = Kvísker. Grænalón glacial lagoon is dry but its location is discernable as a darker colour. By July 2017, both Skeiðará and Súla rivers had joined Gígjukvísl in a single course. Inset: Vatnajökull glacier with subglacial Grímsvötn volcano in the middle. The approximate subglacial route of floods from Grímsvötn is shown as a dotted red line from the caldera to Skeiðarárjökull. Satellite image courtesy of the National Land Survey of Iceland. Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 4 the glacier, has been the major river on the plain since the 16th century at least, usually following a course due south along the eastern part of the plain. In 2009, it switched westwards, flowing along the pro-glacial depression to join river Gígjukvísl (Fig. 2). The westernmost river, Súla, moved its course in 2017 and also joined Gígjukvísl. Many rivers originate from Öræfajökull and flow across the eastern margin of Skeiðarársandur before turning southwards. For the past ca 800 years at least, glacial outburst floods (jökulhlaup in Icelandic) have swept across Skeiðarársandur. These have three different causes. Most floods originate at the subglacial Grímsvötn volcano, deep inside Vatnajökull ice cap (Fig. 2, inset). Meltwater continuously collects above the geothermal area and when it has reached a critical level, the water rushes through the glacier 50 km southwards before bursting out from the edge of Skeiðarárjökull (Björnsson, 2017). These floods may or may not be associated with abrupt melting of ice during eruptions in Grímsvötn. Second, floods may be caused by subglacial eruptions outside the Grímsvötn caldera, and some of the largest floods belong to this category. The third source is the draining of the marginal glacial lagoon at Grænalón, causing floods in Súla and Núpsvötn (Fig. 2). They are much smaller than the other flood types and confined to the westernmost part of the plain. Typical jökulhlaups from Grímsvötn have peak discharges of 0.6–50 x 10³ m³ sec-1 at the glacier margin, a duration of 2–30 days and a total volume of 0.5–4.0 km³ (Björnsson 2002). Estimated sediment load is 100–300 x 106 tons. Apparently, jökulhlaup frequencies have varied greatly in time. From the mid 19th century at least to the 1930s, they occurred at ≤10 yr intervals (Thórarinsson 1974), but between 1938 and 1996 there were only few and small floods. In the early 1970s, gravel dykes were constructed east of Skeiðará to protect farmland in Öræfi, extending from below Skaftafell farm ca 5 km south to the main road, with a second 3 km long dyke west of the river northwards from the main road. There are shorter dykes above the main road by Sæluhúsavatn (1.7 km long) and east of Núpsvötn (1 km) and Figure 3. a) Aerial view of upper part of Skeiðarársandur. Núpsvötn river with bridge in the foreground and in its pre-2017 course, Gígjukvísl with bridge right in the middle distance, left Skeiðarárjökull outlet glacier. The ice-covered Öræfajökull stratovolcano with outlet glaciers in the background. Beyond Gígjukvísl is the vegetated area of the upper part of the plain. b) The upper part of Skeiðarársandur looking west from Skaftafell, right are Skeiðarárjökull and Lómagnúpur mountain. c) Mountain Birch (Betula pubescens subsp. tortuosa) on the upper-central part of the plain (the vegetated area visible in a). d) The sandy flats typical of much of the central part of Skeiðarársandur. e) Low Lymegrass (Leymus arenarius) dunes in dry and unstable sand. Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 5 around the bridges. Except for those dykes in the uppermost part of the plain, the flow of water is not directed or regulated. In 2002, about 15% of Skeiðarársandur between Gígjukvísl and the old Skeiðará river course were moderately to well vegetated (>50% vegetation cover); 12% had a cover of 10–50%, but almost three quarters were very sparsely vegetated with <10% cover (Kofler 2004). Some of the land in the third class is very barren, with a plant cover of only 1–2% (Marteinsdóttir et al. 2010, 2013). The closest climate station with a 30-year record is at Fagurhólsmýri farm, by the southeast edge of Skeiðarársandur (Fig. 2). The 1961–1990 mean air temperature was above zero in all months and highest 10.5°C in July (Icelandic Meteorological Office 2021). Mean annual precipitation was ~1,800 mm, but this may be an overestimate for at least those parts of Skeiðarársandur that lie further from high mountains. The climate on Skeiðarársandur is relatively mild and moist. The regional growing season, estimated about 110 days (Marteinsdóttir et al. 2018), in among the longest in Iceland. Materials and Methods Last Glacial Maximum through Holocene Thermal Maximum up to Settlement Time Our reconstruction largely relies on general climate scenarios for Iceland from the LGM through the HTM to Neoglaciation. Little paleo-environmental reseach has been carried out in the southeast so we supplement with studies from other regions in Iceland. The Post-settlement Environment to the Late 20th Century We used six independent sources to shed light on the post-settlement, medieval Skeiðarársandur environment: 1) For the past evolution of Skeiðarárjökull, coupled models of ice dynamics and mass balance were applied to simulate ice cap geometry (given the bed mapped by radio-echo soundings and degree-day models describing climate changes) (Björnsson 2017). We then consider the implications of this for the disturbance regime of major rivers across the plain. 2) Place names and their subsequent changes, mostly based on comparisons of Íslendingabók, Landnámabók, and selected Sagas with later and contemporary names. 3) Descriptions in Landnáma. 4) Events described in ancient annals. 5) Information on Skeiðarársandur in sagas, notably Sturlunga and Njáls saga and in Biskupasögur. 6) Biskupaannálar Jóns Egilssonar, and last but not least 7) Church cartularia from the 12th to 16th centuries. Landnámabók (Landnáma; the Book of Settlement, see Benediktsson 1968) tells the story of the late 9th century settlement of Iceland. Believed written in the early 12th century, it is preserved in three manuscripts from the late 13th to early and mid 14th centuries. Sturlunga saga (Thorsson 1988) was at least partly written by Sturla Þórðarson in the late 13th century. Njáls saga (Sveinsson 1954) takes place around A.D. 1000, but was written by an unknown author ca 1280. Biskupasögur (The Bishops’ Tales) are largely contemporary 13th–14th century accounts of the lives and deeds of the early bishops (Egilsdóttir 2002 and 2012). Ancient annals of use here are Lögmannsannáll (spanning A.D. 292–1392, believed written in 1362–1392), Flateyjarannáll (A.D. 1044–1650, written in the 17th century), Annálsbrot frá Skálholti (A.D. 1329–1372, contemporary) and Gottskálksannáll (A.D. 1–1578, a late 16th century compilation of older manuscripts). For all annals, see Islandske Annaler indtil 1578 (published 1888). Biskupaannálar (The Bishops’ Annals), compiled by Jón Egilsson in 1605, relate the histories of bishops in Skálholt from the 11th century onwards (Sigurðsson 1856). The cartularia are contemporary inventories of church properties and of church rights to use land and resources, such as forests, and detail the dues individual farms had to pay to their church. The cartularia are accessible in Diplomatarium Islandicum (vol. 1, ed. Sigurðsson 1857–1876 and vol. 2, ed. Thorkelsson 1893). We consider the earliest reasonably detailed maps of Iceland from the late 16th to the 17 th centuries (which curiously have in common that the largest landscape feature in Iceland, Vatnajökull glacier, is missing). From the early 18th century, there are farm inventories, district descriptions and, by the end of the 18th century, information on the vascular flora. The 21st Century Environment Data on the present flora and vegetation of Skeiðarársandur were obtained from several sources: 1) An aerial and ground survey of vegetation on Skeiðarársandur in 1998 (Svavarsdóttir, Thórhallsdóttir, and Sparrow, Arthur Rylah Institute for Environmental Research, Heidelberg, Australia, 1998 unpubl. data); 2) the 1:50,000 vegetation map by Kofler (2004), based on Spot 5 images taken in 2002; 3) vascular species richness was recorded in 2004 and 2012 in 47 Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 6 There are no field studies from the southeast, but elsewhere in Iceland, RSL were depressed by up to 60 m a.s.l. during the Younger Dryas glacier advance. Rapid isostatic uplift left the lowest RSL at -44 m at about 10.0 cal kyrs BP. After that, sea level rose more or less continuously to present level. After its final Preboreal advance, the ice sheet retreated rapidly and by 10 cal. ka BP, much of Iceland had become ice free (Geirsdóttir et al. 2009). In the early Holocene, the sea extended up to Lómagnúpur, and all of the present Skeiðarársandur would have been below sea level. The earlier of the two major phases in the buildup of Skeiðarársandur took place in late glacial to early Holocene time. Seismic soundings revealed sediment thicknesses of 80–100 m in the upper part, increasing seawards to 200–250 m (Guðmundsson et al. 2002). This is similar to sediment thicknesses (235 m) beneath the lowest part of the outwash plain of Mýrdalssandur further west on the south coast, but much thicker than four coastal sites measured between Mýrdalssandur and Skeiðarársandur (106–156 m, Einarsson 1966). Two layers were identified on Skeiðarársandur. The uppermost unconsolidated layer was 70–150 m thick and this was considered to represent uncompacted Holocene deposition, in all ~100 km3. This is equivalent to an average rate of build-up of ~1 km3 per century (Guðmundsson et al. 2002). The Holocene Thermal Maximum to Neoglaciation Combining six climate proxies from seven lakes, Geirsdóttir et al. (2019) concluded that Iceland may have been mostly ice-free by 9 ka. Eyjabakkajökull, a northern outlet glacier of Vatnajökull, did not exist during the HTM (Striberger et al. 2012). In the absence of glaciers, sedimentation would have been negligeable on Skeiðarársandur for a long time, perhaps 4–5 ka. However, it is conceivable that Grímsvötn volcano remained ice covered, and if there was ongoing geothermal and/or eruptive activity, there may have been glacial rivers with outburst floods on Skeiðarársandur all through the Holocene (Guðmundsson et al. 2002). In a stable environment, Skeiðarársandur would have been vegetated like other Icelandic lowlands. Several 20th century jökulhlaups left lumps of peat and birch trunks, some of which have been carbon dated. The oldest peat was a little over 8,000 cal BC yrs old (Sveinbjörnsdóttir 2015). Climate cooling is evident after ~5 ka BP, with accelerated trends from ~4 ka onwards and increasing still after ca 3 ka (Blair et al. 2015; Geirsdóttir et al. 2009, 2013, 2019). Eyjabakkajökull in the northeast part of Vatnajökull had reformed by 4.4 BP (Striberger et al. 2012). In the model of Flowers permanent 25x25 plots in a systematic grid W-E across Skeiðarársandur, extending from the 1890 moraines about half way to the coast. The waterlogged lower part of the plain is difficult to access and was not included. Within each plot, the cover of vascular species, mosses, selected lichen groups and cryptogamic crust was visually estimated in 20 randomly located 0.25m2 quadrats, using a modified version of the Braun-Blanquet Cover Abundance Scale with eight cover classes (<1, 1–5, 6–10, 11–15, 16–25, 26–50, 51–75 and 76–100%). Grain size distributions were also recorded; 4) vascular species inventory of the uppermost part of Skeiðarársandur, updated regularly by the authors since 1998; 5) constraints on ecosystem development were studied by Geissler (2005) and Marteinsdóttir et al. (2010, 2013, 2018); 6) the colonization, growth, population and reproductive biology of Betula pubescens subsp. tortuosa, (Mountain Birch) has been monitored since 2004 (Hiedl 2009, Marteinsdóttir et al. 2007; T.E. Thórhallsdóttir, University of Iceland, Reykjavík, Iceland, and K. Svavarsdóttir, Soil Conservation Service of Iceland, Reykjavík, Iceland, unpubl. data). The 2016 distribution of birch was mapped based on approx. 75 km2 area of remote sensing data collected by drones during the summer of 2016 with a resolution of 5.6 cm/ pixel (Madrigal et al. in prep.). The Last Glacial Maximum to Early Holocene At the LGM (ca 18.6–24.4 ka BP; Norðdahl and Ingólfsson 2015, Norðdahl and Pétursson 2005, Pétursson et al. 2015), Iceland was completely covered (possibly excepting small nunataks) by an ice sheet that extended beyond the present coastline to the coastal shelf at 200 m depth (Norðdahl and Ingólfsson 2015). South and southeast of Vatnajökull, moraines have been identified respectively 50 km (Boulton et al. 1988) and ~77 km out from the coast (Thors and Helgadóttir 2014). During the Bölling warming (15.4–13.9 cal. ka BP), the marine-based part of the ice sheet collapsed, and it was reduced to a quarter of its LGM area. The coastal lowlands became ice-free, but were most probably submerged (Geirsdóttir et al. 2009). In southeast Iceland, glaciers reached beyond the present coastline again during the Younger Dryas (13–11.5 cal. ka BP), but over most of Iceland, they did not advance so far (Ingólfsson et al. 2010). On a geological time scale, the build-up of Öræfajökull over the past 800 ka (Stevenson et al. 2006) has had consequences for the entire region. It has remained outside the active zone of tectonic spread. The Icelandic crust responds very quickly to variations in glacier load, and the late glacial history of relative sea level (RSL) changes closely tracks the repeated episodes of glacier advance and retreat (Pétursson et al. 2015). Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 7 evidence from the early 18th century is available for Breiðamerkurjökull, the second largest southern outlet glacier of Vatnajökull, ~35 km east of Skeiðarárjökull and in a comparable climatic setting and lowland altitude. Breiðamerkurjökull advanced 9 km from 1732 to 1890 (Björnsson 1996), but as the advance probably began >400 years earlier, the difference between its settlement and maximum LIA extent is likely to be well over 10 km. The minimum size of Skeiðarárjökull may be deduced from descriptions in Biskupasögur (Egilsdóttir 2002, 2012) and Sturlunga (Thorsson 1988) of a flood in river Lómagnúpsá (= Núpsvötn) in A.D. 1201. It must have come from Grænalón marginal glacial lagoon (Björnsson 2002, see Fig. 2). Since the glacier had extended sufficiently far to form a dam, the snout cannot have been more than 10–11 km behind the LIA moraines in 1200. All considered, it is most likely that Skeiðarárjökull was ≥10 km shorter in the 9th than in the late 19th century. It would then not have reached beyond Færnes mountains (Fig. 2) and cannot have been much more than 7 km across, less than half of its maximum LIA width. and Björnsson, Vatnajökull only began to assume its present shape about 2 ka ago and Skeiðarárjökull may only have spread over the lowland in the last 1,500–2,000 yrs (Björnsson 2017, Fig. 4). A peat block left on Skeiðarársandur by a jökulhlaup in 1948 was identified as 5 ka old mire vegetation (Jónsson 1960). Carbon dating of birch logs exposed by the retreating Skaftafellsjökull (Fig. 2) after 1930 yielded ages of 201 calBC to 209 calAD (93.8% probability, converted from uncalibrated dates of 2020 +/- 80 yrs in Ives 2007). None of these remains were collected in situ, but they demonstrate that icecovered land in the 20th century was vegetated as recently as 2000 yrs ago. Early Settlement Period: A.D. 900–1362 Skeiðarárjökull The probable size of Skeiðarárjökull at the time of settlement may be gleaned from a few sources. In general, the margins of the large outlet glaciers of Vatnajökull appear to have lain 10–20 km inside their LIA maxima (Björnsson 2017). Historical Figure 4. The formation of Vatnajökull according to the numerical model of Flowers and Björnsson. About 3 – 4 ka ago, there were ice caps on Öræfajökull and high inland mountains (a, b). Skeiðarárjökull did not exist 2 ka ago (c) but had reached the lowlands 1 ka ago (d). The ca 2000 AD size of Vatnajökull is shown with red outlines. Reproduced with permission from Björnsson 2017. Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 8 The Main River on the Plain and Records of Jökulhlaups At the time of settlement, the major river on the plain was not called Skeiðará, but Jökulsá (Benediktsson 1968). Several lines of evidence suggest that this river flowed centrally across the plain or even west of center (not on its eastern flank as later). First, Skeiðarársandur has a convex topography with its highest point in the middle and from this Magilligan et al. (2002) deduced that, for most of the Holocene, the major river had flowed centrally across the plain. Second, the position of the boundary between the properties of Skaftafell (east) and Núpsstaður (west of plain) is likely to have been drawn by the largest river, which on this enormous flat terrain, constituted both the major surface feature and the biggest obstacle to travel (Björnsson 2003). While most of the plain had few resources, the coast had precious driftwood and seals, and it was divided into discrete sections that each farm had the right to use. If the coastal boundary between Skaftafell and Núpsstaður farms reflects the course and mouth of the ancient Jökulsá, it did not flow centrally across the plain around A.D. 900, but ~10 km further west. Third, the oldest written sources (Landnáma, Sturlunga, and Biskupasögur) refer to Lómagnúpssandur, i.e., the sandy part was named after the major landscape feature on the NW corner of the plain which may indicate that the barren section was its central to western part. In his treatise on the pre-1362 Öræfi district, Thórarinsson (1958) reasoned that, during the first centuries after settlement, a major glacial river did not flow down the eastern flank of the plain. Jökulhlaups may have been infrequent before the 14th century. The first records of floods related to Grímsvötn that Thórarinsson (1974) found were in 1332 and 1341, the third not until 1598. However, it should be noted that there are few surviving documents from the intervening 150 year period. During this time then, Skeiðará did not exist or was only a minor harmless river. The name Skeiðará first appears in 1540 when farmers in Skaftafell complain of it ruining their land (Björnsson 2003). From that time on, Jökulsá is no longer mentioned and Skeiðará becomes the major and most destructive river on the plain and, with a few short-lived westward excursions, flowed south along the eastern margin of the plain. The fact that the name Skeiðará only appears in the mid 16th century (Björnsson 2003) begs the question of whether Skeiðarárjökull had another name before then. Almost all outlet glaciers in Öræfi have the same prefix as their main river (e.g., Skaftafellsjökull/Skaftafellsá). Did the landscape setting of Skeiðarárjökull not warrant a placename, i.e., was it not perceived as an independent landscape phenomenon, but only as a section of the main ice cap? This might be the case if it had barely advanced onto the lowland. The Settlement Prior to A.D. 1362 The 1179 and 1343 cartularia allow a unique insight into the pre-1362 community, but there are gaps and limitations. Cartularia are not available for Eyrarhorn church, located on the plain north of Ingólfshöfði and likely to own property nearby, an area of particular interest here. It is not always clear in the terse cartularium text whether a name refers to a farm or a landscape phenomenon and the location of resources is usually not specified. Although much of the property and resources of a church were close by, some could be more distant. Rauðilækur church, for example, owned Bakki farm east of Öræfajökull. We compiled the definate and possible pre- 1362 farms between Skeiðarársandur and Breiðamerkursandur (for details, see Supplemental File 1, available online at https://eaglehill. us/JONAonline2/supplemental-files/043- Thorhallsdottir-S1.pdf). Nineteen farms can be placed with certainty or high probability west of Öræfajökull. Gata is a place name on the plain, and we believe it refers to a farm, but Thórarinsson (1958) did not include it. Another four place names were most probably on the plain, but it is uncertain whether they referred to farms. Subtracting one farm in a side valley and five at the foot of Öræfajökull, east of the plain proper, leaves 13 certain and four possible farms on the plain (see Table S1 in Supplemental File 1). Farms owned by Eyrarhorn church are still missing here. There is no telling how many they were, but probably fewer than the 10 farms owned in 1179 by Rauðilækur, the foremost district church. The total number of farms on the plain is therefore unlikely to have been below 16 to 18. Those figures fit well with Einarsson’s (1918) quote from local people around A.D. 1700 that 15, 16, or 18 farms on Skeiðarársandur had been ruined. In 1746, Stefánsson wrote that 15 farms had previously stood on Skeiðarársandur. Rauðilækur and Eyrarhorn both appear to have stood some distance at least out on the plain. Rauðilækur was one of two settlement estates and the district’s foremost farm with the chief church among the five local churches plus three annex churches. The first settlers could pick the best sites, and their properties were usually extensive. The location of Rauðilækur on the plain further strengthens the conclusion that the plain had not just extensive but excellent agricultural resources. Birch Forests and Riparian Meadows on Skeiðarársandur In both the 1179 and 1343 cartularia, two rights to forest use of Rauðilækur church are Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 9 riparian meadows on the eastern plain where glacial rivers meandered among islands and banks. The cartularia provide evidence that, until the mid 14th century at least, the easternmost part of Skeiðarársandur was vegetated with birch forests and riparian meadows. It is very likely that there were also mesic heathlands on the plain, dominated by willows and graminoids (grasses and rushes), similar to the present heathlands on the adjacent Brunasandur outwash plain (Thórhallsdóttir 2015), but since they were not particularly valuable agriculturally, they were not mentioned in the cartularia. Before the LIA, the surface of the outwash plain was lower and the slope of the alluvial cone descending westwards from the slope of Öræfajökull longer and steeper. The birch forests probably extended southwards from Jökulfell, giving way to wetlands in the lower parts of the plain. It is likely that the alluvial slopes also carried birch forest. That the riparian meadows extended all the way up to Skaftafell is supported by the fact that their remains were still harvested in the 19th century (Tómasson 1980). They were probably most extensive in the lower part of the plain, associated with rivers flowing westwards from Öræfajökull and with glacial waters from Skeiðarárjökull. Finally, it should be noted that while sown grain that must be Hordeum vulgare L. (Barley) is listed in the pre-1362 cartularia, native Lymegrass is not. Ancient cartularia have also survived for the church and later convent at Kirkjubæjarklaustur and several neighbouring kristbú (farms left as legacies and donated to Christ for the provision of the poor). In 12th to 14th c cartularia, Lymegrass grain and flour is listed several times (Sigurðsson 1857–1876, Thorkelsson 1893). This was harvested on Brunasandur and Skaftá outwash plains west of Skeiðarársandur (Thórhallsdóttir 2015). The favoured habitat of Lymegrass is dry and unstable sand, and it only sets appreciable seed under such conditions (Greipsson and Davy 1994). Of course, the absence of Lymegrass resources in the Rauðilækur and Hof cartularia is not proof of its absence in the area, but may nevertheless indicate that it was not abundant enough to constitute a valuable resource, i.e., unstable sands with Lymegrass may have been less extensive on Skeiðarársandur than on the much smaller Brunasandur and Skaftá plains further west. The 1362 Eruption in Öræfajökull The Öræfi district, then called Litla-hérað, was devastated by an eruption in Öræfajökull in A.D. 1362, the biggest explosive eruption in Iceland since settlement. It was accompanied by major floods, enormous pumice deposition and pyroclastic flows specified (Sigurðsson 1857–1876, Thorkelsson 1893). One is in the valley by Jökulfell, the other in Sauðabólsskógar (sauðaból = place where sheep overnight, skógar = forests). Further, in 1179, Rauðilækur owned all forests out from Sauðabólsskógar to the forests belonging to Skammstaðir and in 1343, all the forests out from Sauðabólsskógar to Möðruhólar and all the tongues over (“...allar tungur yfir...”) to the forests that belong to Skammstaðir. The forests on the slope of Svínafell mountain belonged to Svínafell farm. The name of Sandfell (sandy fell) indicates that its steep and unstable slopes were as barren earlier as they are now, as probably were the rhyolitic scree slopes of the mountains between Sandfell and Hof. The forests cannot have been on these mountain slopes. As a landscape term, “tongues” typically refers to strips of land between rivers which again points to the forests being on the plain. The phrasing in the cartularium clearly reflects that at the time, these forests were extensive. The 1343 Rauðalækur cartularium also lists extensive uses of engjar. In the broad sense, engjar are unfertilized meadows cut for hay, but most often they were wet, either intermittently overflown riparian meadows or irrigated land, sometimes mires. Irrigation was practised by the earliest settlers and regulations on irrigation and the diversion of riverwater are detailed in the oldest Icelandic book of law, Grágás (used until ca 1270, Karlsson et al. 1992, Thórhallsdóttir 2015). Icelandic riparian meadows and irrigated land are typically dominated by the tall and productive sedge Carex lyngbyei Hornem., a highly palatable and nutritious plant, sometimes by C. nigra (L.) Reichard, and locally in the southeast, by C. diandra Schrank. The sedges were cut for winter fodder, but cattle also grazed these meadows in summer. Some meadows could be cut every summer, others were harvested every second or third year (Thórhallsdóttir 2015). Good haymaking land was always in short supply, and these productive wetlands were extremely valuable resources. From a description of the traditional use of engjar in Öræfi in the ethnological database of the National Museum, it is clear that the term referred to wet meadows (https://www.sarpur.is/ adfang.aspx?AdfangID=552785). The 1343 cartularium says Rauðilækur church owned engjar in Gegnishólar, Lágey (4 km west of Hof), Litla ey, Kerlingarey, Hrosshólmur, Kolluhvalsey, and all of Starkaðarhólmar. With the exception of Gegnishólar (hólar = hillocks), these place names either end in ey (= island) or hólmur (sometimes hólmi, hólmar in plural, = islet). They show that there were extensive Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 10 of normal river flow. For historical jökulhlaups, a frequent interval between floods prior to 1934 may have been 9–12 yrs (Thórarinsson 1974), i.e., ~10/ century. Based on these calculations for the late LIA, the contribution from floods was of about the same order of magnitude as normal river flow. A contemporary account of the 1362 eruption describes a huge flood of water, mud, and rocks that left a sandy plain where previously had been water 30 fathoms deep (Annálsbrot frá Skálholti, Islandske Annaler indtil 1578). If most of the flood came down the west slope of Öræfajökull as Thórarinsson (1958) reasoned, this implies that a part of what is now Skeiðarársandur was then sea. Several documents mention a fjord or bay on Skeiðarársandur (e.g., Skálholtsannáll in 1315–1320). No fjord is shown in the late 16th century maps of Iceland (Fig. 5a), although this may not be significant since this part of the coast may have been particularly poorly known. Around 1702, Magnússon wrote that Ingólfshöfði was previously surrounded by the sea but that the fjord, which was supposed to lie inland from it, is no longer visible. Beginning in 1730, Knoff led a five year survey of Iceland on command of the Danish government. His maps are considered fairly accurate, especially by the coast (Sigurðsson 1978). On Knoff’s maps, there is no fjord on Skeiðarársandur (Fig. 5b). From comparisons with Knoff’s map, Nummedal et al. (1974) concluded that during the intervening 240 yr period (1730–1970), the seaward advance of Skeiðarársandur had been negligible. Around 1980, the German trawler Friedrich Albert was excavated by the mouth of Skeiðará. After 80 years, the trawler lay at a depth of 12–14 m and 120 m upshore (Jónsson 1984). As evident from the above, a comprehensive investigation of the morphological evolution of Skeiðarársandur still remains to be carried out with estimates of the position of the shoreline at the time of settlement and the magnitude of its subsequent seaward expansion. Mýrdalssandur (60 km further west), an outwash plain also subject to recurrent catastrophic jökulhlaups triggered by subglacial eruptions, may provide a partial analogy. Here, there is good evidence that the coast has advanced and drastically changed since the 9th century. Landnáma (Benediktsson 1968) mentions a fjord inland from Hjörleifshöfði promontory which now lies about 2 km upshore from a flat sandy beach. Sigurðardóttir (2014) estimated that, since the 15th century, sediment deposition on SE Mýrdalssandur has raised the sand surface by an average of 2 m/century. For comparisons with Skeiðarársandur, it should be borne in mind that the largest floods from the subglacial Katla volcano are estimated at 300,000 m3 sec-1 (Larsen 2010), an (Sharma et al. 2008). Thórarinsson (1958) concluded that the pumice deposition, at least 30–40 cm over most of the area, was the main agent of destruction. Lögmannsannáll and Flateyjarannáll (Islandske Annaler indtil 1578) state that the district was totally deserted. It is not known when residents returned, but it may have been after a few decades. Some farms were never rebuilt, e.g., those excavated at Gröf (Gestsson 1959) and Bær (Einarsson 2020). The church farms Rauðilækur and Eyrarhorn were re-established, but the second phase of inhabitation on the plain appears to have been short, of the order of a hundred years. With their land steadily eroded, the property of Eyrarhorn church passed to the church at Hof in 1482, and by 1500, Sandfell had replaced Rauðilækur as the main district church (Thorkelsson 1921). In 1605, Egilsson wrote (see Sigurðsson 1856) that Rauðilækur is a deserted farm, but that its ruins are still visible. The Little Ice Age When Litla hérað (Little-shire, with the shire name indicating prosperity) district was resettled after the 1362 catastrophe, its name changed to Öræfi, meaning wasteland. The eruption was an isolated event, but by then the LIA had set in, with advancing glaciers, increasingly destructive glacial rivers, and deteriorating conditions for vegetation (Geirsdóttir et al. 2009, Hannesdóttir et al. 2015). Climatic conditions in the ensuing centuries were quite variable with alternating colder and warmer periods. Some Icelandic glaciers had reached their maximum extent in the 18th century, but the big outlet glaciers from Vatnajökull obtained their maximum Holocene size in the late 19th century (Björnsson and Pálsson 2008). The first two decades of the 20th century were cold in Iceland and are sometimes included in the LIA period. Morphological Changes of Skeiðarársandur The second of the two major phases in the buildup of Skeiðarársandur took place during the Little Ice Age. Maizels (1991) concluded that the plain had mostly been built up by jökulhlaups, while Marren (2002) considered the contribution of normal river processes to be greater. By extrapolating from Maizels´ (1991) calculations for the much smaller Skógasandur outwash plain, Bahr (1997) estimated that floods had contributed 85% of the build-up of Skeiðarársandur, with regular river sedimentation accounting for 15%. H. Björnsson (unpublished calculations) estimated that outburst floods may be of the order of tenfold the annual contribution Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 11 order of magnitude larger than the biggest 20th century jökulhlaups on Skeiðarársandur (of the order of 50,000 m3 sec-1, e.g., Björnsson 2017). This must be weighed against the greater frequency of jökulhlaups from Vatnajökull (at least one per decade in the 19th century) than from Katla, which in past centuries, appear to have occurred at intervals of 20–90 years. The 1727 Eruption in Öræfajökull Although much smaller than in 1362 (Roberts and Guðmundsson 2015, Sharma et al. 2008), the second historical eruption in Öræfajökull also triggered floods down its west slopes, filling the courses of Virkisá and Kotá rivers on either side of Sandfell farm (Hálfdanarson 1729). Contemporary accounts say it destroyed much of the remaining engjar and grazing land of Sandfell (Fig. 2). Otherwise, it is not clear how extensive the damage was. At that time, Skeiðará may already have wiped out much of the agricultural land that would otherwise have been damaged by the flood. The Öræfi Community, Birch Forests and Riparian Meadows in the 18th Century By 1700, the Öræfi community had mostly settled into its 20th century pattern. West of Öræfajökull, four farmsteads remained inhabited: Skaftafell, Svínafell, Sandfell, and Hof (plus tenant farms from at least two of those). Einarsson’s 1709 inventory (Einarsson 1918) shows that the birch forests were virtually gone, except in Skaftafell and with the exception of two tenant holdings, all farms west and south of Öræfajökull, had a right to forest use in Skaftafell. The valuable riparian meadows were largely buried under sand and gravel. About 1702, Magnússon listed the damage done by glacial rivers from Öræfajökull; Svínafellsá had destroyed Svínafell´s engjar, Kotá those from Hof and Falljökulskvísl had taken almost half of Sandfell’s engjar. In a letter to the authorities in 1756, the farmer in Skaftafell laments destruction of his land, complaining that his former engjar are buried under sediment. His son wrote another letter in 1787 describing how Skeiðará continued to destroy land (Tómasson, 1980). In a 1746 account of Skaftafellssýslur counties, Stefánsson described Skeiðarársandur as uninhabited and devoid of vegetation (Stefánsson 1746). He mentions birch forests on the heathland above Skaftafell and beneath Jökulfell, but says nothing of engjar or other agricultural uses on the plain. The Vascular Flora at the End of the 18th Century In the late 18th century, the physician Sveinn Pálsson lived in Vík on the south coast and frequently crossed Skeiðarársandur as his medical district extended over the whole of south and southeast Iceland. Although best known for his pioneering work on glaciology (Björnsson 2017), Pálsson was also a keen botanist. He describes Skeiðarársandur as a barren waste, utterly without vegetation except for Lymegrass dunes by the coast (Pálsson 1791–1794). In his 1793 diary, he writes that there is hardly a single living plant on Skeiðarársandur except for a few Epilobium latifolium L. (now Chamerion latifolium L. Holub.), Silene uniflora Roth, and Arabidopsis petraea L. Later, he added Honckenya peploides L. Ehrh. Some riparian mead- Figure 5. Enlarged sections of the southeast from the map published in 1590 in the Netherlands by A. Ortelius (left) but believed to be based on a now-lost map by Icelandic bishop Guðbrandur Thorláksson from ca 1570 and the Homann’s map (Insule Islandiae delineatio, right) published in Germany in 1761 and based on Thomas H.H. Knoff’s surveying in 1730-34. Pink brackets mark the approximate west and east limits of Skeiðarársandur. See http://islandskort.is/en/ Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 12 ows were left along the tributaries of Skeiðará and fragmented wetlands below the south slopes of Öræfajökull. Everything else, Pálsson says, is buried under endless piles of gravel, sand, and pumice. All the species mentioned by Sveinn Pálsson are hardy. Epilobium latifolium still occurs on Skeiðarársandur, but is infrequent. Honckenya peploides is local and mostly confined to the westernmost part of the sand. Leymus arenarius, Arabidopsis petraea, and Silene uniflora are characteristic of the most hostile part of Skeiðarársandur, the dry and unstable zone above the middle of the plain where plant cover may be 1–5% (Fig. 3d, e). Pálsson would have been most familiar with the uppermost part, because there lay the traditional route across the plain. Over a third of this part of Skeiðarársandur now has a continuous moss cover (Kofler 2004, Fig. 3c). Since 2000, over 90 species of vascular plants have been recorded there, including birch and willows (Martin 2007; T.E. Thórhallsdóttir, University of Iceland, Reykjavík, Iceland, and K. Svavarsdóttir, Soil Conservation Service of Iceland, Reykjavík, Iceland, unpubl. data). Vegetation in the 19th Century In the early 19th century, fragments of riparian meadows still remained between the tributaries of river Skaftafellsá in the uppermost part of the plain (Einarsdóttir 1995). The names of the engjar still harvested there at the time, reflect their soggy nature; Vondibakki (treacherous bank) and Blautafit (wet islet, see Tómasson 1980). The last riparian meadows of Skaftafell disappeared in the 1861 jökulhlaup (Tómasson, 1980). Shortly before, in 1850, the Skaftafell farm buildings on the plain were abandoned, and people retreated 100 m higher up onto the slopes of Skaftafellsheiði (Einarsdóttir 1995). By the 19th century, Svínafell had lost most of its engjar and the remaining meadows of Hof were destroyed in a flood in 1867 (Björnsson 1976). Among the few 19th century foreign visitors to Öræfi was the Dane Kr. Kålund (1877). His statement that Skeiðarársandur itself is completely without grass should probably be interpreted as meaning without vegetation. Jökulhlaups and Their Impacts Thórarinsson (1974) found definate or probable documentary evidence of seven subglacial eruptions and/or jökulhlaups in the 17th century, seven again in the 18th century and 10 in the 19th century. The first contemporary description is of the 1861 jökulhlaup and since then, there are eyewitness accounts for all major floods. Until 1934, they were only observed from the plain or Skaftafell hill. How well eyewitnesses saw the flood varied by time of year and depended on weather conditions and time of day of the maximum flow. The jökulhlaups of 1838 and 1852 were considered very large (Brandsdóttir and Pálsson 2014); the one in 1861, was called Stórahlaup (Big Flood). Large floods came in 1892, 1903, and 1922. The 1903 flood can be cited as an example of the power of these events. It caused sufficient tremors to break glass windows in the farmhouse of Skaftafell, 100 m above the plain, and the noise was heard over 100 km away in Hornafjörður (Thórarinsson 1974). The 1861 flood has been labelled as exceptionally large (Thórarinsson 1974); however, each of the three following floods (1892, 1903, 1922) was described as being the largest or most spectacular in living memory or over the past century. Ragnar Stefánsson, the last in a long line of ancestors as farmer of Skaftafell, observed all floods from 1922 to 1994 and had extensive knowledge of earlier events from his parents and grandparents. He rated the floods of 1934 and 1938 as being of medium size at the most compared to earlier jökulhlaups (Stefánsson 1982). In the 19th century, Skeiðarárjökull lay atop its great LIA moraines. During jökulhlaups, floodwater burst from the snout in numerous outlets, sometimes from large tunnels. Floodwater rushed onto the plain with full force, carrying blocks of ice and depositing huge sediment loads. Eyewitnesses describe icebergs the size of ocean-going ships rushing down at great speed. In 1861, all of Skeiðarársandur, as seen from Skaftafell, was under water, excepting a small triangular patch by the glacier (Stefánsson 1982). The 1892 and 1903 jökulhlaups completely flooded the eastern and western parts, but accounts did not extend to the centre of the plain. Hannesson, who in 1934 became the first to fly over Skeiðarársandur during a jökulhlaup, said that all the plain was under water, except for a few small strips by the glacier, about a fifth of the uppermost part (Thórarinsson 1974). From these accounts, it seems that the large floods of the 19th century and early 20th century extended over the whole of Skeiðarársandur, excepting only patches in the uppermost zone closest to the glacier. Depositing enormous sediment loads, they probably destroyed almost all plant life in their path. With only about a decade between floods, they had left Skeiðarársandur as an exceptionally barren wasteland by the late LIA. The 20th Century Environment The Retreat of Skeiðarárjökull Skeiðarárjökull started a slow retreat by 1890. In 1904, when Danish surveyor Koch and his military team mapped Öræfi, most outlet glaciers south of Vatnajökull were 200–300 m inside their maximum Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 13 vötn volcano from 1938–1996 translating into a much lower frequency of jökulhlaups than in the 19th century (Björnsson 1997). Third, floods have been reduced in size with a thinner glacier. There has only been one big flood since 1938, the one in 1996. This came in November, the ground was frozen and the damage it did to vegetation was limited and local (Svavarsdóttir, Thórhallsdóttir, and Sparrow, pers. observ.). Fourth, a warmer climate has ameliorated conditions for plant growth. Mean annual air temperature at Fagurhólsmýri was 4.6°C in 1961–1990, 4.8°C in 1991–2000, 5.5 in 2001–2010 and 5.6°C in 2011–2020 (Icelandic Meteorological Office 2021). Finally, protection from livestock grazing in Skaftafell National Park (established 1969, now part of Vatnajökull National Park) is likely to have increased seed production and promoted vascular species richness. This may have lead to enhanced seed rain onto the plain from a seed pool with a greater diversity of species adapted to different conditions. Of these facilitating changes, at least two (higher temperatures and less destructive jökulhlaups due to a thinner glacier) are direct or indirect consequences of a warming climate. However, some consequences of the shrinking of Skeiðarárjökull may not be favorable for plant growth. Bahr (1997) estimated that the deep trough Gígjukvísl has dug, may have lowered groundwater levels on the west part of the plain by up to 50 m and the mid 20th century disappearance of Háöldukvísl and Sæluhúsavatn may have led to drier conditions in the center of the plain. The Present Environment In 1973, continuous vegetation was patchy (Fig. 6). Succession towards closed vegetation had clearly set in on the easternmost part of the plain. Since then, a well vegetated belt, up to 2–5 km wide, has established in this part, extending almost 20 km south from Skaftafell (Fig. 6). This is the area that previously carried the riparian meadows. However, as neither Kofler’s (2004) vegetation map nor Bahr’s (1997) study extended to this part of the plain, it is not included in the following discussion. In this section, we give an overview of the present Skeiðarársandur environment between the 20th century Gígjukvísl and Skeiðará rivercourses. We propose vegetation zones based on the east-west oriented hydrological zones defined by Bahr (1997). Vegetation Zonation The uppermost zone: Moss and dwarf shrub heath with birch. Bahr (1997) classified the uppermost part of Skeiðarársandur as an alluvial cone and not a part of the floodplain. This zone starts at LIA terminal moraines (Hannesdóttir et al. 2015). As Skeiðarárjökull retreated, a depression widened between the large LIA moraines and the snout. Floodwater collected in the depression and only had an outlet through the few gaps that the rivers had opened in the moraines. Sediment and ice were partly deposited in the depression and never entered the plain. The destructive power of 20th century floods was thus much reduced compared to the 19th century jökulhlaups. The maximum extent of a major flood was first mapped in 1996 (Snorrason et al. 1997). Below the middle of the sandur, the plain was submerged except for a central strip. In the uppermost part, the flood was confined to major rivers (Núpsvötn, Gígjukvísl, and Skeiðará) and the normally dry courses of Sæluhúsavatn and Háöldukvísl. Flora and Vegetation in the Early 20th Century Botanist Helgi Jónsson (1906) travelled across Skeiðarársandur in 1901. He saw isolated plants of Epilobium latifolium, Arabidopsis petraea, and Silene uniflora—the same species Pálsson mentioned 110 years before. In an old rivercourse near the middle of the plain, Jónsson recorded a further 14 species. After this, Jónsson did not see any plants at all until arriving at Skeiðará at the eastern edge of the plain. In an account of his 1904 expedition and survey of Skeiðarársandur, Koch (1905) described the sand as extremely barren. Close to the glacier, lichens were occasionally encountered and in a few places, scattered patches with meager moss and grass. These patches, he says, may suffice for half a dozen sheep, but are useless for travellers on horseback. Southwards, there was only Lymegrass. Of three documented vegetated patches on Skeiðarársandur around 1900, two were wiped out by floods in 1903 and 1922. On the Danish General Staff 1:50,000 map from 1905 (Danish General Staff 1904/1905), five vegetated patches are shown, all in the uppermost part, and these must be the patches referred to by Koch in 1905. Their combined area on the map is only 0.3 km². Environmental Transformation in the Late 20th Century We suggest that five major environmental changes combined to transform conditions for plant establisment on the upper part of Skeiðarársandur. The depression between the glacier snout and the LIA terminal moraines is now ~4 km wide and 25 m deep in the western part and ~1 km wide and 14 m deep in the eastern part. Water collects in the depression and is canalized into the few gaps eroded by the rivers through the LIA moraines, thereby restricting the flow of water in the uppermost part. Second, there was a quiescent period in the GrímsJournal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 14 1903 flood. In 2005, most of these kettleholes had an almost continuous vegetation cover (Martin 2007), but in 2011, they were covered by ash from the Grímsvötn eruption. In 2017, the tephra still partly covered some holes. The kettleholes on Skeiðarársandur have a more species rich vascular flora than the surrounding plain with a third of the species being kettlehole-exclusives or very rare on the plain (Martin 2007). The easternmost group of kettleholes are younger, probably from the 1934 or 1938 flood (Klimek 1973) and are mostly barren (Martin 2007). South of the alluvial cone, Bahr (1997) divided the outwash plain into four successive zones parallel to the coastline. Zone A: Dry and unstable with very sparse vegetation. Zone A is 1–4 km wide with a depth to groundwater of 2–3 m to 0.85 m in the south (Bahr 1997). The surface is fine sand with a variable cover of gravel. At two sites, 2 and 4 km south of the road in the central part of the plain, 74% of the surface was sand, 20% gravel, and 6% silt (Marteinsdóttir et al. 2010). Mosses and lichens were virtually absent. In 2006, vascular plant cover was 1–2%, which is probably typical for this zone (Fig. 3d). The most prominent vascular species were Arabidopsis petraea, Poa glauca Vahl, Festuca richardsonii Hooker, Rumex acetosella L., Thymus praecox Opiz subsp. arcticus (E. Durand) Jalas and in places Silene uniflora and Cerastium alpinum L. This zone has changed little in the past decades. A comparison of Koch’s 1905 map with Kofler’s 2004 map indicates that moist parts of the plain have contracted downslope. Further, the 2009 disappearance of Skeiðará may have led to drier conditions on the eastern part of the plain (cf Bahr 1997). It is likely that dry, sandy areas are more widespread now than they were in the 19th and early 20th century, but these constitute the most difficult habitats for plants. the base of the LIA moraines (80–90 m a.s.l.) and extends 2–3 km southwards. The national highway passes near the end of this zone (Fig. 2). Grain size is mixed, gravel and larger stones with sand. The surface is stable and very densely packed. Bahr (1997) estimated that in the lower part of this zone, depth to groundwater was 2–3 m. By the early 1970s, succession towards closed vegetation had been initiated in the uppermost parts of the plain (Fig. 7). The alluvial cone now has a largely continuous moss cover across >30 km². Over 90 vascular species have been recorded, a large increase from the five noted by Pálsson (1981) in 1791–1794, during his repeated crossings in the late 18th century and the 17 recorded by Jónsson (1906) in 1901. Mountain Birch colonized this zone late in the 20th century, possibly mostly around or after 1990 (Hiedl 2009, Marteinsdóttir et al. 2007, Fig. 3c). In 2004, mean plant height varied between 5 and 22 cm at four different sites, the tallest individual found was 75 cm and only a handful of trees had reached reproductive maturity (Marteinsdóttir et al. 2007). In 2016, the tallest trees were ~3.5 m in height and birch had expanded across 35 km² (Madrigal et al. in prep.). If no catastrophies intervene, Skeiðarársandur may, in time, foster one of the largest natural birch forests in Iceland. This uppermost zone is dotted by kettleholes, spherical or conical depressions, 1–4 m deep and 8–30 m in diameter (Martin 2007), created by stranded icebergs that became covered with sediments and slowly melted over a period of weeks, months or years. Martin (2007) distinguished six clusters with a total of >3400 kettleholes. Klimek (1973) assumed that the 1861 flood had wiped out all existing kettleholes and concluded that the oldest visible cluster was from the 1892 rather than the Figure 6. Infra-red aerial Landsat ERTS-1 image of Skeiðarársandur taken 30th July 1973 (left) and a Sentinel image taken 6th September 2017 (right, courtesy of the National Land Survey of Iceland). Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 15 with a groundwater depth of 55–5 cm and high moisture levels, up to 40% in the uppermost substrate layer (Bahr 1997). On Kofler´s (2004) map, the surface becomes moist 10–13 km upshore. By 1973, a patch of continuous vegetation was established centrally in this zone, with long but narrow strips of vegetation along freshwater streams (Fig. 6). Now, Zone B: Lymegrass dunes. Zone B is characterized by Leymus arenarius (Fig. 3e). Depth to groundwater was 0.85– 0.15 m (Bahr 1997). In places, Lymegrass is virtually the only plant visible but elsewhere, accompanying species are the same as in zone A. Zone C: Moist central zone, in places well vegetated. Zone C occupies the middle of the plain, Figure 7. Reconstructed vegetation maps of Skeiðarársandur in AD 900-1100 (A), 1890 (B), 2017 (C), and topographic map of the region (D). White = glacier, light brown = mountains, dark green = birch forest and woodland, grey-green = willow heath, lighter green = grassland, rush (Juncus arcticus) heath and moist graminoid and willow heath at various stages of succession, bright green = riparian meadows, blue = water and sea, dark blue lines = rivers, light grey = dry outwash plain, dark grey = moist and waterlogged part of outwash plain. Mountain areas are shown in brown without distinguishing vegetated or barren areas. Farmsteads are shown by triangles. Red triangles are farms that can be placed with a fair to high degree of accuracy. Yellow triangles represent named or putative farms that can be arranged in a N-S order west of Öræfajökull and south of the mountain but there is no knowledge of their precise location. On the map, we chose to place them by the junction between the birch forest and riparian wetland as this appeared to be a likely choice for farm buildings. West of Skeiðarársandur, Núpsstaður, Rauðaberg, Maríubakki, Indriðagarður and Lundur are shown by red triangles. The last two were abandoned at least by the 16th century but their locations are approximately known. The long abandoned Fagriskógur and Skógarhraun are shown by yellow triangles. There is uncertainty whether the long abandoned Djúpárbakki is the same as Maríubakki. For a discussion of the farms west of Skeiðarársandur, see Thórhallsdóttir (2015). For the assumptions, bases and caveats of the early settlement map, see Supplemental File 2. The 1890 map is based on the 1:200,000 map surveyed in 1904 by Generalstabens topografiske Afdeling (published 1905) except that the margins of Skeiðarárjökull were adjusted to the LIA terminal moraines in line with Hannesdóttir et al., 2015). The birch forest at Bæjarstaðarskógur existed in 1900 but was too small to show on the map. For the 2017 map, glaciers and vegetation were based on infra-red SENTINEL 2 image taken on 6 September 2017, courtesy of the National Land Survey of Iceland. A B C D Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 16 this has expanded to a ~25 km², well vegetated, NE–SW oriented tongue which merges southwards with the coastal vegetation. The vegetation mosaic includes moist willow heath, mossy heath, moist Juncus arcticus Willd. subsp. intermedius Hyl. heath and grassland (Kofler 2004). Comparison with the 1973 image indicates that the vegetation has contracted in the part south of main road, and this may reflect drier conditions as discussed earlier (Fig. 6). Zone D: The coastal plain. Here, the surface is largely saturated, with 0–23 cm depth to groundwater (Bahr 1997). Three patches have coalesced to form a ca 12 km long vegetated area upshore from the coast. In 1973, only the westernmost one was clear (Figure 6). Vegetation in the western part is a mosaic of grassland and heathland with patches of wetland, locally dominated by Carex diandra, a sedge mostly confined to coastal lowlands of the southeast (Kristinsson et al. 2018). Current Vegetation Patterns and Ecosystem Development Excepting the fringes of the plain, continuous vegetation has chiefly established in three areas of Skeiðarársandur: 1) the alluvial cone across the uppermost part, 2) a central NE–SW oriented tongue, and 3) upshore from the coast (Fig. 6). Why has succession proceeded more rapidly in those parts? At least three different hypotheses may be proposed as explanations, relating to i) the physical environment, ii) stochastic processes, and iii) historical factors. Greater surface stability in the coarser substrate of the uppermost part is probably critical for moss establishment. Mosses are absent or very sparse in the dry, unstable sandy zone. In turn, mosses may facilitate vascular plant colonization (Geissler 2005). Together, a stable substrate and continuous moss cover may explain faster rates of vegetation development of the uppermost part of Skeiðarársandur. By the coast, high groundwater levels keep the surface moist and sand from blowing in dry weather, thus alleviating the two most limiting factors for plant establishment higher up—drought due to the low water holding capacity of the sand and an unstable surface and moving sand that damages plant tissues. Considering the dominant easterly wind directions, the regional topography and the vegetation of adjacent areas, the seed source for Skeiðarársandur is much more likely to be on the east rather than west side of the plain. Vascular species richness is higher within Vatnajökull National Park than in the farmland zone further south and seed set probably higher in the absence of sheep grazing. The uppermost part of Skeiðarársandur may, therefore, receive higher densities of a greater diversity of seeds. This may partly account for its elevated species richness compared to the lower zones of the plain. The sizable kettlehole- exclusive vascular flora, i.e., species growing in kettleholes but not on the flat plain, strongly suggests frequent dispersal from Vatnajökull National Park and onto the plain, a distance of the order of 10–15 km (Martin 2007). This leaves the question of why vegetation establishment has proceeded quicker in NE-SW oriented tongue near the middle of the sandur than further east or west. This part of the plain is convex (Magilligan et al. 2002), so a more favourable moisture regime can be discounted. Its central location means that it is farthest away from major rivers, which begs the question of whether the pattern reflects flood impacts. Since 1973, the uppermost strips along freshwater streams have disappeared, but vegetation has expanded downslope (Fig. 6). Vegetation succession had been initiated by 1970, but probably not long before. It, therefore, seems unlikely that it reflects the impact of 19th or early 20th century floods, unless the first facilitating steps in the successional process take decades, with only insignificant aboveground changes. In summary, present landscape-scale vegetation patterns on Skeiðarársandur are shaped by a variety of factors. The different rates and directions of ecosystem development partly reflect spatial heterogeneity of the physical environment (grain size distribution, surface stability, groundwater level), but are also greatly influenced by stochastic processes (variation in the density and species composition of the seed rain, Marteinsdóttir et al. 2018). Finally, we cannot at present discount the hypothesis that the historical imprint of the pre-1940 disturbance regime lasted a long time. The Environmental Transformation of Skeiðarársandur and Öræfi Three maps are presented depicting Skeiðarársandur and Öræfi 1) in early settlement times, 2) towards the end of the Little Ice Age, and 3) in 2017 (Fig. 7). Map 1 is a conceptual reconstruction based on the historical data already discussed. For the assumptions, bases and caveats of the 900–1100 AD reconstructed vegetation and settlement map and for glacier sizes, see Supplemental File 2 (available online at https:// eaglehill.us/JONAonline2/supplemental-files/043- Thorhallsdottir-S2.pdf). Map 2 is simplified from the 1:200,000 map surveyed in 1904 by the Danish Army (Generalstabens Topografiske Afdeling), except that the margins of Skeiðarárjökull were adjusted to the LIA terminal moraines in line with Hannesson et al. (2015). For the 2017 map, glaciers and vegetation Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 17 were based on an infra-red SENTINEL 2 image taken on September 6, 2017 (courtesy of the National Land Survey of Iceland). Around 900 AD, Skeiðarárjökull was relatively small, and outlet glaciers from Öræfajökull did not reach the lowlands (Fig. 7). Two major rivers are shown, Lómagnúpsá to the west (now Núpsvötn) and Jökulsá flowing centrally down the plain, but its course may have lain further west. Birch forests and woodland are depicted in about the uppermost third of the plain, which we regard as a conservative estimate. The lower part probably carried a mosaic of willow, grass, and rush heath with patches of wetland, similar to the present vegetation of the adjacent Brunasandur outwash plain (Thórhallsdóttir 2015). Riparian meadows were more or less continuous south from Skaftafell, widening seawards. Wetlands also dominated the plain south of Öræfajökull. The 15 farms that can be positioned with reasonable certainty west and south of Öræfajökull are shown on the map, but the actual number of farms was probably >20 with at least 13–16 on the plain. By the late LIA, Skeiðarárjökull had almost reached Lómagnúpur mountain, and Skaftafellsjökull and Svínafellsjökull were joined on the plain in front of Hafrafell mountain (Fig. 7). Of the ≥20 pre-1362 farms west of Öræfajökull, only four farmsteads remained. Birch forests only survived on Skaftafell’s land with limited woodland on Svínafell mountain. The riparian meadows were mostly buried under sand and gravel. Continuous vegetation was largely confined to the lower mountain slopes and alluvial cones east of the floodplain proper. It was dissected into isolated fragments by glacial rivers and their deposits of sand and gravel. Skeiðarársandur was a barren wasteland with large stretches with no plants at all. Patches with continuous vegetation were minute, in total ~0.003% of Skeiðarársandur’s area. After the mid 20th century, primary succession was initiated in parts of the plain (Fig. 7). West of the Öræfi farmland, continuous vegetation has established on a 20 km long and 2–5 km wide belt, in all ~40 km². The most species-rich part is the uppermost zone of the plain, in all ~40 km² of continuous vegetation. Vegetation has established on a central tounge from the middle of the plain down to the sea (~60 km²). This sums up to about 140 km², 14% of the total area of Skeiðarársandur. In the absence of major disturbances, the largest natural birch forest in Iceland may establish across an area of >35 km² in the uppermost part. The dry and unstable central zones (A and B cf above) have changed less and may remain sparsely vegetated for longer. Summary: The Environmental History of Skeiðarársandur The late-glacial melting of the Icelandic ice sheet contributed to the first major phase in the build-up of the 1000 km² outwash plain of Skeiðarársandur. For much of the Holocene, the plain probably remained wholly or mostly vegetated. Glaciers reformed as neoglaciation set in, but Skeiðarárjökull may only have spread onto the plain in the last 1.5 ka. From the time of settlement (~ A.D. 900) until the mid 14th century, the eastern side of Skeiðarársandur was vegetated with extensive birch forests and productive riparian meadows. There were at least 13–16 farms on the plain west of Öræfajökull. The western fringe of the plain was also forested and sustained farms. There is no evidence that the central part was vegetated and most of Skeiðarársandur has never been inhabited. The 1362 eruption in Öræfajökull temporarily laid the district to waste, but some farms on the plain were rebuilt. However, this second settlement phase barely lasted beyond a century, and the farms on Skeiðarársandur had been deserted by the end of the 15th century. By the mid 16th century, Skeiðará had become the largest river on the plain and a major recipient of floodwater. It ran down the eastern side close to the farms, and in jökulhlaups destroyed everything in its path. By A.D. 1700, the regional birch forests were gone, except in Skaftafell, and farmers mourned the loss of valuable riparian meadows. By the late 18th century, Skeiðarársandur was an exceptionally barren wasteland with isolated plants of a few hardy vascular species in the uppermost part and some Lymegrass dunes further south. Ten jökulhlaups swept across the plain during the 19th century, at least some of them covering more or less the whole plain, depositing icebergs and enormous sediment loads. The intensity, scale, and frequencies of jökulhlaups would destroy any terrestrial ecosystem, and the short intervals between disturbances precluded recovery. In summary, at least three causes of the conversion of the eastern part of the Skeiðarársandur outwash plain from vegetated to barren can be identified. First, the short-term, catastrophic eruption event that temporarily devasted the Öræfi region, but from which the vegetation would probably have recovered in a matter of decades (at least with light to moderate livestock grazing pressure). The primary force of destruction is the Little Ice Age with all its detrimental impacts, first through expanding glaciers and more destructive glacial rivers and second through increasingly catastrophic jökulhlaups as a consequence of a thicker glacier. Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 18 Parts of Skeiðarársandur are now undergoing rapid vegetation succession. We suggest that this transformation can be attributed to 1) the decoupling of the floodwater from the plain, 2) fewer and smaller outburst floods coupled with, 3) a period of low activity in Grímsvötn volcano, 4) ameliorated conditions for plant growth, and 5) increased seed rain of greater floristic diversity with protection from grazing in adjacent Vatnajökull National Park. The most important changes are a direct or indirect consequence of a warmer climate. Skeiðarárjökull’s retreat has decoupled floods from the plain and its thinning has led to smaller floods. Rapid increases in vegetation cover in the past 30 years coincide with higher summer temperatures. In the arctic, regime shifts attributed to higher temperatures are increasingly being reported (Mekonnen et al. 2021). Increased growth of woody species has converted tundra to shrubland and permafrost degradation has both caused the conversion of terrestrial ecosystems to ponds and lakes as well as the reverse process where terrestrial vegetation has replaced drying aquatic ecosystems (Karlsson et al. 2011). Skeiðarársandur adds a sub-arctic example to these arctic studies. We postulate that changes similar to those on Skeiðarársandur may take place by retreating arctic and high alpine glaciers. There, rates of change are likely to vary depending on conditions, including precipitation, the penetrability and grain size of the substrate, and distances to seed sources of different vegetation types. The post-settlement history of the Skeiðarársandur plain is an example of a regime shift that turned forests, wetlands, and heathlands into an exceptionally barren desert. This happenened in successive steps, but each one was probably abrupt and associated with discrete events. In the absence of these disturbances, the original ecosystems would certainly have survived the LIA. While there is little doubt that the cold periods had a detrimental impact on vegetation, this may not have made much difference as the frequency and intensity of the disturbance regime precluded recovery. Parts of Skeiðarársandur appear to be reverting back to ecosystems similar to those of the early-settlement period; birch forests and willow and graminoid heathland. At present, however, there is little sign of the re-establishment of riparian meadows. In the upper part of the plain, it is evident that a major threshold has been crossed with colonization by birch, the only native forest-forming species and the presumed “climax“ vegetation of the lowlands of Iceland. The large-scale establishment of the birch is set to transform at least the uppemost part of the plain. Acknowledgements We gratefully acknowledge support for our research on Skeiðarársandur from the following funding agencies: The Icelandic Research Fund (grants 040263031, 090255021 and 173688-051), Vinir Vatnajökull, Náttúruverndarsjóður Pálma Jónssonar, and Kvískerjasjóður. We thank Helgi Björnsson for discussions and for allowing us to use unpublished data, Kolbeinn Árnason at the National Land Survey of Iceland, and Finnur Pálsson and Joaquin Maria Munoz Cobo Belart at the Institute of Earth Sciences, University of Iceland for assistance in finding and processing aerial images. Ashley Sparrow took part in the initial fieldwork. We thank Árný Erla Sveinbjörnsdóttir and Hreggviður Norðdahl, both at the Earth Science Institute University of Iceland, for assistance with radiocarbon date conversions and glaciation history respectively. We are indebted to Anna María Ragnarsdóttir, the landowner of much of Skeiðarársandur, for her cooperation and interest in the project and to the park managers of southern Vatnajökull National Park. Finally, we wish to acknowledge the many students that have contributed to plant ecological knowledge on Skeiðarársandur, notably Bryndís Marteinsdóttir, Jamie Ann Martin, Jasmin Geissler, Magdalena Milli Hiedl, Ólöf Birna Magnúsdóttir, Oliver Bechberger, Birgitta Steingrímsdóttir, Dagný Rúnarsdóttir, Jón Ásgeir Jónsson, Rannveig Ólafsdóttir, Sigrún Huld Halldórsdóttir, Þorfinnur Hannesson, Hlynur Steinsson, Benedikt Traustason, Guðrún Óskarsdóttir, Hulda Margrét Birkisdóttir, Jóhannes Bjarki Urbancic Tómasson and Vigdís Helmutsdóttir. Literature Cited Bahr, T. 1997. Hydrogeologische Untersuchungen im Skeiðarársandur (Südisland). Münchner Geologische Hefte, Reihe B, 3, XIV. Angewandte Geologie. Ludwig-Maximilians-Universität.142 pp. Barnosky, A.D., E.L. Lindsey, N.A. Villavicencio, E. Bostelman, E.A. Hadly, J. Wanket, and C.R. Marshall. 2016. Variable impact of late-Quaternary megafaunal extinction in causing ecological state shifts in North and South America. PNAS 113:856–861. Benediktsson, J. (Ed.). 1968. Íslenzk fornrit I. Íslendingabók. Landnámabók. Hið íslenzka fornritafélag, Reykjavík, Iceland. 525 pp. Beschta, R.L., and W.J. Ripple. 2012. The role of large predators in maintaining riparian plant communities and river morphology. Geomorphology 157–158:88– 98. Björnsson, H. 1996. Scales and rates of glacial sediment removal: A 20 km long and 300 m deep trench created beneath Breiðamerkurjökull during the Little Ice Age. Annals of Glaciology 22:141–146. Björnsson, H. 1997. Grímsvatnahlaup fyrr og nú. Pp. 61–77, In H. Haraldsson (Ed.). Vatnajökull. Gos og hlaup 1996. Vegagerðin, Reykjavík. 184 pp. Björnsson, H. 2002. Subglacial lakes and jökulhlaups in Iceland. Global and Planetary Change 35:255–271. Björnsson, H. 2017. The Glaciers of Iceland. A Historical, Cultural and Scientific Overview. Atlantis Advances in Quaternary Science. Atlantis Press. 613 pp. Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 19 Björnsson, H., and F. Pálsson. 2008. Icelandic glaciers. Jökull 58:365–386. Björnsson, S. 1976. Byggðasaga Hofshrepps, 3. Pp.11– 148, In Byggðasaga Austur-Skaftafellssýslu. Bókaútgáfa GuðjónsÓ, Reykjavík. 328 pp. Björnsson, S. 2003. Skeiðarársandur og Skeiðará. Náttúrufræðingurinn 71:120–28. Blair, C.L., Á. Geirsdóttir, and G.H. Miller. 2015. A high-resolution multi-proxy lake record of Holocene environmental change in southern Iceland. Journal of Quaternary Science 30:281–292. Boulton, G.S., K. Thors, and J. Jarvis. 1988. Dispersal of glacially derived sediment over part of the continentalshelf of south Iceland and the geometry of the resultant sediment bodies. Marine Geology 83:193–223. Brandsdóttir, B., and F. Pálsson. 2014. Umbrot tengd Bárðarbungu og Grímsvötnum 1838–1903. Jökull 64:91–106. Danish General Staff. 1904/5. Special map of Skeiðarársandur and Öræfi, 1:200 000. The topographic department of the Danish General Staff, Copenhagen. Danish General Staff. 1905. Sheets 77 SA (Lómagnúpur– Núpsstaður), 87 SV (Öræfajökull– Svínafell), 87 SA (Öræfajökull–Hvannadalshnúkur), 88 NA (Ingólfshöfði–Fagurhólsmýri), 88 NV (Ingólfshöfði– Svínafellsfjara) and 78 NV (Kirkjubæjarklaustur– Prestbakki), all 1:50 000. The Topographic Department of the Danish General Staff, Copenhagen, Denmark. deMenocal, P., J. Ortiz, T. Guilderson, J. Adkins, M. Sarnthein, L. Baker, and M. Yarusinsky. 2000. Abrupt onset and termination of the African Humid Period: Rapid climatic responses to gradual insolation forcing. Quaternary Science Review 19:347–361. D’Odorico, P, A. Bhattachan, K.F. Davis, S. Ravi, and C.W. Runyan. 2013. Global desertification: Drivers and feedbacks. Advances in Water Resources 51:326–344. Egilsdóttir, Á. (Ed.). 2002 and 2012. Íslenzk fornrit 16. Biskupa sögur I and II. Hið íslenzka fornritafélag, Reykjavík, Iceland. 380 pp. Einarsdóttir, H.K. 1995. Ragnar í Skaftafelli. Endurminningar og frásagnir. Hörpuútgáfan, Reykjavík, Iceland. 203 pp. Einarsson, B.F. 2020. Bær við Salthöfða: eyðibýli í Öræfum. 7. Fornleifafræðistofan, Reykjavík, Iceland. 320 pp. Einarsson, Í. 1918. Jarðabók Ísleifs sýslumanns Einarssonar um Austur-Skaptafellsþing, er hann gerði 1708 og 1709 í umboði Árna Magnússonar. Blanda I, 1–38. Sögufélag, Reykjavík, Iceland. Einarsson, T. 1966. Suðurströnd Íslands og myndunarsaga hennar. Tímarit Verkfræðingafélags Íslands 51:1–18. Geirsdóttir, Á., G.H. Miller, A. Yarrow, and S. Ólafsdóttir. 2009. Holocene and latest Pleistocene climate and glacier fluctuations in Iceland. Quaternary Science Reviews 28:2107–2118. Geirsdóttir, Á., G.H. Miller, D.J. Larsen, and S. Ólafsdóttir. 2013. Abrupt Holocene climate transitions in the northern North Atlantic region recorded by synchronized lacustrine records in Iceland. Quaternary Science Reviews 70:48–62. Geirsdóttir, Á., G.H. Miller, J.T. Andrews, D.J. Harnig, L.S. Anderson, C. Florian, D.J. Larsen, and Th. Thordarson. 2019. The onset of neoglaciation in Iceland and the 4.2 ka event. Climate of the Past 15:25–40. Geissler, J., 2005. Small-scale vegetation patterns in early successional stage at Skeidararsand, SE-Iceland. Diplom thesis Wetfälische Wilhelms-Universität Münster. Institute für Landschafsökologie, Germany. 73 pp. Gestsson, G. 1959. Gröf í Öræfum. Árbók Hins íslenska fornleifafélags 56:5–87. Greipsson, S., and A.J. Davy. 1994. Characteristics and uses of a dune-building grass. Icelandic Agricultural Sciences 8:41–50. Guðmundsson, M.T., A. Bonnel, and K. Gunnarsson. 2002. Seismic soundings of sediment thickness on Skeiðarársandur, SE-Iceland. Jökull 51:53–64. Hannesdóttir, H., H. Björnsson, F. Pálsson, G. Aðalgeirsdóttir, and S. Guðmundsson. 2015. Variations of southeast Vatnajökull ice cap (Iceland) 1650–1900 and reconstruction of the glacier surface geometry at the Little Ice Age maximum. Geografiska Annaler: Series A, Physical Geography 97:237–264. Hálfdanarson, E. 1729. Frásögn síra Einars Hálfdanarsonar um hlaupið úr Öræfajökli 1727: “Af jöklinum er öræfin hljópu.” Published in Blanda 1: 54–59, 1918–1920. Hiedl, M.M., 2009. Development of early successional populations of mountain birch (Betula pubescens Ehr.) on Skeiðarársandur, Southeast Iceland. Institute of Life and Environmental Sciences, University of Iceland. 11 pp. Holtemeier, F.K., and G. Broll. 2006. Radiocarbon-dated peat and wood remains from the Finnish Subarctic: evidence of treeline and landscape history. The Holocene 16:743–751. Hughes, T.P., S. Carpenter, J. Rockström, M. Scheffer, and B. Walker. 2013. Multiscale regime shifts and planetary boundaries. Trends in Ecology and Evolution 28:389–395. Ingólfsson, Ó., H. Norðdahl, and A. Schomaker. 2010. Deglaciation and Holocene glacial history of Iceland. Developments in Quaternary Sciences 13:51–68. Icelandic Meteorological Office. 2021. Weather summary report on mean parameters for Fagurhólsmýri, SE Iceland. Available online at http://www.vedur.is/ Medaltalstoflur-txt/Stod_745_Fagurholsmyri.Man- Medal.txt and https://vedur.is/vedur/vedurfar/manadayfirlit/. Accessed April 2021. Islandske Annaler indtil 1578. Published in 1888. Det norske historiske Kildeskriftfond. Christiania. Ives, J. 2007. Skaftafell in Iceland: A Thousand Years of Change. Ormstunga, Reykjavík, Iceland. Jepsen, J.U., M. Biuw, R.A. Ims, L. Kapari, T. Schott, O.P.L. Vindstad, and S.B. Hagen. 2013. Ecosystem impacts of a range expanding forest defoliator at the forest-tundra ecotone. Ecosystems 16:561–575. Jónsson, H. 1906. Bæjarstaðarskógur. Sumargjöf 2:13–21. Jónsson, J. 1960. Mórinn á Skeiðarársandi. Náttúrufræðingurinn 30:36–37. Jónsson, J. 1984. Sandur á sjávarströnd. Náttúrufræðingurinn 53:69–72. Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 20 Karlsson, G., K. Sveinsson, and M. Árnason. (Eds.). 1992. Grágás. Lagasafn íslenska þjóðveldisins. Mál og menning, Reykjavík, Iceland. 567 pp. Karlsson, J.M., A. Bring, G.D. Peterson, L.J. Gordon, and G. Destouni. 2011. Opportunities and limitations to detect climate-related regime shifts in inland Arctic ecosystems through eco-hydrological monitoring. Environmental Research Letters 6:014015. Kålund, K.P.E. 1877. Bidrag til en historisk-topografisk Beskrivelse af Island. Gyldendalske Boghandel, Copenhagen, Denmark. 638 pp.] Klimek, K. 1973. Geomorphological and geological analysis of the proglacial area of Skeidarárjökull. Extreme eastern and Extreme western sections. Geographica Polonica 26:89–114. Koch, J.P. 1905. Fra Generalstabens topografiske Afdelings Virksomhed paa Island. Geografisk Tidsskrift 18:1–14. Kofler, K. 2004. Large Scale Vegetation Patterns on a Sandur Plain: A Digital Vegetation Map of Skeiðarársandur dreived from Satellite Imagery. University of Salzburg, Austria. 67 pp. Kristinsson, H., J.B. Hlíðberg, T.E. Thórhallsdóttir. 2018. Flóra Íslands. Blómplöntur og byrkningar (The Flora of Iceland). Vaka-Helgafell Reykjavík, Iceland. 741 pp. Larsen, D.J., G.H. Miller, Á. Geirsdóttir and Th. Thordarson. 2011. A 3000-year varved record of glacier activity and climate change from the proglacial lake Hvítárvatn, Iceland. Quaternary Science Reviews 30:2715–2731. Larsen, G. 2010. Katla–Tephrochronology and eruption history. Pp. 12–40, In A. Shomacker, J. Krüger, and K. Kjær. (Eds.). The Mýrdalsjökull ice cap, Iceland: Glacial processes, sediments and landforms on an active volcano. Development in Quaternary Science 13. Elsevier, Amsterdam. 224 pp. Magilligan, F.J., B. Gomez, L.A.K. Mertes, L.C. Smith, N.D. Smith, D. Finnegan, and J.B. Garvin. 2002. Geomorphic effectiveness, sandur development and the pattern of landscape response during jökulhlaups: Skeiðarársandur, southeastern Iceland. Geomorphology 44:96–113. Magnússon, Á. ca 1702. Chorographica islandica. Pp. 8–120, In Ó. Lárusson (Ed.). Safn til sögu Íslands og íslenskra bókmennta, 2. Published in 1955. Hið íslenska bókmenntafélag, Reykjavík, Iceland. 652 pp. Maizels, J. 1991. The origin and evolution of Holocene sandur deposits in areas of jökulhlaup drainage, Iceland. Pp. 267–302, In J.K. Maizels and C. Caseldine (Eds.). Environmental Change in Iceland. Past and Present. Kluwer Academic Publishers, Dordrecht, The Netherlands. 332 pp. Marren, P. 2002. Fluvial–lacustrine interaction on Skeiðarársandur, Iceland: Implications for sandur evolution. Sedimentary Geology 149:43–58. Marteinsdóttir, B., K. Svavarsdóttir, and T.E. Thórhallsdóttir. 2007. Landnám birkis á Skeiðarársandi. Náttúrufræðingurinn 75:123–129. Marteinsdóttir, B., K. Svavarsdóttir, and T.E. Thórhallsdóttir. 2010. Development of vegetation patterns in early primary succession. Journal of Vegetation Science 21:531–540. Marteinsdóttir, B., K. Svavarsdóttir, and T.E. Thórhallsdóttir. 2018. Multiple mechanisms of early plant community assembly. Ecology 99:91–102. Marteinsdóttir, B., T.E. Thórhallsdóttir, and K. Svavarsdóttir. 2013. An experimental test of the relationship between small scale topography and seedling establishment in primary succession. Plant Ecology 214:1007–1015. Martin, J. 2007. The Ecology of Kettleholes in Successional Environments: Skeiðarársandur, Iceland. M.Sc. Thesis, University of Iceland, Reykjavík, Iceland. 80 pp. Mekonnen, Z.A., W.J. Riley, L.T. Berner, N.J Bouskill, M.S. Torn, G. Iwahana, A.L. Breen, I.H. Myers-Smith, M.G. Criado, Y.L. Liu, E.S. Euskirchen, S.J. Goetz, M.C. Mack and R.F. Grant. 2021. Arctic tundra shrubification: a review of mechanisms and impacts on ecosystem carbon balance. Environmental Research Letters 16:053001 Mitchell, F.J.G. 2011. Exploring vegetation in the fourth dimenstion. Trends in Ecology and Evolution 26:45– 52. Norðdahl, H., and H.G. Pétursson. 2005. Relative sea level changes in Iceland. New aspect of the Weichselian deglaciation of Iceland. Pp. 25–78, In C. Caseldine, A. Russell, J. Harðardóttir, and Ó. Knudsen (Eds.). Iceland–Modern Processes and Past Environments. Elsevier, Amsterdam, The Netherlands. 397 pp. Norðdahl, H., and Ó. Ingólfsson. 2015. Collapse of the Icelandic ice sheet controlled by sea-level rise? Arktos 1:13. Nummedal, D., A.C. Hine, L.G. Ward, M.O. Hayes, J.C. Boothroyd, M.F. Stephen, and D.K. Hubbard. 1974. Recent migrations of the Skeidarársandur shoreline, Southeast Iceland: Final Report, Contract N60921- 730C-258, Naval Ordinance Laboratory, 156 pp. Pálsson, S. 1791–1794. Ferðabók Sveins Pálssonar. Dagbækur og ritgerðir 1791–1794. Published in 1981. Bókaútgáfan Örn og Örlygur, Reykjavík, Iceland. Pétursson, H.G., H. Norðdahl and Ó. Ingólfsson. 2015. Late Weichselian history of relative sea level changes in Ieland during a collapse and subsequent retreat of marine based ice sheet. Cuadernos de Investigación Geográfica 41:261–277. Roberts, M.J., and M.T. Gudmundsson. 2015. Öræfajökull volcano: Geology and historical floods. Pp. 17–44, In E. Pagneux, M.T. Gudmundsson, S. Karlsdóttir, and M.J. Roberts (Eds.). Volcanogenic floods in Iceland: An assessment of hazards and risks at Öræfajökull and on the Markarfljót outwash plain. IMO, IES-UI, NCIPDCPEM, Reykjavík, Iceland. 164 pp. Rocha, J.C., G.D. Peterson and R. Biggs. 2015. Regime Shifts in the Anthropocene: Drivers, Risks and Resilience. PLoS ONE 10:e0134639. Scheffer, M., S. Carpenter, J.A. Foley, C. Folke, and B. Walker. 2001. Catastrophic shifts in ecosystems. Nature 413:591–596. Scheffer, M., and S. Carpenter. 2003. Catastrophic regime shifts in ecosystems: linking theory to observation. Trends in Ecology and Evolution 18:648–656. Journal of the North Atlantic T.E. Thórhallsdóttir and K. Svavarsdóttir Vol. 12, 2022 No. 43 21 Sharma, K., S. Self, S. Blake, T. Thordarson, and G. Larsen. 2008. The AD 1362 Öræfajökull eruption, SE Iceland: Physical volcanology and volatile release. Journal of Volcanology and Geothermal Research 178:719–739. Sigurðardóttir, S.S. 2014. The Eldgjá lava flow beneath Mýrdalssandur, S-Iceland. Mapping with magnetic measurements. M. Sc. Thesis, University of Iceland, Reykjavík, Iceland. 53 pp. Sigurðsson, H. 1978. Kortasaga Íslands frá lokum 16. aldar til 1848. Bókaútgáfa Menningarsjóðs og Þjóðvinafélagsins, Reykjavík, Iceland. Sigurðsson, J. 1856. Biskupa-annálar Jóns Egilssonar. Pp. 15–136, In J. Sigurðsson, K. Gíslason, and G. Brynjúlfsson (Eds.). Safn til sögu Íslands, I. Hið íslenska bókmenntafélag, Copenhagen, Denmark. 184 pp. Sigurðsson, J. (Ed.). 1857–1876. Íslenskt fornbréfasafn, Diplomatarium Islandica, 1. Hið íslenska bókmenntafélag, Copenhagen, Denmark. 857 pp. Snorrason, Á., P. Jónsson, S. Pálsson, S. Árnason, O. Sigurðsson, S. Víkingsson, Á. Sigurðsson, and S. Zóphóníasson. 1997. Hlaupið á Skeiðarársandi haustið 1996. Pp. 79– 137, In H. Haraldsson (Ed.). Vatnajökull: Gos og hlaup 1996. Vegagerðin, Reykjavík, Iceland. 184 pp. Stefánsson, S. 1746. Austur-Skaftafellssýsla. Pp. 1–23, In Sýslulýsingar 1744–1749. Sögurit XXVIII. Sögufélag, Ísafoldarprentsmiðja h.f., Reykjavík, Iceland. 348 pp. Stefánsson, R. 1982. Skeiðarárhlaupin. Skaftfellingur. Þættir úr Austur-Skaftafellssýslu 3:99–118. Stevenson, J.A., D.W. McGarvie, J.L. Smellie, and J.S. Gilbert. 2006. Subglacial and ice-contact volcanism at the Öræfajökull stratovolcano, Iceland. Bulletin of Volcanology 68:737–752. Striberger, J., S. Björck, S. Holmgren, and L. Hamerlik. 2012. The sediments of Lake Lögurinn–A unique proxy record of Holocene glacial meltwater variability in eastern Iceland. Quaternary Science Review 38:76–88. Sveinbjörnsdóttir, Á.E. 2015. 14C aldursgreiningar á jurtaleifum frá Breiðamerkur- og Skeiðarársandi varpa ljósi á umhverfisbreytingar á Nútíma. The Autumn Conference of the Geoscience Society of Iceland, 20th November 2015. Available online at http://www.jfi.is/ radstefnur/. Accessed November 17, 2017. Sveinsson, E.Ó. (Ed.). 1954. Íslenzk fornrit 12. Brennu- Njáls saga. Hið íslenzka fornritafélag, Reykjavík. 514 pp. Thorkelsson, J. (Ed.). 1893. Diplomatarium Islandicum. Volume 2. Hið íslenska bókmenntafélag, Copenhagen, Denmark. 1098 pp. Thorkelsson, J. 1921. Kirkjustaðir í Austur-Skaptafellsþingi. Blanda II 1:246–268. Thors, K. and G. Helgadóttir. 2014. Hryggir í Lónsdjúpi. Náttúrufræðingurinn: 84:43–48. Thorsson, Ö. (Ed.). 1988. Sturlunga saga I & II. Svart á hvítu, Reykjavík, Iceland. 937 pp. Thórarinsson, S. 1958. The Öræfajökull eruption of 1362. Acta Naturalia Islandica 2:1–99. Thórarinsson, S. 1974. Vötnin stríð. Saga Skeiðarárhlaupa og Grímsvatnagosa. Bókaútgáfa Menningarsjóð, Reykjavík, Iceland. Thórhallsdóttir, T.E. 2015. Saga gróðurs og umhverfis á Brunasandi. Pp 71–140, In J. Hjartarson (Ed.). Brunasandur: Mótun lands og samfélags í yngstu sveit á Íslandi. Dynskógar, Rit Sögufélags Vestur-Skaftfellssýslu, Vík, Iceland. 368 pp. Tómasson, Th. 1980. Skaftafell. Þættir úr sögu ættarseturs og atvinnuhátta. Þjóðsaga, Reykjavík, Iceland. 264 pp.