I would like to thank Graham Bowden for drawing the figures. I would like to thank Laurence Totelin for translating the abstract from English to French and Martin Hess for translating the Roth von Telegd quote from German to English. I would also like to thank the anonymous reviewers for helpful comments on an earlier draft of this manuscript.

1 - The mountains of the Mediterranean region

1The Mediterranean Sea is almost encircled by mountains, with the exception of the southeastern area in North Africa (Fig. 1).

2Definitions of what constitutes a Mediterranean mountain area can include areas that drain into the Mediterranean basin. However, this would exclude important areas of southern Europe such as the Picos de Europa in Iberia and the Toubkal Massif in the High Atlas. Furthermore it would also include the central Alps, which are often not considered as Mediterranean, although the southernmost Maritime Alps are. This review follows HUGHES et al., (2006a) and HUGHES and WOODWARD, (2009) and includes a review of glaciers in the countries that border the Mediterranean Sea. This flexible approach also allows for inclusion of countries such as Bulgaria, which has similarities with nearby Balkan countries.

3In the north, the watershed of the Mediterranean basin culminates in the Alps, containing some of the highest mountains in Europe (including Mont Blanc, 4818 m a.s.l.). Further west, the Pyrenees dominate followed by the numerous mountain chains of the Iberian peninsula, where in both areas several mountains exceed 3000 m a.s.l. To the south the Atlas Mountains dominate northwest Africa, reaching 4167 m a.s.l. at Jbel Toubkal, gradually giving way to the east to the flattest and low-lying areas of the Mediterranean basin in Libya and Egypt. In the eastern Mediterranean, the watersheds are marked by the Lebanese Mountains culminating at Qornet es Saouda (3083 m a.s.l.). Continuing north into Turkey, the Taurus Mountains dominate along with several other chains such as the Kaçkar Mountains and also numerous isolated volcanoes such as Mount Ararat, which reaches 5137 m a.s.l. Westwards, into Europe, the southeastern Balkans are dominated by the Pirin, Rhodope and Balkan Mountains whilst in the west the great mountain ranges of the Pindus Mountains and Dinaric Alps run from the Peloponnese to Slovenia, reaching over 2500 m a.s.l. in several places, and eventually merging into the European Alps. In Italy, the Apennines dominate the main peninsula and reach 2912 m a.s.l. at C10 Corno Grande in the Gran Sasso. Finally, the Mediterranean islands also contain high mountains, especially in Corsica, Sicily and Crete where the highest peaks reach well over 2000 m a.s.l. and over 3000 m a.s.l. at Etna on Sicily.

4During the Pleistocene the Mediterranean mountains were extensively glaciated. Whilst large ice masses covering whole mountain chains were rare and restricted to the Alps and Pyrenees, substantial ice caps and valley glaciers did form in many of the Mediterranean mountains. Some of the largest Mediterranean large ice fields formed over the mountains of Iberia and the Dinaric Alps in Montenegro and Albania. Even as far south as North Africa, large valley glaciers radiated out from ice caps centred in the Toubkal and Djurdjura Massifs of Morocco and Algeria, respectively. The timing of the Pleistocene glaciations varied across the region. Some of the clearest glacial landforms date from the last glacial cycle, although in several areas older Middle Pleistocene glacial landforms are well-preserved (HUGHES and WOODWARD, 2008).

5Today, few glaciers survive in the Mediterranean mountains. In the mountain ranges described above most of the larger glaciers are in the Alps. The Pyrenees are also home to modern glaciers with 21 glaciers with ice patches also surviving in the Picos de Europa, in northern Spain, and permafrost present in the Sierra Nevada of southern Spain (GONZÁLEZ-TRUEBA et al., 2008). The difference between glaciers and ice patches is that glaciers display evidence of movement and internal deformation whereas ice patches are static. SERRANO et al. (2011) defined two types of ice patch – one originating from previously dynamic glacier ice and another originating from a snow patch. In some areas, such as the Maritime Alps (France/Italy) and in the mountains of Turkey, active rock glaciers were present in the Little Ice Age and many of these remain active today. The role of debris cover on glacier development is considered in the later discussion.

6No glaciers are present today in North Africa or in the Middle East. However, several glaciers do survive in Turkey, some of which are the largest of the Mediterranean region. Elsewhere, in Bulgaria, Albania, Montenegro, Slovenia and Italy tiny glaciers are still present. Whilst glacier extent is very modest in these areas today, glaciers were much more numerous in the Little Ice Age. Knowledge of Little Ice Age glacier extent is variable around the Mediterranean. The best records are those from the Pyrenees where there has been a longstanding tradition of glacier observation. In areas such as the Balkans, the presence of Little Ice Age glaciers has only recently become apparent. Appreciating that the state of knowledge is variable, this paper reviews the evidence for recent glaciation in the Mediterranean mountains and examines glacier changes from the Little Ice Age to the present.

2 - Defining the Little Ice Age from a glacier perspective

7Glaciers throughout Europe were larger than today during the “Little Ice Age”. This was a period characterised by mean annual air temperatures c. 1ºC lower than today. Traditionally, the Little Ice Age is defined as the period between 1550 and 1850 AD (JONES and BRADLEY, 1992). In Europe, the most prolonged cold period of the Little Ice Age was between 1550 and 1750 AD although the period from the 1830s to 1860s saw some of the coldest summer temperatures (BRADLEY and JONES, 1993). However, authors such as NESJE and DAHL (2003) and MATTHEWS and BRIFFA (2005) noted that the term ‘Little Ice Age’ can be defined by glaciological as well as climatic criteria. MATTHEWS and BRIFFA (2005) argue that the glacier advances which characterised the Little Ice Age were forced by increased precipitation in addition to lower temperatures. Consequently, they recognised a Little Ice Age in the European Alps spanning the period 1300‑1950 AD. In fact, CLAGUE et al., (2009) argued that the term ‘‘Little Ice Age’’ should be restricted to glacier activity and not used in a climatic context. Whichever time span is used to define the Little Ice Age, this paper takes a broader view and examines the changing glaciers of the Mediterranean mountains over the past few centuries of the Little Ice Age right up to the present-day.

3 - Modern glaciers in the Mediterranean mountains: Little Ice Age to present

8There less than 100 glaciers in the Mediterranean mountains, excluding the main Alpine chain. Most of these have retreated in the past 150 years from more extensive 19^th Century positions. In many cases it is known that ice masses have completely disappeared since this time and the transition from the Little Ice Age to the current period has been marked by one of the great phases of glacier retreat. Despite overall retreat there are known instances of glacier stabilisation or even advance, in particular during the 1960s and 1970s. Whilst the recent history of Mediterranean glaciers (last 150 years) is well documented (GRUNEWALD and SCHEITHAUER, 2010) the history of glaciers during the Little Ice Age itself it not so clear. Whilst there are good records from some areas like the Pyrenees (e.g. GROVE, 2004), in many areas of the Mediterranean mountains we simply have no record of how glaciers advanced and retreated prior to 1850. Nevertheless, it does seem that in some parts, the glaciers of the Late 19^th Century were some of the biggest of, not just the Little Ice Age, but the entire Holocene.

3.1 - Iberia

9In Iberia, outside of the Pyrenees (which is covered in another section below), there are no glaciers. However, at least four ice patches exist today in the Picos de Europa in the Cantabrian Mountains (Fig. 1) (SERRANO et al., 2011). The largest of these is the Jou Negro ice patch which is situated at c. 2220 m a.s.l. in a north-facing cirque below the highest peak of the Picos de Europa, Torre Cerredo (2648 m a.s.l.). There has been some debate as to whether this ice mass represents a true glacier or simply a relict ice patch from the Little Ice Age (GONZÁLEZ SUÁREZ and ALONSO, 1994, 1996; FROCHOSO and CASTAÑÓN, 1995). The consensus is that the feature now represents an ice patch (GONZÁLEZ-TRUEBA et al., 2008; SERRANO et al., 2011). However, the Jou Negro and other ice patches in the area were recently true glaciers which have lost mass throughout the 20^th century, especially after the 1980s, with considerable losses during recent years, such as in 2009 (SERRANO et al., 2011).

10During the Little Ice Age, the Picos de Europa was the only glaciated massif of the Cantabrian Mountains (GONZÁLEZ‑TRUEBA, 2006). At the end of the Little Ice Age (1890-1900 AD) the Jou Negro glacier had a surface area of 5.2 ha. By 2008 AD the feature had stopped moving and shrunk to cover an area of just 0.6 ha. It is clear therefore that the Jou Negro ice patch is a dead glacier formed by changes in the glacial morphodynamic system.

11Further south, in the Sierra de Gredos, there is evidence of substantial increase in snow cover in the Sierra de Gredos during the Little Ice Age (e.g. TORO et al., 1993). On the northern slopes of Pico Almanzor (2592 m a.s.l.) the highest in the Gredos, SANCHO et al. (2001) used lichenometry to show that a protalus rampart, situated at the base of a snow patch, formed during the Little Ice Age.

12In the Sierra Nevada (Fig. 1), buried ice is present today at the foot of the north wall of Picacho del Veleta (3398 m a.s.l.) in the Corral Veleta cirque. Drilling investigations have shown that ice still exists at altitudes of 3050-3100 m a.s.l. buried within the talus in this cirque (GÓMEZ et al., 2001, 2003). The buried ice in the Corral del Veleta represents the southernmost permafrost remnant in Europe (GÓMEZ et al., 2001) and a small rock glacier has developed (PALADE et al., 2011). The topoclimatic position of this sporadic permafrost, in a shaded north‑facing cirque and covered by rock debris, means that it is now partly decoupled from climate. In these circumstances the ice bodies can survive apparently unfavourable climatic conditions. This is evidenced by the presence of active rock glaciers, which require some interstitial ice within the debris mass, long after the end of the Pleistocene (as late as 7,000 years ago) in the Sierra Nevada (GÓMEZ-ORTIZ et al., 2012).

13The Little Ice Age in the Sierra Nevada has been reviewed by GÓMEZ‑ORTIZ et al,. (2009). Glacier ice in the Corral Veleta cirque was first reported scientifically in the early 19^th century by BOISSIER (1837) and later by MADOZ (1849). A quote from MADOZ (1849) was reported in GÓMEZ‑ORTIZ et al., (2009) and is repeated below:

“The regions of these two lofty mountains, Mulahacen and Veleta, and their immediate surroundings are covered with eternal and hardened snows, with layers or stratifications which can be counted easily …. The snow, blown by the wind, which can be found collected in the great repository of this cirque or corrie (referring to el Veleta) is petrified to such an extent that it has the consistency of the hardest marble” (MADOZ, 1849 cited in GÓMEZ‑ORTIZ et al., (2009, p. 285)

14There are no records of when and for how long this ice mass was a true glacier, in the sense that it was an actively deforming ice mass responding to changes in mass balance. However, in 1916, Obermaier noted that the feature was a mass of dead ice with no movements. Whether the Corral Veleta ice patch became reactivated forming a glacier at a later date is not clear. GARCÍA-SAINZ (1947) considered the feature to be a glacier comparable to those in the Pyrenees. MESSERLI (1967, p.162) noted that the Corral Veleta glacier was the “sensation of the Sierra Nevada” but that by the 1960s it was “merely a firn dot”.

3.2 - Pyrenees (France, Spain & Andorra)

15Today, there are 21 glaciers and c. 29 ice patches in the Pyrenees (GONZÁLEZ-TRUEBA et al., 2008) (Fig. 1). This contrasts with at least 115 glaciers which were present in the Little Ice Age. A good overview of these glaciers with photographs and illustrations illustrating their changing extents in recent times is provided by RENÉ (2013). From 1880 to 1980 at least 94 glaciers became extinct, while 17 glaciers have disappeared on the Spanish side since the 1980s (GONZÁLEZ-TRUEBA et al., 2008). In the Cirque de Troumouse, on the northwestern flank of La Munia (3134 m a.s.l.) Little Ice Age glaciers are recorded by moraines at c. 2360 and at c. 2650 m a.s.l., close to the cirque headwall (GELLATLY et al., 1992). Similar moraines are also present on the northeastern and eastern flank of La Munia in the Cirque de Barroude and Cirque de Barrosa, respectively. The Little Ice Age moraines in the Cirque de Troumouse on La Munia, are not far from the existing glaciers of Glacier de la Munia and the Pene Banque ice patch (Fig. 2).

16In the Cirque de Barrosa, snow was still present in October 2013 in hollows behind moraines (which are thus presumed to be Little Ice Age in age) at altitudes of c. 2400-2600 m a.s.l. (Fig. 3).

17The evidence from La Munia illustrates that Little Ice Age glaciers formed in NW, NE and E facing cirques. Across the Pyrenees, moraines associated with the most extensive Little Ice Age glacier expansion occur at similar altitudes to those on La Munia, with lowest moraine altitudes ranging from 2200 to 2630 m.

18On La Munia, it is clear that the Little Ice Age glaciers were not much bigger than the ice and snow that exists there today. Downvalley of the Little Ice Age moraines in the Cirque de Troumouse radiocarbon ages have been obtained from two cores in a bog (2099 m a.s.l.) near Lac d’Aires (Fig. 2). A radiocarbon age of c. 5190 ¹⁴C years BP indicates that the moraines down-valley of this site must be at least mid-Holocene in age (GELLATLY et al., 1992). Neoglacial events are indicated by younger silt-rich sediments in this bog sequence dated between 4955 ± 90 and 4654 ± 60 ¹⁴C years BP and associated with two younger moraine sets on the cirque floor at altitudes of c. 2250-2350 m a.s.l. This means that the Little Ice Age moraines (which are higher at altitudes >2350 m a.s.l., Fig. 2) were not the largest of the Holocene. However, the geochronological sequence of Holocene moraines is not yet clearly established in the Pyrenees with GELLATLY et al.’s 1992 paper one of few that presentings geochronological data .

19The largest Little Ice Age advance period was during the late 1600s and the early 1700s AD. This maximum phase of Little Ice Age expansion is document in the chronicles of Ramond de Carbonnières in 1797 and 1802 (RAMOND de CARBONNIÈRES, 2010). In addition to these historical observations lichenometry and dendrochronology has also placed the maximum advance in the Pyrenees to the 1600s and 1700s (JULIÁN and CHUECA, 1998; SAZ and CREUS, 2001; both cited in GONZÁLEZ-TRUEBA et al., 2008).

20Glacier retreat between 1750 and 1800 was followed by a later readvance between 1800 and 1850 (GONZÁLEZ‑TRUEBA et al., 2008). This was followed by period of major glacier retreat starting c. 1870 AD. At Glacier du Pays Baché on the eastern side of the Néouville massif in the French Pyrenees, Michelier estimated that the glacier lost c. 8,400,000 m³ in volume as a result of reduced precipitation and warmer winters (MICHELIER, 1887). Despite this retreat, the late 19^th century was characterised by another advance and BONAPARTE (1890) presented photographic evidence of increased ice thickness between 1880 and 1890. GROVE (2004, p. 199) also cites evidence for increased accumulation for the period 1885-1895 at Glacier d’Ossoue on the eastern slopes of Vignemale (3298 m a.s.l.), covering caves above the glacier (grottes) once inhabited by Count Henry Russell, forcing their abandonment with new caves dug below the glacier. Fig. 4 shows the positions of the glacier front in 1789, 1865 and 1927.

21Since the mid 19^th century the Glacier d’Ossoue has undergone dramatic retreat, moving 1.5 km up-valley, rising 500 m in altitude and reducing in size from 110 hectares in 1850 to 58 hectares in 2002 (GROVE and GELLATLY, 1995; GROVE, 2004; ECLIPSE, 2003). Whilst recession dominated the since the mid 19^th it was not continuous. Periods of glacier advance or stabilisation occurred in the years 1889-1899, 1905-1927, 1945-1946 and 1962-1986. The period between 1962-1986 was a major phase of positive mass balance with the glacier front advancing by a distance of over 150 m in 20 years (ECLIPSE 2003, p. 15). The nearby Glacier des Oulettes on the north side of Vignemale has a similar history to the Glacier d’Ossoue with the glacier reducing in size from 38 hectares in 1850 to 16 hectares in 2002.

22The largest modern glaciers in the Pyrenees are found in the Aneto-Maladeta Massif where 5 glaciers cover and area of 152 hectares (Table 1).

23The glacial geomorphology of this massif has been studied in detail and good overviews of recent (including Little Ice Age) glacier extent are provided by COPOONS and BORDONNAU (1994) and LAMPRE VITALLER (1994). The largest Maladeta glacier shrank in surface area by 35.7% from 152.3 hectares in 1820-1830 to 54.5 hectares in 2000 and the ELA rose by 255 m over the same period (CHUECA CÍA et al., 2005). However there were at least three phases of significant stability during the intervals 1820‑1830 to 1857; 1914‑1920 to 1934‑1935; and 1957 to 1981.

24SERRANO et al., (2002) identified six phases of Little Ice Age glacier behaviour with the maximum extension of glaciers reached between 1600 and 1750 (see review in HUGHES and WOODWARD, 2009). A minor advance occurred between 1905 and 1920, whilst the interval 1920‑1980 was characterised by glacier retreat and a particularly rapid wastage between 1980 and 2000 (SERRANO et al., 2002). It is estimated that the Posets Massif has lost -39.2% of ice surface area since the Little Ice Age. The comparable figures for the rest of the Spanish Pyrenees are -32.0% for the whole Aneto-Maladeta massif, -24.9%, for Monte Perdido, and -33.6% for Infiernos (CHUECA CÍA et al., 2005)

3.3 - Maritime Alps (France & Italy)

25Fifteen small glaciers were present in the Maritime Alps on the border of France and Italy (Fig. 1) at the end of the 20^th century (FEDERICI and PAPPALARDO, 1995). These are the southernmost glaciers in the European Alps and some are situated only 50 km from the Mediterranean Sea at Monaco. Thirteen of the glaciers are found in the Argentera Massif, which contains the highest peaks of the Maritime Alps, including Argentera itself (3297 m a.s.l.) and a further seven peaks above 3000 m a.s.l. The equilibrium line altitude (ELA) of the six largest Argentera glaciers was c. 2800 m a.s.l. at the start if this century (FISINGER and RIBOLINI, 2001). Glaciers have retreated in the last few decades and this has been the result of temperature rises, a decrease in winter snowfall and a shift of in dominant precipitation from autumn to spring (FEDERICI and PAPPALARDO, 2010).

26Rock glaciers are active in the Argentera Massif and the lower limit of permafrost and associated rock glacier activity is situated between 2500 m and 2600 m a.s.l. (RIBOLINI and FABRE, 2006). During the Little Ice Age, glaciers were much more substantial with ELAs 100-150 m lower than in the late 20^th century (FEDERICI and PAPPALARDO, 1995; Pappalardo, 1999). Holocene cold phases such as that of the Little Ice Age and Sub Boreal are thought to be responsible for the permafrost found in rock glaciers today (RIBOLINI and FABRE, 2006). This important because it means that when ice becomes buried by debris it can become decoupled from climate and the effects of colder periods such as the Little Ice Age can survive into warmer phases centuries later.

3.4 - Apennines (Italy)

27Today one of the southernmost glaciers in Europe, the Calderone glacier, survives in a cirque on the north face of the highest peak of the Apennines, Corno Grande (2912 m a.s.l.), in the Gran Sasso massif (Fig. 1). GIRAUDI et al. (2011) reviewed the evidence for the Holocene history of the Calderone glacier. The Calderone glacier disappeared in the first half of the Holocene. This is known because soils were able to develop on cirque floors. The Calderone Glacier re-formed just after 4520–4290 cal. years BP. This is based on the age of a reworked soil found within moraines (GIRAUDI, 2004, 2005). The glaciers subsequently experienced several phases of expansion at c. 2855‑2725 cal. years BP, c. 1410‑1290 cal. years BP, 640‑580 cal. years BP followed by the largest advance, which dates from the Little Ice Age. The glacier retreated dramatically through the 20^th century. From 1916 to 1990 the volume of the glacier shrank by c. 90% and the area by c. 68% (GELLATLY et al., 1994; D’OREFICE et al., 2000). The reasons for this are clear: over the period 1932‑2005 mean summer temperatures rose by c. 1°C whilst winter precipitation declined by c. 100 mm over a similar period (1919-2003). As a consequence, average annual mass balance for the period 1920 to 2006 was -0.74 mm w.e. In 2000 the glacier split into two portions (PECCI et al., 2008). The current state of the glacier is unknown to the author, although there are reports that the g lacier has become overwhelmed with debris. As with the Corral Veleta site in the Sierra Nevada (Spain) the Calderone glacier is likely to become a collection of buried ice patches and possibly evolve into a small rock glacier.

3.5 - Julian Alps (Slovenia)

28Two ice patches are present today in the Julian Alps of Slovenia (Fig. 1) along with numerous permanent and semi-permanent snow patches. The last remaining ice patches survive below the peaks of Triglav (2864 m a.s.l.) and Skuta (2532 m a.s.l.). The Triglav glacier (sometimes known as Zeleni Sneg) is situated between c. 2400 and 2500 m a.s.l. on the northern cliffs of Slovenia’s highest peak. The Skuta glacier is lower, situated between 2020 and 2120 m a.s.l.

29The Triglav glacier has undergone dramatic retreat throughout the 20^th century (Fig. 5) (Table 2).

30Between 1937 and 2005 the glaciers had reduced in area from 27 to 0.7 ha and reduced in volume from 8000 to just 20 m³ (GABROVEC, 2008). However, there was not continuous annual retreat, since the glacier expanded slightly after heavy winter snowfalls in 2001 and a similar situation occurred in the 1970s (ŠIFRER, 1987; GABROVEC and PERŠOLJA, 2004). Despite these brief expansions, the glacier reached its minimum of 0.6 ha in 2003. Since then there has been little change in the extent of the glacier. The Skuta glacier is situated in a shaded cirque and is largely covered in debris. As with the Triglav glacier the Skuta glacier has also suffered similar rapid retreat (PAVŠEK, 2004).

31The Triglav glacier is thought to have reached its maximum Little Ice Age positions at a similar time to the glaciers of the eastern Alps. GABROVEC (2008) provides an interesting history of observations at this site. GABROVEC (2008) notes the work of ŠIFRER (1963) who suggested that the oldest moraines were formed by a large ice advance in the 17^th or 18^th centuries. This was followed by a smaller advance in the mid 19^th century, and a third moraine accumulation dating from c, 1920 AD when, due to heavy winter snowfall, the retreat of the glacier temporarily stopped (ŠIFRER, 1963). Photos from the 1920s reveal deep transverse crevasses on the Triglav glacier (see Fig. 1 in TRIGLAV-ČEKADA et al., 2012). Systematic measurements were begun on these crevasses in 1946 and a crevasse depth of 8.6 m was measured in 1950 and did not reach to the bottom of the glacier. The Skuta Glacier also had crevasses and in 1913 they were between 10 and 15 m deep. On the uppermost section of the Skuta Glacier transverse crevasses were still observed in 1973 (TRIGLAV‑ČEKADA et al., 2012).

3.6 - Dinaric and Albanian Alps (Montenegro and Bosnia)

32Small glaciers still survive today in Montenegro and Albania. In the Durmitor Massif, Montenegro (Fig. 1), the Debeli Namet glacier covers several hectares and has an ELA of c. 2100 m a.s.l. (DJUROVIĆ, 2013). This glacier has been present in its cirque basin throughout the 20^th century (DJUROVIĆ, 2013), although has been close to extinction on several occasions in recent years (HUGHES, 2008; GACHEV and STOYANOV, 2012). In the 19^th century the glaciers was only slightly larger than it is today and moraine crests have been dated using lichenometry to c. 1878 AD and 1904 AD (HUGHES, 2007) (Fig. 6).

33Elsewhere in the Durmitor Massif, there are no surviving glaciers today, yet moraines are present in many cirques at similar altitudes to those in front of the Debeli Namet glacier (Fig. 7).

34Again, these moraine surfaces have been dated using lichenometry and reveal similar histories with evidence of glaciers in these cirques in the late 19^th century at the end of the Little Ice Age (HUGHES, 2010).

35DJUROVIĆ (2013) noted that the Debeli Namet glacier has reduced in size at a rate slower than that of other southern European glaciers and has maintained a similar size and position over the past 50 years – despite significant short term contractions in 1998, 2003 and 2007 (HUGHES, 2008; GACHEV and STOYANOV, 2012). Its resilience means that it is now one of the largest glaciers of southern Europe (DJUROVIĆ, 2013). The fact that the surviving Debeli Namet is not significantly smaller than it was in the 19^th century, yet in other cirques the glaciers have vanished, illustrates that once a glacier does melt completely it is unable to then re-grow. This is because it takes several years, possibly decades, of continuous positive mass balance to reach snow depths sufficient to compress basal accumulations into ice, although melting and re-freezing within the snowpack can accelerate this process. The fact that the Debeli Namet has not completely melted appears to be due to very strong local topoclimatic controls, especially avalanching from the surrounding cliffs and windblown snow accumulation off the high plateau above the glacier. Whilst other cirque sites are prone to avalanching they lack similar plateau surfaces above them, hence their ice masses have disappeared.

36There are no other glaciers in Montenegro today. However, there is evidence of small glacier occupation in many cirques and HUGHES et al., (2011) demonstrated that these post-date larger Younger Dryas glaciers. In the Sinjajevina Massif, moraines ascribed to the Little Ice Age are present in the northern cirque of Gradišta at 42.8839°N, 19.3699°E, c. 1925 m a.s.l. (Fig. 8).

37Moraines are also present in the northern cirque of Torna (2217 m a.s.l.), the highest peak of this massif at an altitude of c. 1850-1900 m a.s.l. (42.8679°N, 19.3762°E). Lichenometry suggests these moraines formed over 100 years ago with surface ages corresponding to the final phase of moraine building at the end of the 19^th century (HUGHES et al., 2011). Similar moraine ridges and pronival ramparts are present in the highest cirques of other mountain areas, including several cirques of Kapa Morača (2226 m a.s.l.) and Komovi (2487 m a.s.l.).

38Today the greatest concentration of permanent snowfields in Montenegro occurs in the Prokletije Mountains (Fig. 1) and here many cirques contain moraines and pronival ramparts that formed during the Little Ice Age. The moraines and ramparts often enclose areas of permanent and semi-permanent snow. On the Albanian side of the Prokletije several small glaciers still survive (Fig. 9).

39MILIVOJEVIĆ et al. (2008) and HUGHES (2009) identified glaciers in the area of the highest summit Maja Jezerce (2694 m a.s.l.). Three glaciers were reported by MILIVOJEVIĆ et al. (2008) whilst a fourth, on the remote eastern face of Maja Jezerce, was reported in HUGHES (2009). Some of these glaciers had disintegrated by the early autumn of 2011 when GACHEV and STOYANOV (2012) made more recent observations. However, they found evidence for several more small ice/snow masses with frontal moraines such as in a cirque of Maja e Made (2576 m a.s.l.) and the total number of existing ice masses in Albania is still very much undiscovered. The Little Ice Age extent of glaciers in this area is even more of a mystery. HUGHES (2007) suspected the presence of glaciers because of the presence of a glacier in the Durmitor massif in Montenegro and also because of reports of permanent firn fields in the Prokletije by ROTH VON TELEGD (1923). On the western and north-western slopes of Maja Jezerce (2692 m a.s.l.), ROTH VON TELEGD reported ‘Firnmasse’ longer than 1 km:

‘… füllte an der Maja Jezerce, an der NW-seite ihres 2692 m hohen Nebengipfels eine breite Firnmasse, die länger als 1 km war, den oberen Teil des gegen die Jezerceer Seen gerichteten Tales aus und einen nicht viel kleineren Firnfleck fand ich auch am Westfuße dieser Spitsen, in dem gegen die Čafa Pajs gerichteten Becken.’
(ROTH VON TELEGD, 1923, p. 471)

English translation: ‘at the Maja Jezerce, on the NW flank of its 2692 m high secondary peak a wide firn longer than 1 km, filled the upper part of the valley facing the Jezerce Lakes, and a not much smaller patch of firn I also found at the western foot of these peaks, in the Čafa Pajs-facing basin’.

40The largest area of firn occupied the cirque floor to the east of the modern Maja e Kolacit glacier and covered the area of ice and rock described by MILIVOJEVIĆ et al., (2008) as a rock glacier. ROTH VON TELEGD (1923) also mapped a large perennial snow field in the wide cirque floor to the southwest of Maja Jezerce summit in the Dolu Popluks cirque which drains towards the Cafa e Pejes pass (also known as Čafa Pajs or Qafa e Pejës). He also reported the presence of numerous perennial snow patches on the northern slopes of Maja Kolats (2554 m a.s.l.) and Maja Rošit (2522 m a.s.l.) to the north of Maja Jezerce. However, ROTH VON TELEGD (1923) did not mention the Llugu i Zajave cirque where HUGHES (2009) found three glaciers and this area remained blank on his maps. Of course, it is difficult to assess which of Roth von Telegd’s perennial snow and firn fields were true glaciers. Nevertheless, the presence of much larger perennial firn fields than is the case today is certainly significant and the Prokletije glaciers are likely to have been significantly larger.

3.7 - Bulgaria

41The southernmost glacier in Europe is found in the Pirin Mountains of Bulgaria (GRUNEWALD et al., 2006, GACHEV, 2010) (Fig. 1). The Sneschnika glacier covers and area of 1 hectare (in 2005) and is present below the northern cliffs of Vihren (2914 m a.s.l.), the highest peak in the Pirin. Another glacier is present below the peak of Kutelo (2908 m a.s.l.) a few kilometres north of Vihren. These glaciers and several other perennial snow fields are found in north-facing cirques at altitudes between 2400 and 2750 m a.s.l. Little is known about the extent of Little Age glaciers in Bulgaria. However, radiocarbon dating shows that soil has been developing on the moraine crest directly in front of the Sneschnika glacier for 600 to 400 years, since 428‑611 AD (GRUNEWALD and SCHEITHAUER, 2008). This indicates a Late Holocene re-activation of glaciation well before the onset of the Little Ice Age.

3.8 - Greece

42Today in Greece, there are no modern glaciers. This is true of both the Pindus Mountains and the Olympus Massif, which lies to the east. However, evidence for Holocene glacier activity has been reported from Mount Olympus (2917 m a.s.l.), the highest mountain in Greece (Fig. 1). Here, moraines are present in the north-facing Megali Kazania, the largest cirque of the area. SMITH et al., (1997) suggested that these could be Holocene neoglacial moraines and snow patches have been known to survive the summer in recent years. The Megali Kazania moraines are situated at an altitude of c. 2200-2250 m. This is similar to the elevations of the glaciers in Albania and Montenegro to the north. However, unlike these areas, Mount Olympus is significantly drier, more southerly, and on the leeside of the Pindus Mountains. Precipitation is 2500-3000 mm in the high inland mountain areas of Montenegro (BOŠKOVIĆ and BAJKOVIĆ, 2004) and only 1800 mm on Mount Olympus (MANAGEMENT AGENCY OF OLYMPUS NATIONAL PARK, 2013). Thus it is questionable whether Olympus could have supported glaciers during the Little Ice Age. Nevertheless, perennial snowfields are likely to have been present given that snow patches occasionally survive the year today.

43In the Pindus Mountains, the highest cirque moraines have been correlated with the Younger Dryas. HUGHES et al., (2006b) argued that cirque moraines between altitudes of c. 2300-2400 m a.s.l. on Mount Smolikas (2637 m a.s.l.), the highest of the Pindus chain, must have been formed under temperatures that were much lower than today. Even under precipitation similar to today a temperature depression of more than 6°C is required to sustain glaciers (HUGHES et al., 2006). Direct precipitation values are smaller in the Pindus Mountains (2000-2500 mm, FOTIADI et al., 2000) compared with northern Albania and Montenegro (see reference above).Even with substantial local inputs, such as avalanching and windblown snow, sufficient snow accumulation to form glaciers is unlikely to have been achieved in Greece during the Little Ice Age. However, further research is needed in order to determine the age of the youngest moraines on Mount Olympus and in the Pindus Mountains to test the possibility of Holocene glaciers. Despite the uncertainty on the presence or absence of Little Ice Age glaciers, permanent (or semi-permanent) snowfields are likely to have occurred during the Little Ice Age in the mountains of Greece. Observations by the author of snow patches surviving as late as late July as far south as the Peloponese make this an entirely reasonable assumption.

3.9 - Turkey

44At least thirty eight modern glaciers have been identified in the mountains of Turkey and the largest Mediterranean glaciers occur in the mountains of eastern Turkey (AKÇAR and SCHLÜCHTER, 2005). In 1980, glaciers covered a total area of c. 22.9 km² (KURTER and SUNGUR, 1980). The snowline in Turkey in the 1950s was situated at between 3100 and 4200 m with the lowest snowlines on the northern slopes of the Pontic Mountains, the highest in the eastern interior near the borders with Iran and intermediate snowlines in the central Taurus range (ERINÇ, 1952). SARIKAYA et al. (2011) provide a comprehensive overview of the glacial history of Turkey and a summary of the main changes since the Little Ice Age is outlined below.

45In the Uludağ (2542 m a.s.l.), in NW Turkey 80 km south of Istanbul, Birman (1968) noted the presence of a small glacieret at an elevation of just 2300 m a.s.l. The glacieret was about 100 m in width and length and had an elevation difference of about 60 m from the upper to the lower margin. The ice had retreated about 50 m from the base of a well-developed moraine the crest of which is about 25 m high. This description bears close resemblance to the small glaciers found in cirques elsewhere in the Mediterranean mountains. ERINÇ (1952) noted that the very highest moraines of Uludağ could have formed during the Little Ice Age. More recent work by ZAHNO et al. (2010) used cosmogenic exposure ages to date glacial surfaces. No AMS measurement could be made on a sample from a small aplitic boulder on the innermost moraine ridge (sample TRU-12 in ZAHNO et al., 2010, p.1179). The fact that no cosmogenic nuclides could be detected is consistent with a young exposure age and combined with the observations of ERINÇ (1952) and BIRMAN (1968) supports a recent age for moraine formation, probably at the end of the Little Ice Age.

46More than 20 glaciers are present in the southeastern Taurus, in easternmost Turkey. This accounts for two thirds of all of the glaciers in Turkey (KURTER, 1991). The largest occur on Mount Cilo (Fig. 1) (also known as the Cilo Sat Mountains) around the highest peak of Uludoruk (4135 m a.s.l.). The longest of these emanates from an eastern cirque and extends for several kilometres (Fig. 10).

47This glacier was nearly 4 km long with an area of 8 km² in the late 20^th century and the altitude of the glacier terminus rose in altitude by c. 400 m between 1937 and 1991 (ÇINER, 2004). BOBEK (1940) mapped Pleistocene glaciers that were 9 km long in this area. Fig. 10 shows that the current glacier is less than 3 km long.

48Smaller glaciers occur in the central Taurus (Orta Toroslar), further west on the mountains of Aladağ (3756 m a.s.l.) and Bolkardağ (3524 m a.s.l.). The snowline in these areas is situated at c. 3450 m a.s.l. and glacier survival in such marginal conditions is likely to be controlled by local climatological and physiographic conditions (KURTER, 1991). In the Aladağ massif, a small glacier covering an area of < 1 km² is considered to be a glacier remnant consisting largely of dead-ice buried by rock debris and, in one karstic shaft at c. 3400 m, up to 120 m thickness of perennial snow, firn and ice is present (BAYARI et al., 2003).

49The northern slopes of the Pontic Mountains drain into the Black Sea and receive some of the highest precipitation levels in Turkey, especially in the east where mean annual precipitation exceeds 2000 mm (KURTER, 1991). As a result, the snowline on the northern slopes is between 3100 and 3400 m, rather lower than on most other mountains in Turkey. In total, at least 12 glaciers exist in the Pontic Mountains and cover an area of 2.54 km² (ÇINER, 2004). However, in the Kavron and Verçenik valleys in the Eastern Black Sea Mountains, moraines associated with a Little Ice Age advance appear to be absent with only Pleistocene moraines present in these valley systems (AKÇAR et al., 2007, 2008). This is despite the valleys draining peaks as high as 3932 m a.s.l. in the Kavron Valley (Mt Kaçkar) and 3907 m a.s.l. in the Verçenik Valley (Mt Verçenik). AKÇAR et al. (2008) suggest that dry and cold climatic conditions during the Little Ice Age might explain the absence of evidence for glaciers in these areas. Whilst Little Ice Age moraines are not present, small modern glaciers do exist (KURTER and SUNGUR, 1980; KURTER, 1991). This apparent paradox can be explained by the presence of large rock glaciers which are evident in places such as the upper Kavron Valley on Mt Kaçkar (Fig. 1) (AKÇAR and SCHLÜCHTER, 2005; AKÇAR et al., 2008) and in several places on Mt Verçenik (KURTER, 1991). It is likely that cold and dry conditions during the Little Ice Age resulted in an excess of debris supply over snow accumulation. The resulting rock glaciers, which are still active today, appear to have overrun any previous Holocene advances in this valley (AKÇAR et al., 2007). All that is left of real glaciers is restricted to the uppermost cirque headwalls, where in 1980 two glaciers covered 14 hectares on Mt Verçenik and four glaciers covered a total area of 6 hectares on Mt Kaçkar (KURTER and SUNGUR, 1980; KURTER, 1991).

50Glaciers are found on several Turkish volcanoes, including Mount Ararat (5137 m a.s.l.), Mount Süphan (4058 m a.s.l.) and Mount Erciyes (3917 m a.s.l.) (Fig. 1). Mount Ararat is covered by an ice cap of about 10 km² which extends down to c. 4100 m a.s.l. and has a snowline altitude of c. 4300 m a.s.l. (KURTER and SUNGUR, 1980). On Mount Süphan, several glaciers occur in the crater, the largest covering an area of 3 km² (KURTER, 1991; ÇINER, 2004). On Mount Erciyes, a glacier retreated from a length of 700 m in 1905 to a length of only 380 m in 1983 (ÇINER, 2004). Between 1902 and 2008 measurements have shown that the glacier retreated at a rate of 4.2 m per year, which corresponds to a warming rate of 0.9 to 1.2 °C per century (SARIKAYA et al., 2009).

3.10 - High Atlas (Morocco)

51No glaciers survive today in the Atlas Mountains (Fig. 1), despite reaching over 4000 m a.s.l. in several places. In the highest Toubkal Massif (4167 m a.s.l.) perennial snow accumulations do survive. The most notable survives in a gully on the north-facing cliffs of Tazaghart (3980 m a.s.l.) near the Lepinéy Refuge Hut (Fig. 11). This snow patch is identified with the name Névé Permanent on Moroccan topographic maps and is a well-known feature in trekking and mountaineering literature (e.g. SMITH, 2009). Long‑lasting snow fields (neiges éternelles) of the Tazaghart and others were mentioned by DRESCH (1941, p. 577). The snow patch is bounded on its lateral edges by clear sediment ridges that resemble moraines, although a stream cuts through these ridges directly in front of the snow patch (Fig. 11).

52It is likely that these moraines are recent in age and given the presence of a permanent snow field today then a Little Ice Age origin for the moraines is possible (HANNAH, 2014). HOOKER and BALL (1878) recalled stories claiming the existence of perpetual snow in the Atlas Mountains. Based on their observations of summer snowfall they noted that a “mere increase in the amount of precipitation, with little change in the general conditions of temperature of this region, might produce glaciers” (HOOKER and BALL, 1878, p. 291).

53The snow line, where snow accumulation is balanced by melting can be estimated for the High Atlas (and other mountains) using a glacier-climate model. By applying a simple degree-day model approach (HUGHES, 2008, 2009), modern temperature data from Marrakech (466 m a.s.l.) (WORLD METEOROLOGICAL ORGANISATION, 1998) can be extrapolated to higher altitudes using a standard atmospheric lapse rate of 0.6°C per 100 m altitude. Temperature data at different altitudes can then be used to calculate melting by applying a common degree-day factor of 4 mm day^-1 °C^-1 (BRAITHWAITE et al., 2006). At 4000 m a.s.l. the mean annual temperature is estimated at -1.6°C and this requires c. 2649 mm of snow accumulation (water equivalent) in order to balance melting. At Marrakech (466 m a.s.l.), the 1961-1990 mean precipitation was 281.3 mm. At the higher site of Oukaimeden (2700 m a.s.l.), just 10 km from the highest peak of Toubkal, the mean annual precipitation, albeit measured over a short period (1990-1993), was still only 362.5 mm (JABIRI et al., 2000). However, at Agaiouar (c. 1800 m a.s.l.) average precipitation was 592 mm (period unknown) with the High Atlas experiencing >600 mm per annum according to MESSERLI (1967). Whilst the High Atlas seems too dry for glacier survival, avalanching and windblown snow can dramatically increase accumulation and this explains the presence of the lowest modern-day glaciers in the Mediterranean mountains (e.g. HUGHES, 2008, 2009). The Tazaghart Névé Permanent sits below the highest cliffs of the High Atlas and the most extensive high plateau (2 km² above >3800 m a.s.l.) in North Africa. Thus, this site is prone to very large accumulations of the snow (DRESCH, 1941, photo 2, planche XXIX) and the last glaciers of the High Atlas are likely to have been present here, as recently as the Little Ice Age.

4 ‑ Future glacier prospects in the Mediterranean mountains

54It is clear that many glaciers in the Mediterranean mountains have retreated considerably during the 20^th Century. This is consistent with the retreat of glaciers around the world. In the Mediterranean mountains as well as many other parts of the world glacier retreat accelerated between 1971 and 2009 (IPCC, 2013). Rising temperatures have resulted in a reduction in spring snow cover in the northern hemisphere and the IPCC (2013, p. 17) state that this and associated the retreat of glaciers is likely to be caused by anthropogenic influences. Few real glaciers still survive in the Mediterranean mountains and instead we are left with numerous static ice patches. Many of these ice patches have formed as a direct result of glacier decay. GRUNEWALD and SCHEITHAUER (2010) noted that if temperature rises are as large as predicted, then all southern European glaciers will disappear. The most recent IPCC report (2013) states that global mean surface temperature change for the period 2016–2035 relative to 1986–2005 is likely to be in the range of 0.3°C to 0.7°C (medium confidence). The change between the end of the Little Ice Age and the end of the 21^st century is much greater with the global surface temperature change for the end of the 21st century likely to exceed 1.5°C relative to 1850 to 1900. However, precipitation changes are equally important and these were a major control on glacier expansion and retreat during the Little Ice Age (see below). This is more difficult to predict and the IPCC (2013, p.18) report states “Changes in the global water cycle in response to the warming over the 21st century will not be uniform. The contrast in precipitation between wet and dry regions and between wet and dry seasons will increase, although there may be regional exceptions”. Climate change projections modelled by GIORGI and LIONELLO (2008, their Fig. 6) suggest that the period 2071-2100 AD will see drier winters in the southern Mediterranean region although the northern Mediterranean winters will be wetter. Their prediction maps for winter precipitation suggest that the glaciers of the Pyrenees, Alpes Maritimes, Slovenia and NE Turkey could be least at risk from climate change, whereas the glaciers of the Apennines, Montenegro, Albania, Bulgaria and SE Turkey could experience the effects of both higher summer temperatures and lower winter precipitation.

55It is clear that the larger Little Ice Age glaciers in Europe were the result of higher precipitation than today. For example in the Alps, winter precipitation was 25% higher than the 20^th century average during the final phase of Little Ice Age glacier advance between 1760 and 1830 AD. Subsequent glacier retreat after 1830 AD was driven by a reduction in precipitation (VINCENT et al., 2005). There is widespread evidence that the Little Ice Age was wetter in both the eastern and western parts of the Mediterranean region (e.g. XOPLAKI et al., 2001; MORELLÓN et al., 2011) and it is very likely that larger glaciers in the mountains of this region were the result of wetter conditions as well as cooler temperatures than today (cf. GROVE et al., 2001).

56Most, if not all, of the surviving glaciers in the Mediterranean mountains can be considered remnants of the Little Ice Age (GRUNEWALD and SCHEITHAUER, 2010). However, whether Little Ice Age glaciers can be considered remnants of the Pleistocene glaciations is another question and is difficult to test, although under certain local topoclimatic conditions it is possible that ice masses have survived through the Holocene. Whilst this paper has highlighted the evidence for Little Ice Age glaciers around the Mediterranean region, there is great uncertainty in many areas as to the number of Little Ice Age glaciers that were present. This is simply because in many areas research has not been undertaken investigating this issue.

57The current Mediterranean glaciers survive in niche localities and there is evidence that they are currently decoupled from the regional climate and are resistant to changes in the wider prevailing temperature and precipitation patterns. This is because of very strong local topoclimatic controls, such as avalanching or windblown snow, which can dramatically increase accumulation (see references in HUGHES, 2008, 2009).

58In some cases, debris supply from surrounding rock walls exceeds snow accumulation. If glaciers become buried then the most likely life path for cirque glaciers in decline is from glacier to ice patch to permafrost (buried ice) and, potentially, the formation of a rock glacier if the ice-debris mass is able to move and flow down-valley (e.g. HUMLUM, 1998). The final phase of buried ice can potentially survive climate warming because debris-covered ice is known to be decoupled from atmospheric climate change (REZNICHENKO et al., 2011). Permafrost is able to maintain active processes such as rock glacier dynamics long after the ground initially became frozen (e.g. RIBOLINI and FABRE, 2006). This is important because it means that ice may have survived throughout the Holocene in the cirques of the Mediterranean mountains. The presence of buried ice bodies would have crucial for re-activation of glaciers during sustained periods of positive mass balance such as the Little Ice Age, and will be crucial for any future re-growth of glaciers. Once the ice is gone, and ice patches themselves disappear, then the potential for glaciers reforming in periods of positive mass balance is much reduced. In this scenario ice patches must build up through the transformation of snow packs to ice masses. The transition of snow to static ice and then to dynamic ice is inherently slower in this process than when static ice is already present.

5 - Conclusions

59The Little Ice Age was a period of glacier expansion in the Mediterranean mountains. In some areas there is evidence that glaciers formed for the first time since the Pleistocene, whilst in others glaciers reactivated or expanded from existing ice patches and smaller glaciers. Since the end of the 19^th Century glaciers have been retreating all over the Mediterranean mountains, although minor advances and stabilisations are evident over this period, especially in the 1960-1980 period. Some of the southernmost glaciers have now disappeared and all that remains are ice patches buried under rock debris. The transition from active glacier to patches of sporadic permafrost is important because the latter can be decoupled from regional climate trends as a result of strong local topoclimatic controls. Many Mediterranean glaciers that still survive today are also the result of very strong local topoclimatic controls and many of the glaciers are climatic anomalies and should not exist considering the prevailing regional climates. In many areas, current glaciers are considered to be relics of the Little Ice Age and this period of climate cooling was a major period of landscape evolution in the highest cirques of the Mediterranean mountains. Whilst there are several areas around the Mediterranean where research has been undertaken, there remain many more areas where the number, extent and timing of Little Ice Age glaciers remains unknown. Thus, there is a clear need for further research on Little Ice Age glaciation in the Mediterranean mountains.