Small-scale moisture availability increase during the 8.2-ka climatic event inferred from biotic proxy records in the South Carpathians (SE Romania)

Sediment sampling

The sediment core TDB-1 was taken in the central part of Lake Brazi with a modified Livingstone piston corer in 2007. Sediment lithology was described in the laboratory (details in Magyari et al., 2009).

Radiocarbon dating

In all, 20 radiocarbon dates are available from Lake Brazi (Table 1). For this study, seven new AMS radiocarbon dates were obtained in the Hertelendi Laboratory of Environmental Studies at ATOMKI in Hungary (Molnár et al., 2013). ¹⁴C dates were calibrated by CALIB Rev 6.1.0 (Stuiver et al., 2011). Outlier ¹⁴C dates were detected in a Bayesian age–depth modeling program BACON (Blaauw and Christen, 2011).

OPEN IN VIEWER

Table 1.

Radiocarbon dates from Lake Brazi, core TDB-1.

Core	Laboratory code	Dated material	Depth (cm)	14C age years BP	Calibrated range years BP	Error of the average years BP	Remarks
TDB-1	Poz-26103	Picea abies needles	119	725 ± 30	652–723		outlier
TDB-1	I/338/1#	>180 µm fraction, plant macrofossil	127	375 ± 25	319–503	411 ± 92
TDB-1	I/338/2#	>180 µm fraction, particular organic matter	127	1018 ± 23	913–970		outlier
TDB-1	Poz-26104	Pinus mugo cone scale	160	1735 ± 30	1562–1712	1637 ± 75
TDB-1	I/338/3#	P. mugo shoot	204	2611 ± 23	2724–2763	2743.5 ± 19.5
TDB-1	Poz-206106	P. mugo cone	238	3045 ± 30	3205–3356	3280.5 ± 75.5
TDB-1	I/338/4#	>180 µm fraction, plant macrofossil	280	3962 ± 30	4381–4520		outlier
TDB-1	I/338/5#	>180 µm fraction, particular organic matter	280	3987 ± 26	4416–4521	4468.5 ± 52.5
TDB-1	Poz-26107	Pinus twig	315	5040 ± 40	5708–5902	5805 ± 97
TDB-1	Poz-26108	P. abies needles	355	6320 ± 40	7163–7324	7243.5 ± 80.5
TDB-1	I/338/6#	>180 µm fraction, plant macrofossil	391	6925 ± 30	7683–7828	7755.5 ± 72.5
TDB-1	Poz-26109	P. abies needles	393	6130 ± 40	6926–7160		outlier
TDB-1	Poz-26110	P. abies needles and seed	450	8240 ± 50	9072–9326	9199 ± 127
TDB-1	Poz-26111	P. abies needles	505	8810 ± 50	9670–10,155	9912.5 ± 245.5
TDB-1	Poz-31714	P. mugo needles	521	9150 ± 50	10,226–10,433	10,329.5 ± 103.5
TDB-1	Poz-26112	P. abies cone	545	9610 ± 50	10,766–11,167	10,966.5 ± 200.5
TDB-1	Poz-31715	P. mugo needles	557	9980 ± 100	11,216–11,826	11,521 ± 305
TDB-1	Poz-31716	charcoal	569	10,870 ± 70	12,598–12,925	12,761.5 ± 163.5
TDB-1	Poz-27305	Pinus sp. needles (2)	578	11,590 ± 60	13,287–13,620	13,453.5 ± 166.5
TDB-1	Poz-26113	P. abies cone scales	591	9690 ± 50	11,067–11,225		outlier

Since sediment accumulation rates were very different in the Lateglacial and Holocene sections of the core (Magyari et al., 2009, 2012), we calculated age–depth relationships for the Holocene section of the core separately (110–505 cm). In this paper, we use the results of the linear age–depth modeling as provided by the software Psimpoll 4.27 (Bennett, 2007), which uses linear interpolation between the median values of the 2σ calibrated age ranges. For this age–depth model, we used 12 radiocarbon dates between 550 and 110 cm.

Sediment chronology

Figure 3 shows the results of linear age–depth modeling. Overall, we excluded five dates from the age–depth modeling using the outlier detection function of the Bayesian age–depth modeling software Bacon. These five dates (shown in gray in Table 1) were stratigraphically inconsistent with the majority of the ¹⁴C dates (Figure 3). The sediment section between dated levels 391 cm (7755 cal. yr BP) and 450 cm (9200 cal. yr BP) includes the 8.2-ka event. Between these two radiocarbon dates, spanning more than 1000 years, we assumed uniform sediment accumulation rate. Results of the LOI analysis (Figure 3) suggest that abrupt changes in sediment accumulation did not characterize this time interval; only one sample showed higher LOI value at 8100 cal. yr BP, suggesting that sediment accumulation was likely not linear in this sediment section. The deposition time is 24.5 yr/cm in this section, that is, each sediment sample represents ~25 years. We note, however, that the modeled calibrated BP ages of the sediment samples have some uncertainty because of the relatively low number of ¹⁴C dates around the 8.2-ka event. A smooth spline age–depth model was used in a separate study focusing on the Holocene fire history, but the two models agree well in the early Holocene (maximum age difference <30 years; Finsinger et al., 2014). We only use average fire return intervals calculated on this timescale from this study.

OPEN IN VIEWER

Figure 3.

Age–depth model for the Holocene section of the Lake Brazi TDB-1 core (between 111 and 521 cm). The model is based on 12 ¹⁴C dates, calibrated using CALIB Rev 6.1.0 and age–depth modeling using linear interpolation in Psimpoll 4.27. Note that the top of the sediment is at 111.14 cm; sediment depth calculation included the lake water column. Loss-on-ignition values are also shown on the right. The red rectangles (gray in the printed version) highlight the sediment section that encompasses the 8.2-ka event.

Pollen stratigraphy, pollen-inferred vegetation changes and fire history

The pollen percentage diagram is presented in Figure 4 and Supplementary Appendix 2 (available online). Figure 4 includes the major terrestrial pollen types, with particular attention to those that show distinct changes around the 8.2-ka event. Eight statistically significant local pollen assemblage zones (B-4 to B-11) were identified in the Holocene part of the record (Supplementary Appendix 2, available online). Here, we describe in detail only three pollen assemblage zones: one which predates (B-6), one which includes (B-7), and one which postdates (B-8) the 8.2-ka event (Figure 4). The main characteristics of these pollen zones, such as arboreal and non-arboreal pollen, micro- and macrocharcoal, total terrestrial pollen accumulation rates (PARs), and the dominant trees and shrubs, are summarized in Supplementary Appendix 1 (available online), while fire return intervals (FRIs) and background charcoal are displayed in Figure 5.

OPEN IN VIEWER

Figure 4.

Changes in relative frequencies of the main pollen taxa and stomata, as well as the micro- and macrocharcoal accumulation rates, core TDB-1, Lake Brazi, Retezat Mts, Romania. Macrocharcoal accumulation rates were recalculated to constant sample interval (40 years).

OPEN IN VIEWER

Figure 5.

Background and fire return intervals (FRIs, years between consecutive detected fire episodes) as previously quantified in Finsinger et al. (2014). Calculated on interpolated (40 years) macroscopic charcoal accumulation rates that are based on charcoal-area (CHARa) and charcoal-number (CHARc) measurements.

B-7 pollen zone, 436–334 cm, 8870–6520 cal. yr BP

This pollen assemblage zone is characterized by continuing high relative frequencies of arboreal pollen (95%). On the basis of the pollen and stomata records, the forest around the lake was dominated by P. abies with admixture of P. mugo and P. cembra as attested by the occurrence of their stomata. The absence of A. alba stomata suggests its withdrawal from the lakeshore. The pollen percentages of P. mugo rapidly decreased from values of 10% to significantly lower values at about 8700–8600 cal. yr BP. Other trees, such as Quercus, F. excelsior, and Ulmus, had stable pollen values, and we infer that they played a significant role at lower altitudes, reflecting the stability of the mixed oak forest zone. The main feature of the zone is the increasingly high pollen frequencies of Corylus (maximum 36% at 8000 cal. yr BP), suggesting its expansion in the lower altitude mixed-deciduous forest zone and likely also in the spruce zone similar to other mountain ranges in the Eastern Carpathians (Feurdean, 2005; Feurdean et al., 2008; Tanţău et al., 2011).

Macrocharcoal accumulation rate values were generally lower and showed distinct, but lower amplitude peaks than in the preceding pollen zone. The microcharcoal record showed small peaks between 7400 and 7100 cal. yr BP and distinct peaks in the high-resolution section spanning across the 8.2-ka event. Interestingly, the two charcoal records showed different patterns around the 8.2-ka event: whereas macrocharcoal accumulation rates were high between 8450 and 8300 cal. yr BP (see also the lowest FRI values on Figure 5 around 8500–8300 cal. yr BP; 100–200 years) and thereafter rapidly declined until 7700 cal. yr BP, five large microcharcoal peaks were detected between 8300 and 8050 cal. yr BP. This likely suggests that although regional fire activity increased between 8300 and 8050 cal. yr BP, the surroundings of Lake Brazi were not affected by local forest fires at that time. In association with the 8.2-ka event, we found characteristic changes in the relative frequencies and PARs of several deciduous trees, shrubs, and herbs (Figure 6). Monolete fern spores and Poaceae show distinct pollen percentage and PAR peaks during most of the microcharcoal peaks, suggesting the opening up of the vegetation and early succession after the forest fires, likely at lower altitudes. The percentages of deciduous pollen taxa (Quercus, F. excelsior, Ulmus) show comparable declines; however, PAR values increase along with microcharcoal peaks between 8300 and 8200 cal. yr BP. Since this increase is present in nearly all pollen taxa, this likely does not indicate real population increases, but an abrupt change in sediment accumulation rates.

OPEN IN VIEWER

Figure 6.

Changes in (a) pollen percentages and (b) pollen accumulation rates of the selected pollen taxa, micro- and macrocharcoal accumulation rates between 8000 and 8500 cal. yr BP from Lake Brazi, core TDB-1.

The stomata record attests to changes in the local vegetation around the 8.2-ka event as well (Figure 4). Around 8250 cal. yr BP, P. abies stomata temporarily disappeared, while P. cembra stomata reached maximum relative frequencies. Around 8150 cal. yr BP, the relative frequencies and PARs of Carpinus betulus increased temporarily (from 1% to 6%), suggesting its rapid short-lived population expansion at lower altitudes between 8200 and 8100 cal. yr BP. C. betulus likely colonized forest openings. However, it seems to have failed expanding further, as pollen percentages decreased rapidly back to values <1%. During this interval, two other taxa, Corylus and F. excelsior, showed significant relative frequency decreases. If we look at their PAR values (Figure 6), it becomes clear that F. excelsior showed real population size decrease, while Corylus populations likely did not decrease on the northern slope since its percentage decrease is not accompanied by PAR decline; on the contrary, PAR values increase, suggesting minor population increase. Also, a typical feature of this pollen zone was the episodic occurrence and pollen percentage increase in C. betulus three more times: at 8750, 8550, and 7500 cal. yr BP. These episodic increases were associated with Corylus declines, suggesting that the short-term growth of C. betulus was likely connected to forest disturbances, but these were not always associated with detectable fire events.

Macrofossil-inferred local vegetation between 9100 and 6900 cal. yr BP

According to the macrofossil diagram (Figure 7), three conifer species were present around the lake between 9100 and 6900 cal. yr BP: P. abies, P. mugo, and P. cembra. Macrofossils of two more conifer species, Larix decidua and A. alba, were present prior to 9900 cal. yr BP, suggesting that the early Holocene forest around Lake Brazi was species rich, and likely had a more open character.

OPEN IN VIEWER

Figure 7.

Changes in the concentration of woody plant macrofossils between 4900 and 10,500 cal. yr BP in core TDB-1, Lake Brazi, Romania.

P. abies was the most abundant in the studied period, pointing to its dominance on the lakeshore. P. mugo needles were found in low concentration; however, its male blossoms were more abundant, suggesting continuous presence. P. cembra needles were found less regularly without other vegetative parts, suggesting low abundance on the lakeshore. Around 8200 cal. yr BP, compositional change was not seen in the macrofossil record; P. abies bud scales attained exceptionally high concentration at 8300 cal. yr BP; this peak, however, represents a loose bud in the sample. The only notable changes are the temporary disappearance of P. mugo after 8200 cal. yr BP, between 8000 and 7800 cal. yr BP, and the absence of P. cembra between 8700 and 8050 cal. yr BP. The stomata-inferred increase in P. cembra and decrease in P. abies around 8250 cal. yr BP are not supported by the macrofossil record; only P. abies shows slightly decreased concentrations. Note, however, that the macrofossil record has lower time-resolution. P. abies concentrations increase steadily from 8130 cal. yr BP, likely indicating denser spruce forests around the lake after this time.

LOI, BiSi, and siliceous algae records

The siliceous algae record is presented in Figure 8 along with DI-pH, C:D ratio, BiSi, LOI, and major diatom life form groups. Here, we describe two diatom assemblage zones (DAZ-8 and DAZ-9; Buczkó et al., 2013), of which the first one precedes and the second one includes the 8.2-ka event. Changes in LOI and BiSi are also discussed.

OPEN IN VIEWER

Figure 8.

Relative frequencies of selected diatom taxa, loss-on-ignition, biogenic silica, diatom life form groups, and Chrysophyte:Diatom (C:D) ratios from Lake Brazi, core TDB-1. On the right, diatom-inferred pH (DI-pH) values are also shown. DAZ: Diatom Assemblage Zone. Aulacoseira valida and Aulacoseira pfaffiana relative frequency curves have blue fillings (gray in the printed version); these two planktonic/tychoplanktonic taxa show the strongest response at the 8.2-ka event.

Interpretation of the ecosystem changes around the 8.2-ka event

The high-resolution pollen, stomata, and charcoal records allowed identifying distinct vegetation composition changes in the Retezat Mts during the studied RCC event. The early Holocene forest between c. 9600 and 8900 cal. yr BP was composed of P. abies, P. cembra, and P. mugo around Lake Brazi likely with a minor admixture of A. alba. Since this latter tree species has an upper elevation limit of 1400 m in the Retezat today (Nyárádi, 1958), we infer that summer temperatures were higher than at present in the Retezat in this period, likely in association with higher than present summer insolation (Berger and Loutre, 1991; Feurdean et al., 2013). At lower altitudes, mixed-deciduous forests were dominated by Ulmus, Quercus, Tilia, and F. excelsior. A major compositional change occurred around 8900 cal. yr BP. The increase in Corylus pollen suggests the formation of a Corylus dominated mixed open forest zone. The spread of Corylus around 9500 cal. yr BP, which displays in the continental interior of Europe an altogether different pattern compared with Northwest Europe (Giesecke et al., 2011), was detected in several other Carpathian pollen diagrams (Feurdean, 2005; Feurdean et al., 2013; Tanţău et al., 2011) and was connected to macroclimate change (Feurdean et al., 2008). Finsinger et al. (2006) found a positive correlation between the increase in Corylus pollen and fire activity in the Southern Alps. The coincidence between the prominent peak in the micro- and macrocharcoal records and the onset of the Corylus pollen increase in the Retezat Mts around 9500 cal. yr BP (Figure 4) may suggest that fires possibly played an important role in the initial population expansion of the light-demanding Corylus. However, the charcoal record does not support the view that fires played an important role in the long-term population increase that peaked about 1500 years later, as suggested by Huntley (1993). Despite the decreasing trend of the summer insolation curve after 9500 cal. yr BP, a chironomid-based summer temperature reconstruction from the same sediment record indicates that July temperatures were highest between 9500 and 8700 cal. yr BP (Tóth et al., 2015). We may thus hypothesize that warmer/drier summers were sufficient to favor the expansion of Corylus even in the absence of higher fire activity.

Within this Picea–Corylus dominated pollen phase, distinct changes occurred between c. 8300 and 8100 cal. yr BP, when microcharcoal accumulation rates increased repeatedly, without marked macrocharcoal increases, suggesting episodic forest fires in the region, likely in the lower elevation mixed-deciduous forest zone, but lower fire activity around the lake. Moreover, we found ambiguous changes in the representation of P. cembra around the lake with the stomata record indicating its increasing abundance, while the macrofossil record suggested sporadic appearance. The increase in C. betulus at ~8150 cal. yr BP occurred coincidently with the third microcharcoal peak, but the appearance and first increase in C. betulus were coincident with the first microcharcoal accumulation peak at 8300 cal. yr BP (Figures 4 and 6), and some of its earlier and later temporary appearances also coincided with increased micro- and macrocharcoal intervals (e.g. at 7600, 8500, and 10,900 cal. yr BP, but not at 8750 cal. yr BP; see Figure 4 and Supplementary Appendix 2, available online). One possible interpretation of these changes is that C. betulus colonized the forest openings of the mixed-deciduous forests, but was soon overtaken by Corylus and F. excelsior as its PAR values and relative frequencies both decreased. Alternatively, the closure of the forest openings decreased its pollen production and uphill pollen transport in low fire activity periods; hence, it became less visible in the subalpine pollen record.

In the context of the 8.2-ka RCC event, the coincident vegetation response was the short-lived further expansion of C. betulus. It is evident from the organic content, vegetation (Figure 8), and chironomid-inferred July temperature records of the same deposit (Tóth et al., 2015) that the 8.2-ka event is incised in the most productive, warmest summer interval of the Holocene in the Retezat Mts, when reconstructed July temperatures were 1.5–1.9°C above modern values. In this context, the temporary expansion of C. betulus can be interpreted in two alternative ways. According to the first interpretation, increased forest fire activity and secondary succession in the forest areas affected by fire facilitated the establishment and temporary expansion of C. betulus on the northern slopes of the Retezat Mts from 8300 cal. yr BP. Studies on fire sensitivity have shown that C. betulus generally benefits from ground fires (Tinner et al., 2000). As it was favored against F. excelsior and Corylus avellana that are also early successional, fire-adapted species, other interspecific competitions, or climatic factors might have also played a role in the temporary increase in C. betulus at 8150 cal. yr BP and also at the earlier temporary expansion. Bioclimatic parameters of the affected woody species suggest that C. betulus has the lowest tolerance to winter cold (T_c min: −8°C), while its drought resistance is weaker than the other two species (α* = 0.7 against 0.65 for F. excelsior and 0.55 for C. avellana; Sykes et al., 1996). These parameters suggest that the temporary expansion of C. betulus during the 8.2-ka climatic anomaly was likely a response to increasing moisture availability between 8200 and 8100 cal. yr BP, the interpretation of which is supported by the coincident increase in planktonic/tychoplanktonic diatoms suggesting water-level increase in the same period. Under this scenario, the role of the repeated regional forest fires was the creation of space for its early establishment (~8300 cal. yr BP), while the subsequent increase in available moisture helped its temporary spread. This interpretation is also supported by the macrocharcoal record that shows low fire activity between 8300 and 7800 cal. yr BP (average fire return interval increased from 200–300 years to 1200 years, Figure 5) that may be related to moister summer conditions in the subalpine zone during the 8.2-ka event. As far as the summer temperatures are concerned, the chironomid-inferred July temperature reconstruction from Lake Brazi does not indicate decrease around 8200 cal. yr BP, but before, between 8700 and 8500 cal. yr BP, when C. betulus showed also episodic advances at two times, while the diatom-based δ¹⁸O record indicated winter moisture increase between 9000 and 8500 cal. yr BP (Magyari et al., 2013; Tóth et al., 2015). Note, however, that the resolution of the δ¹⁸O_DIAT and chironomid reconstructions is much lower (~80 years) than the pollen, diatom, and charcoal records (~20–25 years). The terrestrial proxies-based interpretation of the ecosystem and climatic changes around the 8.2-ka event is also supported by coincident changes in the lake ecosystem. The high-resolution proxy records for organic and BiSi content and for diatom compositional change (Figure 8) suggest that the 8.2-ka event disrupted or ended a phase of decreasing diatom production (indicated by decreasing BiSi values) between 8900 and 8250 cal. yr BP. Second, the temporary increase in planktonic/tychoplanktonic diatoms at ~8150 cal. yr BP, exactly at the time when C. betulus pollen increased, suggests that during the major diatom bloom period in spring the lake had high turbulence and increased water depth, which is inferred from the rapid increase in A. valida. An increase in this floating and strongly silicified diatom species has also been detected in North American lake deposits during the 8.2-ka event, where it was also interpreted to be indicative of water turbulence and/or increased water depths in spring (Spooner et al., 2002). Moreover, a recently built training set, based on diatom distribution in 40 South Carpathian lakes, clearly shows positive and significant correlation between increasing water depth and increasing relative frequencies of A. valida (Buczkó, unpublished) further supporting that the temporary increase in A. valida indicates water depth increase at ~8150 cal. yr BP. On the other hand, increasing water turbulence may also explain the increase in A. valida without an accompanying water-level increase, as it was formerly discussed in Buczkó et al. (2013) using examples from prairie lakes in North America. However, a local training set was not yet available at that time. All in all, the diatom data suggest that during the spring season, the lake received increased water discharge (either by increasing spring rainfall or snowmelt) at the same time when vegetation reorganization pointed to increasing moisture availability. The proxy records also suggest that after this short episode, the lake system reverted to benthic diatom assemblages, decreased BiSi, and increasing LOI values, all suggesting rapid decrease in water depth and expanding lakeshore mire vegetation. It may also be inferred from these data that available moisture likely increased in the early part of the vegetation season, while at least in some years late summers were warm and dry allowing for the prevalence of occasional fires in the region.

Although the above interpretation of the high-resolution proxy records is convincing since terrestrial and lake proxies are in agreement, we cannot exclude an alternative explanation of the short-lived C. betulus expansion between 8200 and 8100 cal. yr BP. Although we pointed out some differences in the climatic tolerance of the woody species that showed relative frequency and accumulation rate changes during the short climatic perturbation, the difference between the available moisture tolerance of the two most antagonistically behaving species, F. excelsior and C. betulus, is relatively little. F. excelsior is often more abundant in wet alluvial soils, given its tolerance to seasonal inundation (Borhidi et al., 2012), but it is also a characteristic component of low-built scree forests in the Carpathian foreland that are exposed to strong winds, high summer insolation, and poor soils, that is, extreme habitats (Borhidi et al., 2012). Furthermore, F. excelsior requires less accumulated heath during the vegetation season and tolerates lower winter temperatures than C. betulus – these properties allow the species to expand much further north in Europe (Sykes et al., 1996). Therefore, the expected cooling at 8200 cal. yr BP is unlikely to have been the main factor alone that favored the expansion of C. betulus (Alley et al., 1997; Wiersma and Renssen, 2006). We discussed above that C. betulus benefits from increasing soil moisture, while the light-demanding F. excelsior and C. avellana can better cope with enduring drought stress under strongly continental climatic conditions that characterized the early Holocene in the Carpathians (Feurdean et al., 2013). Climate simulations suggest that early summers were up to 4°C warmer and much drier in the Carpathian region in the early Holocene (Feurdean et al., 2013). C. betulus cannot cope with hot and dry summers, but its seedlings are light-demanding; therefore, its establishment requires canopy gaps that were overall frequently provided in the early Holocene high fire activity and low fire return interval ecosystems of the Carpathian region (Feurdean et al., 2012, 2013; Finsinger et al., 2014). So an alternative explanation of the short-lived spread of C. betulus at 8200 cal. yr BP can be the alteration of relatively dry/warm summer years that triggered canopy fires with cool/moist summer years that in association with the gradually increasing winter temperatures enabled the (1) establishment and (2) temporary spread of C. betulus at low–mid altitudes. This second alternative interpretation of the proxy records differs from the first in emphasizing the complexity of the climate change that involved alternating warm/dry summer years and cool/moist summer years during the 8.2-ka event. Irrespective of which data interpretation is accepted, the main increment of the climatic perturbation was the establishment and temporary expansion of a new canopy component in the early Holocene forests, which finding agrees well with the conclusion of other studies (Tinner and Lotter, 2006) that pointed out the importance of short-lived climatic perturbations in the establishment of new canopy components in climax forests.

Regional comparisons

The onset of the increased regional biomass-burning rates in the deciduous forest zone of the Retezat Mts at ~8300 cal. yr BP predates the maximum cooling of 3.3 ± 1.1°C in Greenland at c. 8175 cal. yr BP (Kobashi et al., 2007), but agrees with the outburst date of Lake Agassiz, 8470 ± 300 cal. yr BP (Hillaire-Marcel et al., 2007), even when we take into account the age–depth model’s uncertainty (Figure 3). The spread of C. betulus and increase in the planktonic A. valida at 8150 cal. yr BP, on the other hand, lag maximum cooling above Greenland only by 25 years, which is within the dating uncertainty of the Lake Brazi record. Overall, the terrestrial ecosystem changes in the Retezat Mts seem to start around 8300 cal. yr BP and culminate at 8150 cal. yr BP, which timing is consistent with several other European biotic proxy records. For example, a high-resolution isotopic record and sedimentological changes in a Danish lake (Bjerring et al., 2013) also found a two-stage lake-level response around the 8.2-ka climatic anomaly. Their results indicated a lake-level decrease followed by an abrupt increase, with both events taking place within the window of 8390–8210 cal. yr BP. Since their results partially disagreed with Magny’s inference of increased humidity during the 8.2-ka event in Northern Europe (Figure 1). Bjerring et al. (2013) concluded that the hydrological effects of the 8.2-ka event may be substantially more complex than suggested by the latitudinal borders. In a Swedish lake (Lake Kälsjön), the increase in Aulacoseira species in the planktonic diatom assemblages was dated between 8500 and 8200 cal. yr BP and was explained by wind-induced turbulence (Randsalu-Wendrup et al., 2012) similarly to Lake Brazi, but over a prolonged time interval. Cooling was inferred from the spread of a centric planktonic diatom at c. 8200 cal. yr BP as well as from the temporary decrease in Betula and Pinus PARs.

Comparing the vegetation responses established in this study to the climatic anomaly with similar records from Europe reveals several similarities with responses in the Carpathians and Balkans (Feurdean et al., 2008; Panagiotopoulos et al., 2013), but antagonistic vegetation changes in Northern and Northwestern Europe all emphasizing the role of cooling further north and west (Kofler et al., 2005; Ralska-Jasiewiczowa et al., 1998; Seppä, 1998, 2007; Tinner and Lotter, 2001; Veski et al., 2004).

In Northern Europe, the pollen influx decline of spring-temperature sensitive trees was recorded in Lake Rõuge in Estonia between 8250 and 8150 cal. yr BP with a simultaneous increase in Betula accumulation rates suggesting lower temperatures in early spring (Veski et al., 2004). This inference was also corroborated by quantitative pollen-based climate reconstruction that showed a 2°C drop in annual mean temperatures. Seppä et al. (2007) found a major decrease in pollen percentages of thermophilous tree taxa (especially Corylus, Alnus, and Ulmus) south of 61°N and interpreted it as reflecting decreasing temperature in spring. These data suggest that Northern and Northeastern Europe experienced climate cooling during the 8.2-ka event, consistently with general circulation modeling results (Wiersma et al., 2011).

Further south in the Alps, the pollen percentage of Corylus decreased significantly (from 40% to 16%) and large-scale expansion of Pinus, Betula, Tilia, Fagus sylvatica, and A. alba was detected at Soppensee and Schleinsee (Tinner and Lotter, 2001). The authors suggested that climatic cooling reduced drought stress and this allowed more drought-sensitive and taller growing species to out-compete C. avellana by forming denser forest canopies. This vegetation reorganization had a long-lasting consequence and suggested increasing available moisture during the vegetation season, partially in agreement with our results in the South Carpathians for the period limited to the 8.2-ka event. However, while at Soppensee charcoal was not correlated with pollen, the microcharcoal record from Lake Brazi shows good correlations with some of the pollen types, suggesting causal relationships between episodic fires and vegetation composition between 8355 and 8000 cal. yr BP. Furthermore, the terrestrial vegetation response was much stronger in the Alps and involved the expansion of five tree species, all pointing to cooler and moister summer conditions, while in the South Carpathians we found the expansion of a single tree species (C. betulus) that attested to modest available moisture increase, but no significant summer temperature decrease could be inferred from its spread. We infer from this comparison that the differences in the terrestrial vegetation response between the two large European mountain ranges are attributable to their different continentality levels; summer cooling in the Alps was likely more significant during the 8.2-ka event leading to the expansion of tall-growing but cool-summer tree species, while in the South Carpathians the expanding tree species suggested the alteration of warm/dry summer years with cool/moist summers and available moisture increase in the early part of the vegetation season. More significant cooling in the Alps is also attested by stalagmite oxygen isotope data (e.g. Boch et al., 2009) that indicate rapid cooling by ~3°C at 8175 cal. yr BP.

Pollen and plant macrofossil records that show similar changes to Lake Brazi include the Steregoiu peat bog in the Northwest Carpathians, where the pollen records suggested that between 8300 and 8400 cal. yr BP, P. abies, Ulmus, and Corylus increased in combination with the episodic expansion of Fagus sylvatica (Feurdean and Bennike, 2004; Feurdean et al., 2008). The latter species has similar ecological requirements to C. betulus, but it is late successional and has higher moisture requirement. These findings suggest more significant moisture availability increase in NW Romania during the 8.2-ka climatic oscillation, which inference is well supported by the climatic differences between the two areas. The mountains of NW Romania are influenced more strongly by the Atlantic westerlies and less by Mediterranean water sources (Drăguşin et al., 2014), which difference means that in times of intensifying westerly circulation, this area receives more precipitation from the Atlantic ocean. Another NW Romanian low altitude site that showed vegetation response around 8200 cal. yr BP is Turbuta. Here, the temporary spread of C. betulus shows up in the pollen record around 8200 cal. yr BP, similarly to Lake Brazi, but the authors do not interpret this pollen compositional change (Feurdean et al., 2007). Similarly to Brazi, the episodic expansion of C. betulus likely responds to forests disturbance at low altitude and the species was likely facilitated by increased winter/spring moisture in this area.

In connection with the fire events, a recent Holocene fire regime study from lowland Transylvania clearly showed that the 8.2-ka event appeared in an early Holocene fire zone (10,100–7100 cal. yr BP) that was characterized by frequent high-intensity fires (mean fire return interval: 112 years, fire frequency: 9 fires/100 years; see Feurdean et al., 2013). This study also showed modeled early summer temperatures with the highest values between 11,500 and 8300 cal. yr BP, that is, the driest/warmest summer interval ended by the time of the 8.2-ka event. Although fire activity was high before 8200 cal. yr BP, this study recognized the 8.2-ka event as a short-lived decline in biomass burning that the authors associated with cool/moist summers. It is notable that these inferences are based on the macrocharcoal record that agrees well with the Lake Brazi macrocharcoal record and suggest decreasing fire activity in the vicinity of both sites. Other Holocene RCC events were also recognized as low fire intervals (e.g. 4200, 2800 cal. BP) by Feurdean et al. (2013) and associated with cool/moist summers. Overall, the regional picture of fire histories seems to be consistent when the same proxies are applied, but the Lake Brazi microcharcoal record deviates, and at least in the Retezat Mts suggests the episodic occurrence of fires during the 8.2-ka event. Climatic changes around the 8.2-ka event have also been inferred from stalagmite isotope records in the Romanian Carpathians. These show a minor decrease or no significant change in δ¹⁸O composition (e.g. Ursilor and V11 caves in Romania; Onac et al., 2002; Tămaş et al., 2005), suggesting no significant change in annual temperatures. However, δ¹³C values increased in the V11 and Sofular Caves, the latter of which is located further south in the southern Black Sea coast. This record was interpreted as being indicative of a prominent decrease in moisture availability over a prolonged time interval from c. 8400 to 7800 BP (Göktürk et al., 2011). This period agrees well with increased organic content in Lake Brazi, what we interpreted as a trend of gradually decreasing lake levels in the same period. However, we detected a short and modest lake-level increase within this period, at 8150 cal. yr BP; a correspondent decrease in δ¹³C values is, however, missing in the Sofular record (Göktürk et al., 2011), suggesting that springs were not moister during the 8.2-ka event in the southern Black Sea coast. In summary, the isotopic records from the region suggest that the 8.2-ka event appears within a longer dry period in the southern Black Sea area and remains undetectable in the isotope records.

Finally, we mention the findings of pollen and diatom studies from the Rila Mountains that are located south of the Retezat (Figure 1). These mountains are under relatively strong Mediterranean climatic influence, but still receive precipitation from Atlantic water sources (Tonkov et al., 2008). In these mountains, a short-term steep decline was detected in arboreal PARs at 8230 cal. yr BP, mainly in Pinus Subgenus Diploxylon, suggesting either dying or depressed flowering of pine trees and shrubs at high altitude in this period (Tonkov et al., 2016). In addition to this likely direct vegetation response, A. alba expanded directly after 8200 cal. yr BP, and vegetation disturbance around 8200 cal. yr BP likely facilitated its expansion (Tonkov et al., 2008). Since A. alba is the most demanding species regarding its moisture requirement among conifers (Tinner and Lotter, 2001; Tonkov et al., 2008), these data suggest that the Rila Mts also experienced climatic cooling and associated available moisture increase during the 8.2-ka event, which facilitated the spread of the more moisture demanding late successional A. alba. Climatic conditions after the 8.2-ka event, however, likely remained less continental in the Rila, further helping the advance of A. alba. It is also notable that A. alpigena, the planktonic diatom species showing the strongest response in lake Brazi at 8150 cal. yr BP, increased in abundance in Lake Sedmo Rilsko (2250 m a.s.l.; Lotter and Hofmann, 2003) around 4500 cal. yr BP. Although its increased abundance was longer lasting and appeared much later than the 8.2-ka event, it was associated with the expansion of Fagus sylvatica and increasing diatom productivity in the Rila Mts, and the authors interpreted these changes as indicative of increasing moisture availability (Lotter and Hofmann, 2003), which interpretation agrees well with our inference at Lake Brazi.

Did available moisture increase in the South Carpathians during the 8.2-ka event? Testing the hypothesis of Magny (2007)

The interpretation of the 8.2-ka event is complicated by the fact that both temperature and hydrological conditions appear to have been altered; therefore, it is difficult to disentangle whether hydrological or temperature changes are the main source of available moisture changes during this event. Since most early Holocene Central European hydrological studies agree on that water levels in lakes of forested regions were controlled primarily by winter precipitation (Carcaillet and Richard, 2000; Roberts, 1998) and a recently published diatom oxygen isotope record from Lake Brazi (Magyari et al., 2013) also suggests that fluctuation in the δ¹⁸O_DIAT values reflects alternating contribution by winter precipitation, we can safely infer that as long as the planktonic/tychoplanktonic diatom maximum indicates water-level increase at 8150 cal. yr BP in Lake Brazi, it was a response to increased winter/spring precipitation. This interpretation would agree with Magny’s meta-analysis results of lake-level anomalies in Europe around the 8.2-ka event that indicated a more humid climate accompanied by lake-level increases in mid-Central Europe (Figure 1, see Magny et al., 2003). It is obvious, however, that our lake proxies are strongly skewed toward the winter half year, whereas the terrestrial proxies are skewed toward the vegetation season (mainly summer), and the two systems show slightly different change. Our inference from this result mirrors the conclusion of Bjerring et al. (2013) that climate change was more complex and likely seasonally different during the climatic anomaly. This inference is also supported by the pollen studies of Seppä et al. (2007) that showed a strong vegetation response in Northern Europe up to 61°N, but no response in the sub-arctic areas. They suggested that this might be explained by cooling mostly during the winter and spring, to which the ecosystems in the south responded sensitively since cooling occurred at the onset of the growing season. In contrast, in the sub-arctic area, where the vegetation remains dormant longer, the cold event is not reflected in pollen-based or lake sediment–based records. Such interpretation is consistent with our results, but our proxies rather sensitively showed the changes in available moisture during winter/spring in positive direction.

Overall, if the increase in planktonic diatoms at 8200 cal. yr BP was not merely the result of increasing wind turbulence but a response to increased water depth, then our biotic proxy-based climatic and lake-level inferences support Magny’s interpretation of this climatic anomaly in that we infer increasing lake levels in the mid-European sector. However, we argue on the basis of our other biotic proxies (pollen and charcoal) that this lake-level rise principally resulted from increased precipitation during winter or spring, which was followed by alternating dry and moist summers, that is, the continentality of the area was maintained, but there were climatic years with weaker continentality, or available moisture increased mainly in the early part of the vegetation season.