Early land use and centennial scale changes inlake-water organic carbon prior tocontemporary monitoringCarsten Meyer-Jacoba,1, Julie Tolua, Christian Biglera, Handong Yangb, and Richard Bindlera

aDepartment of Ecology and Environmental Science, Umeå University, 901 87 Umeå, Sweden; and bEnvironmental Change Research Centre, UniversityCollege London, London WC1E 6BT, United Kingdom

Edited by Edward A. Boyle, Massachusetts Institute of Technology, Cambridge, MA, and approved April 23, 2015 (received for review January 23, 2015)

Organic carbon concentrations have increased in surface watersacross parts of Europe and North America during the past decades,but the main drivers causing this phenomenon are still debated.A lack of observations beyond the last few decades inhibits abetter mechanistic understanding of this process and thus a reliableprediction of future changes. Here we present past lake-waterorganic carbon trends inferred from sediment records across centralSweden that allow us to assess the observed increase on a centen-nial to millennial time scale. Our data show the recent increase inlake-water carbon but also that this increase was preceded by alandscape-wide, long-term decrease beginning already A.D. 1450–1600. Geochemical and biological proxies reveal that these dynamicscoincided with an intensification of human catchment disturbancethat decreased over the past century. Catchment disturbance wasdriven by the expansion and later cessation of widespread sum-mer forest grazing and farming across central Scandinavia. Ourfindings demonstrate that early land use strongly affected pastorganic carbon dynamics and suggest that the influence of histor-ical landscape utilization on contemporary changes in lake-watercarbon levels has thus far been underestimated. We propose thatpast changes in land use are also a strong contributing factor inongoing organic carbon trends in other regions that underwentsimilar comprehensive changes due to early cultivation and graz-ing over centuries to millennia.

lake-water quality | carbon cycling | land use | Holocene | paleoecology

Over the past three decades, monitoring programs have re-corded a widespread increase of organic carbon (OC) con-

centrations in surface waters in parts of Europe and NorthAmerica (1–4). OC in lakes and rivers plays a major role in theglobal carbon cycle by transporting carbon from terrestrial tofreshwater and marine environments (5), determining drinkingwater quality and associated treatment costs (6), and affectingaquatic ecosystem functioning. In aquatic ecosystems, OC influ-ences energy mobilization, light conditions (7), water acidity (8),as well as the transport of metals and pollutants (9). The wide-spread occurrence of this increase in surface water OC, also re-ferred to as browning or brownification due to an associatedincrease in color, suggests that regional rather than local factorsare the drivers behind this phenomenon, but at present the un-derlying mechanisms are still controversial.Several hypotheses have been proposed to explain this recent

OC increase, from climate change to changes in anthropogenicforcing such as declining atmospheric acid deposition or alterationsin landscape utilization. A number of climate-sensitive mecha-nisms have been suggested to cause an increase in OC export fromthe terrestrial to the aquatic environment; for example, increasedtemperatures enhance decomposition rates in organic-rich soils(10) and promote vegetation cover (11). Changes in the amountand timing of precipitation potentially lead to alterations of hy-drological flow paths and intensities, controlling OC leachingfrom soils (12–14). An increased frequency of severe droughtsfollowed by rewetting was also proposed to increase terrestrial

carbon loss by stimulating microbial and bacterial activities (15).Other studies suggest that declining rates of acid deposition maybe the key driver behind the observed OC increase in surfacewaters. Decreasing sulfur deposition rates allow catchment soilsto recover from acidification, where the resulting decrease in acidityand/or ionic strength of the soil solution may increase the solu-bility of soil organic matter (OM) and consequently OC leaching(4, 16, 17). Some studies have linked the OC rise to recent changesin land-use practices, such as the drainage and burning of peat-lands (18, 19), or to changes in land cover (20). Other proposedexplanations are the increase in atmospheric carbon dioxide levels(21) and the elevated atmospheric deposition of nitrogen (22, 23),both of which support increased plant productivity.All proposed hypotheses are based on the recent OC trends

observed in monitoring programs, which reflect the developmentin surface waters only over the past few decades. However, thelack of long-term monitoring data—over centuries or even mil-lennia—leaves us with an ambiguous understanding of the pasttrajectory of OC concentrations. For example, many of the cli-mate-related hypotheses imply some change in ecosystem andsoil functioning, which in turn leads to new and possibly un-precedented OC levels in surface waters. In contrast, if the de-cline in acid deposition were the main trigger for the recent OCdynamics, then the observed increase would largely represent arecovery process that returns surface-water OC to more natural,preindustrial concentrations. These two trajectories—unprecedentedincrease or recovery to past levels—imply substantially different

Significance

Monitoring programs have recorded increases in organic car-bon concentrations in northern lakes, which have importantimplications for water quality and ecosystem functioning.Current hypotheses interpret this trend in light of recent en-vironmental changes such as acidification and climate but donot include an examination of long-term changes and theircauses. We inferred past trends from sediment archives acrosscentral Sweden, allowing us to assess recent changes on amillennial scale. Our data demonstrate that a long-term de-cline beginning already in the 15th century preceded the re-cent organic carbon increase. This was a response to spatiallyextensive human–landscape interactions that included forestgrazing and mire exploitation, which were common acrossEurope and altered carbon cycling between terrestrial andaquatic ecosystems.

Author contributions: C.M.-J. and R.B. designed research; C.M.-J., J.T., C.B., H.Y., and R.B.performed research; C.M.-J., J.T., C.B., and H.Y. analyzed data; and C.M.-J. and R.B. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: carsten.meyerjacob@gmail.com.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1501505112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1501505112 PNAS | May 26, 2015 | vol. 112 | no. 21 | 6579–6584

ENVIRONMEN

TAL

SCIENCE

SSU

STAINABILITY

SCIENCE

Dow

nloa

ded

by g

uest

on

Oct

ober

1, 2

020

future scenarios: either an advancing increase in OC concentra-tions strongly coupled to anthropogenic forcing or a stabilizationat naturally higher, prehuman impact levels. Thus, it is importantto understand the long-term trajectory of past OC dynamics toidentify the key drivers behind the recent OC increase. A bettermechanistic understanding of this process will improve our abilityto predict future OC changes in surface waters as well as theirenvironmental implications. To assess the likelihood and likelyintensity of future OC changes, it is a prerequisite to establish thenatural, predisturbance baseline conditions and how surface-waterOC has evolved in response to climate change and superimposedhuman impacts over tens to possibly thousands of years.Lake-sediment studies have shown that contemporary eco-

system responses to environmental problems such as surface-water acidification, eutrophication, and climate change have beeninfluenced in part by human activities such as land use over sev-eral hundred to even several thousand years. Land use, includingcultivation as well as low-intensity grazing, has been demonstratedto have affected not only forest cover and ecology in much ofEurope—for example, upland woodland areas in Scotland (24)—but also lake trophic status in, for example, Ireland and Denmarkalready during the Neolithic and Bronze Ages (25, 26). In south-west Sweden, land use led in some areas to a change in lake-waterpH by causing an alkalization of acid-sensitive lakes during the IronAge and into the late 19th century (27) and possibly also changes inlake-water OC levels (28). We hypothesize therefore that the im-pact of early land use, which entailed an extensive, if not intensive,utilization of the landscape, on OC levels in surface waters hasbeen underestimated and has instead played an important role inlong-term OC dynamics.One of the important strengths of using sediments as an ar-

chive is the ability to put recent environmental changes into alonger term perspective. The few decades of overlap that nowexist between monitoring and sediment records allow us to makequantitative comparisons between recent trends measured inmonitoring programs and the long-term patterns preserved insediments. Here, we present reconstructions of past lake-watertotal OC (TOC) concentrations in four lakes across the centralSwedish boreal landscape. These reconstructions are based onvisible-near-infrared (VNIR) spectroscopic measurements ofsediment records from the study lakes. VNIR spectroscopy issensitive to changes in OM quality and is widely used for qualitycontrol in industrial processes, and also in environmental andecological studies to determine, for example, plant and animaltissue composition or different soil constituents (29, 30). Surfacewaters receive different inputs of autochthonous and allochthonousOM depending on climate and catchment characteristics suchas vegetation, hydrological flow patterns, or nutrient supply (31).These qualitative and quantitative differences in OM ultimatelyleave their fingerprint in the lake’s sediment and are consequently

reflected by the sediment’s VNIR spectrum. To infer past lake-water TOC, we applied a calibration model between spectra oflake surface sediments, the most recently accumulated material,and corresponding TOC concentrations in the water column (32,33) (Materials and Methods and Fig. S1).The four study lakes—Långsjön, Lång-Älgsjön, Fickeln, and

Gipsjön—are located in the spruce and pine-dominated borealforest of central Scandinavia (Fig. 1 and Table S1). This land-scape has been noticeably used by humans over many centuries,notably through a widespread transhumant system of summerforest grazing and farming [fäbod or säter in Swedish, sæter inNorwegian; similar to shielings in the United Kingdom orAlmwirtschaft (German) and alpage (French) in the Alps], sinceat latest the 15–16th centuries, with earlier traces in the Iron Age(34). In conjunction with natural processes, human landscapeutilization formed a diverse forest structure and vegetationcomposition. During the 20th century, this forest type has beenaltered by modern forestry, resulting in a younger, denser, and lessdiverse forest (35). All four study lakes are closely connected tomires, leading to humic waters with TOC concentrations above10 mg·L−1. Gipsjön and Långsjön are part of the freshwatermonitoring program of the Swedish Agency for Marine andWater Management, and both have exhibited increasing TOCconcentrations since the beginning of their monitoring in 1987(Fig. 2, Insets). In boreal Northern Europe, the dissolved OC(DOC) fraction (∼97%) greatly dominates the TOC pool (36, 37),and therefore, the TOC concentrations presented here can beconsidered as virtually equivalent to DOC.Our aim was to assess whether past OC dynamics are lake-

specific or if there are similarities across landscapes, which wouldsuggest a common driver behind these dynamics. TOC moni-toring data exist for two of our study lakes since the late 1980sand were used to validate the reconstructed TOC values. Wefurther conducted a multiproxy analysis of one of the studied

Fig. 1. Overview maps of Scandinavia (Left) and south-central Sweden withregistered historical summer forest farming and grazing sites (red circles)(Swedish National Heritage Board database) and locations of the study lakesLångsjön, Lång-Älgsjön, Fickeln, and Gipsjön (Right).

Fig. 2. Lake-water TOC reconstructions for short sediment cores from thestudy lakes Långsjön, Lång-Älgsjön, Fickeln, and Gipsjön over the past 1,500 y.Insets show the agreement between sediment-inferred (open circles) andobserved (blue) lake-water TOC concentrations for the two monitoring lakesLångsjön and Gipsjön.

6580 | www.pnas.org/cgi/doi/10.1073/pnas.1501505112 Meyer-Jacob et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

1, 2

020

lake’s entire sediment sequence extending ∼10,500 y back in timeto exemplify the long-term Holocene lake-water TOC develop-ment with the aim of elucidating potential driving factors behindthe reconstructed TOC changes. These analyses include organicand inorganic geochemistry as well as biological indicators(Materials and Methods).

Results and DiscussionLake-Water OC Dynamics During the Past. In the two monitoringlakes, Långsjön and Gipsjön, sediment-inferred lake-water TOCconcentrations agree well with the observed TOC levels and re-cent trends, corroborating the possibility to correctly reconstructpast lake-water TOC dynamics using sediment archives andVNIR spectroscopy (Fig. 2, Insets). Our lake-water TOC re-constructions show a consistent trajectory at the four sites overthe past 400 y indicating two distinct phases: (i) a long-term de-crease since at latest A.D. 1600, culminating in the lowest con-centrations during the early to mid-20th century, and (ii) a recentincrease of TOC concentrations starting circa (c.) A.D. 1940–1970 (Fig. 2). In the four lakes, TOC concentrations decreasedfrom values in the range of 12–18 mg·L−1 to 8–10 mg·L−1 in theearly/mid-20th century and have subsequently increased to levelsof 10–16 mg·L−1 inferred for the top surface sediment. In contrastto the other three sites, the TOC decrease in Gipsjön was lesspronounced and of lower magnitude. Before the onset of thesedynamics, stable TOC levels prevailed in three of the four lakesfor c. 1,000 y. For Långsjön, the recovered sediment sequence

covers only the past c. 400 y, which precludes reconstruction ofTOC dynamics before A.D. 1600.The lake-water TOC reconstruction for Lång-Älgsjön over the

past 10,500 y since deglaciation shows that TOC levels rose from3 to 17 mg·L−1 within the first c. 1,000 y after lake formation.After this sharp increase, TOC concentrations remained rela-tively stable around ∼19 mg·L−1 for almost 9,000 y (Fig. 3A),before the onset of the distinct decline from c. A.D. 1450. Duringmost of the Holocene, TOC levels fluctuated between 17 and22 mg·L−1.

Driving Factors Behind Past Lake-Water OC Dynamics.Early lake ontogeny (8400–7400 B.C.) and natural long-term stability (7400B.C.–A.D. 1450). Rapid changes in the landscape following de-glaciation of central Sweden around 8500 ± 500 B.C. (SwedishGeological Survey database, maps2.sgu.se/kartgenerator/maporder_sv.html) drove the initial lake-water TOC dynamics at Lång-Älgsjön. Within the first 1,000 y after lake formation, lake-waterTOC concentrations rapidly increased, reflecting catchment sta-bilization through development of soils and vegetation. Counts ofleaf-like appendages of Sphagnum in the sediment record indicatethat the surrounding peatlands developed already during the firstcenturies after lake formation (Fig. S2C). The build-up of soilOM, including peat, led to an increased export of organic sub-stances from the catchment to the lake, as indicated by the si-multaneous increase in sediment TOC (from 0.9% to 15%) andtotal lignin (from 0.1 × 106 to 1.0 × 106; Fig. 3 B and C), whichoriginates predominately from terrestrial vascular plants. In

Fig. 3. Comparison of (A) the VNIR-inferred lake-water TOC development to geochemical and biological proxy data analyzed in the sediment record of Lång-Älgsjön for the past 10,500 y; (B) sediment TOC; (C) total lignin, indicating changes in the input of terrestrial-derived OM; (D) bSi; (E) K/Al ratio, indicatingchanges in the weathering degree of the sediment and consequently changes in catchment erosion; (F) diatom-inferred pH reconstruction; (G) groupeddiatom assemblage composition; and pollen counts of (H) apophytes (pollen from native plants favored by human disturbance) and (I) anthropochores (pollenfrom cultivated plants).

Meyer-Jacob et al. PNAS | May 26, 2015 | vol. 112 | no. 21 | 6581

ENVIRONMEN

TAL

SCIENCE

SSU

STAINABILITY

SCIENCE

Dow

nloa

ded

by g

uest

on

Oct

ober

1, 2

020

addition to the enhanced allochthonous input of OM, anincreased in-lake production likely contributed to elevatedlake-water TOC levels and higher carbon sequestration rates.Enhanced in-lake production is indicated by increasing concen-trations of biogenic silica (bSi; derived mainly from diatoms)in the sediment (from 0.4% to 19%; Fig. 3D). Rapidly decliningK/Al ratios from 0.6 to 0.2 further illustrate the catchment sta-bilization (Fig. 3E). The K/Al ratio provides information aboutthe weathering degree of the sediment because labile K is pre-ferentially leached from aluminosilicates compared with in-soluble Al during chemical weathering (38). As a consequenceof declining erosion rates, chemical weathering in soils had astronger relative imprint on the minerogenic matter deposited atthe lake bottom. The diatom-based pH reconstruction shows thatthe lake progressively acidified after formation (Fig. 3F), which isconsistent with an increase in lake-water TOC and decline inbase cation export from the catchment (39).After the initial catchment development and stabilization, lake-

water TOC concentrations in Lång-Älgsjön reached a stablestate of natural variability, which prevailed for the following 8,850 y.During most of this period—that is, until A.D. 200—none of theanalyzed biological and geochemical proxies show substantialvariation (Fig. 3). From c. A.D. 200, sediment TOC increased toconcentrations above 35% and bSi dropped to ∼14%, indicating astrong decline in diatom production that also coincided with adiatom community shift from benthic to tychoplanktonic speciesdominance (Fig. 3 B, D, and G and Fig. S3). These changes weremost likely a result of an expansion of the mires bordering Lång-Älgsjön. The lake’s present catchment has extensive areas ofwetland, and floating Sphagnum mats cover most of the shore-line. An advancing peatland development with Sphagnum matsgrowing farther into the lake would have reduced the habitat forbenthic production, which had dominated the diatom assem-blage until c. A.D. 200. We suggest that because peatlands werealready a dominant control on lake-water TOC, these naturalshoreline dynamics did not have a measureable effect on thelake-water TOC quantity and quality in Lång-Älgsjön.Anthropogenic disturbance (A.D. 1450–1950). During the last 550 y,lake-water TOC dynamics in Lång-Älgsjön appear to be stronglycoupled to dynamics in human landscape utilization. Coincidingwith the lake-water TOC decline from 18 to 8 mg·L−1 starting c.A.D. 1450, pollen counts show both an increase in apophytesfrom 2.6% to 6.6% and a continuous occurrence of anthro-pochores (cultivated plants; e.g., cereals) that increase from 0.1%to 0.8% (Fig. 3H and I). Apophytes are native plants indicative ofa more open landscape, which are favored by human landscapedisturbances such as forest grazing and cultivation. There is evi-dence of limited land use already during c. 500 B.C.–A.D. 1450,such as smaller increases in apophyte pollen and isolated occur-rences of anthropochore pollen in Lång-Älgsjön, which are seenin other regional pollen records (40, 41). During this earlier pe-riod, there is also a slight decline, ∼3 mg·L−1, in inferred lake-water TOC relative to mid-Holocene values, but given the iso-lated occurrence for anthropogenic indicators and the possibletransport of pollen from outside the catchment, we do not in-terpret this decline as such. The pronounced decrease in lake-water TOC from A.D. 1450 is supported by changes in the di-atom community toward a dominance of benthic diatom species(Fig. 3G), indicating improved light conditions (7).Clear evidence of land use in the form of forest grazing and

cultivation exists for at least 1,000 y in central Sweden, withtraces further back into the Iron Age. This evidence occurs mainlyin the form of disturbance indicators, resulting from the gradualopening and alteration of the forest vegetation (40–43). How-ever, the region underwent more fundamental changes in agri-cultural practices after the medieval agrarian crisis. During the15–16th centuries, a more widespread system of summer forestgrazing and farming expanded, which was driven by changes in

the rural society and increasing demands for winter fodder (34).These seasonal settlements were established at some distancefrom permanent settlements in areas where resources were lessexploited and fodder not limited. Activities at these sites in-cluded the grazing of the livestock in the forest, production andprocessing of milk, haymaking on mires for fodder, and occa-sionally also cultivation of cereals (e.g., barley and rye) (40). Toimprove forest grazing, the canopy cover was opened and un-wanted field-layer vegetation such as dwarf shrubs and mosseswas grazed, removed, or burned, leading to an altered, moreopen forest composition (42).The increase in apophytes and anthropochores in the sedi-

ments of Lång-Älgsjön from A.D. 1450 reflects the increasingimpact of summer grazing in the area. Pollen studies in the re-gion show a similar timing for the expansion of this type oflandscape utilization around A.D. 1300–1500 (41, 42), and his-torical documents mention summer farms in the direct vicinity ofLång-Älgsjön for the first time around A.D. 1650 (Swedish Na-tional Heritage Board database, www.fmis.raa.se/cocoon/fornsok/search.html). The increased presence of humans is further sup-ported by a higher abundance of larger charcoal particles (>25 μm)in the sediment record (Fig. S4), reflecting more frequentlocal fires associated with selective forest clearance by slashand burn. In addition to the structural changes of the forest, thegrazing in the forest and on mires disturbed the ecosystemaround the summer settlements. We suggest that trampling bylivestock (sheep, goats, and cattle) physically disturbed vege-tation and soils in the catchment and exposed mineral soilhorizons. This physical disturbance contributed to an enhancedtransport of higher vascular plant debris and less intensivelyweathered minerogenic matter to the lake, which is reflected inthe sediment record by increases in total lignin and K/Al ratios,respectively. Simultaneously with the decrease in lake-waterTOC, total lignin increased from ∼2.7 × 106 to ∼3.6 × 106 andthe K/Al ratios from <0.1 to ∼0.2 (Fig. 3 C and E).Three important effects of summer grazing and farming on

OC dynamics in the boreal forest are the following: First, openingof the forest canopy, haymaking on mires, and grazing reduce theaboveground biomass and litter production (42, 44). This conse-quently decreases the labile soil OC pool, which codeterminesOC concentrations in surface waters (45). Second, the regularremoval of biomass lowers the depth to the water table on mires(42, 46), reducing the possible OC generation in the soil solutionupon rehydration (47). Third, the exposure of fresh, less weath-ered mineral-soil horizons, associated with soil disturbance by thelivestock, facilitates an enhanced adsorption of DOC to mineralsurfaces (48). Together, these processes would lead to a de-creased export of OC from the terrestrial to the aquatic systemand thus to a decline in OC concentrations in surface waters. Thefew existing studies aimed at estimating the impact of moderngrazing on OC dynamics show no significant changes in the soilsolution or discharge due to grazing (44, 46, 49). However, thereare no analog studies that reflect the changes associated with thecomprehensive shift from a natural, undisturbed to a low-intensive,human-altered boreal landscape. Existing studies investigatedonly recent decennial effects of grazing in peatlands but not long-term effects over centuries in the boreal forest as a whole andwere conducted in areas that have previously been altered byhumans over millennia.Inventories of historical sites for summer forest farming and

grazing emphasize the spatially extensive significance of this low-impact type of landscape utilization (Fig. 1), which would explainthe common past OC trajectory across central boreal Sweden.Within a 3-km distance to Lång-Älgsjön, five summer enclosuresalone are documented, where each enclosure included several totens of summer cabins and sheds for livestock. Långsjön, wherethe lake-water TOC declines from 15 to 10 mg·L−1 during A.D.1600 to the mid-20th century, is similarly located in an area that has

6582 | www.pnas.org/cgi/doi/10.1073/pnas.1501505112 Meyer-Jacob et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

1, 2

020

an extensive registry of historical summer farms in the lake’s closeproximity. Even though no historical summer settlements are reg-istered near Gipsjön, local place names reflect past land use, andhistorical documents show that the lake is located along a pathwaylinking the town of Malung to three summer farms. Compared withthe other lakes, the lake-water TOC decrease in Gipsjön (2 mg·L−1)is somewhat lower. No data about historical summer settlementsexist for the county in which the third lake, Fickeln, is located.However, pollen counts of apophytes and cereals in the lake’ssediment indicate also that past OC dynamics in Fickeln, with adecline from 14 to 8 mg·L−1, were strongly coupled to forest grazingand also cereal cultivation (Fig. S5).Recovery (A.D. 1950 to present). Since the mid-20th century, lake-water TOC concentrations increased continuously in Lång-Älgsjön, as well as in the other study lakes. This increase in Lång-Älgsjön is accompanied by decreasing total lignin values, lessabundant pollen of apophytes and anthropochores, as well asfewer charcoal particles in the lake’s sediment, indicating re-duced catchment disturbance and human presence (Fig. 3 andFig. S4). In central boreal Sweden, summer farming and grazingstarted to become less important, decreasing gradually in the19th century and more rapidly from the 1890s due to newchanges in agricultural practice following the industrial revolu-tion (34). Historical documents describe the first abandonmentsof summer farms in the vicinity of Lång-Älgsjön at the end of the19th century, but some sites remained active until the 1940s(Swedish National Heritage Board database, www.fmis.raa.se/cocoon/fornsok/search.html). Together with the development ofmodern forestry, the cessation of forest farming and grazing alloweda forest regrowth that has led to a denser forest with an increasedstand volume (35). These changes in forest composition and struc-ture reversed the effects of summer farming and grazing on the OCdynamics in the boreal forest and may thus contribute to the recentincrease in OC concentrations in surface waters in the study area.

Recent Lake-Water OC Dynamics and Implications for Future Change.Our results indicate that changes in lake-water TOC in centralboreal Sweden are strongly coupled to historical patterns inlandscape utilization that included widespread but low-intensivesummer forest grazing as well as small-scale farming. The factthat not only does the reconstructed long-term TOC decreasecoincide with the expansion of this type of land use but also thatthe recent TOC increase coincides with its cessation stronglysuggests that the observed recent increase is to a large extent aresponse to changes in human landscape utilization. Based onour findings, we cannot isolate the additional and likely interactingroles of other hypothesized drivers for recent TOC dynamics. Aciddeposition during the 20th century seems to have not significantlyaffected Lång-Älgsjön. The diatom-based pH reconstruction re-veals that after the natural postglacial acidification of the lakefrom 6.5 to ∼4.9, the pH remained consistently acidic thereafter,with a negligible (0.1 pH units) decline c. A.D. 1950 (Fig. 3F). Noconsiderable pH changes occurred concurrently with the recon-structed lake-water TOC variations during the past centuries.An explicit separation between climate and land-use effects on

lake-water TOC is complicated by the temporal overlap betweenthe expansion of summer farms and climate cooling during theLittle Ice Age (LIA), when temperatures were ∼1 °C cooler thanat present (50). The climate deterioration during the LIA mayhave contributed to the TOC decline during the period A.D.1450–1950; however, because of the limited variation in TOCduring the preceding 8,850 y despite other periods with signifi-cant changes in climate, we suggest that climate was not the maindriver. For example, no pronounced trends in lake-water TOCoccurred during the Holocene Thermal Maximum (HTM; 6000–2800 B.C.), when annual temperatures were 2–2.5 °C warmer inNorthern Europe than at present, nor during the millennia fol-lowing the HTM, when temperatures declined by >1 °C, or

during a cold phase c. 1500 B.C. of equal magnitude to the LIA(51) (SI Materials and Methods and Fig. S6). Regional climatereconstructions indicate that the scale of climate change duringthe LIA was not unprecedented (50), whereas the decline inlake-water TOC (∼10 mg·L−1) over the past 500 y had noprecedent following the early Holocene landscape development.Cooling during the LIA did play an indirect role by encour-

aging the expansion of summer forest grazing and farming.Historians note that farms in marginal growing areas switchedfrom cultivation to pasturage and that colder winters facilitatedthe transport of fodder from the summer forest farms back to thevillages to feed the increasing number of overwintering livestock(34). Nevertheless, the recent increase in lake-water TOC inGipsjön, which exceeds the TOC decline since A.D. 1600 andreaches TOC levels higher than what occurred during the preceding1,500 y (Fig. 2), indicates that land use, too, is not the only driver ofthe recent OC increase in central boreal Sweden.As shown for surface-water eutrophication and acidification in

Europe (25–27), our data demonstrate that early land use alsoplayed, and still plays, an important role in long-term OC dy-namics and contemporary changes in surface-water TOC. Basedon our findings in central Sweden and an earlier study of onelake in southwestern Sweden (28), we suggest that past changesin land use are also a contributing factor in ongoing OC trends inother regions that underwent similar comprehensive changes dueto long-term human landscape utilization. Even though the typeof early land use described here may be specific for forestedupland and mountainous areas in Europe, our findings exemplifythe potential magnitude of past human land-use activities on OCdynamics in surface waters. Globally, humans have extensivelyaltered the natural vegetation cover over centuries (NorthAmerica, following European settlement) to millennia (Europe,Asia) and with it the long-term terrestrial carbon storage (52). Inother regions with observed changes in OC levels, other driverssuch as the recovery from acidification in southern Scandinaviaand the United Kingdom (4) may dominate current OC dy-namics. However, also these regions have long land-use historiesthat may need to be considered when assessing recent changes inOC concentrations and determining appropriate reference con-ditions for water management.

Materials and MethodsWe recovered sediment cores that span several centuries to millennia fromLångsjön, Fickeln, and Gipsjön and the last ∼10,500 y in the case of Lång-Älgsjön. Corresponding age–depth models were constrained by radiocarbonmeasurements. We applied 210Pb-dating to improve the age control for thenear-surface sediment from Långsjön and Gipsjön, which allowed a com-parison between our sediment-inferred trends and lake-water monitoringdata. For Lång-Älgsjön, we used spheroidal carbonaceous fly ash particles(SCPs) and a correlation between significant changes in Pb isotope signa-tures and established historical patterns of Pb pollution from varve-datedsediment records to further constrain the age–depth model (SI Materials andMethods, Table S2, and Figs. S2 and S7). Lake-water TOC reconstructions arebased on a partial least squares regression (PLSR) model between VNIRspectra of surface sediment samples and corresponding TOC concentrationsin the water column. We updated a previously published model (32, 33),which consists of surface sediments from 140 lakes distributed throughoutSweden, including nemoral, boreal, and subarctic sites. The model covers alake-water TOC gradient from 0.7 to 22 mg·L−1, comprises 7 PLSR compo-nents, and exhibits a cross-validated R2 of 0.65 and a root mean square errorof cross-validation (RMSECV) of 3.0 mg·L−1.

We used standard methods in paleolimnology and soil science in themultiproxy study of the sediment record from Lång-Älgsjön. Organic andinorganic geochemical sediment properties were determined by means ofelemental analysis for sediment TOC, pyrolysis-gas chromatography/massspectrometry for total lignin (53), Fourier transform infrared spectroscopyfor bSi (54), and wavelength dispersive X-ray fluorescence spectroscopy fortotal Al and K (55). Pollen and diatom analyses followed standard pro-cedures, and diatom-inferred pH reconstructions are based on data from theSurface Water Acidification Program. More specific details about the

Meyer-Jacob et al. PNAS | May 26, 2015 | vol. 112 | no. 21 | 6583

ENVIRONMEN

TAL

SCIENCE

SSU

STAINABILITY

SCIENCE

Dow

nloa

ded

by g

uest

on

Oct

ober

1, 2

020

methods, analytical procedures, instrument setups, and relevant referencesare provided in SI Materials and Methods.

ACKNOWLEDGMENTS. We thank the anonymous reviewers for helpfulcomments and suggestions to clarify some sections of the original manuscript.

We also thank Ingemar Renberg, Anna Ek, Jan-Erik Wallin, and others in thepaleolimnology group who have contributed to the analyses over time.We thank the Swedish Research Council, the university-supported researchenvironment, The Environment’s Chemistry, and the “Svenska Sällskapetför Antropologi och Geografi” for financial support.

1. Driscoll CT, Driscoll KM, Roy KM, Mitchell MJ (2003) Chemical response of lakes in theAdirondack Region of New York to declines in acidic deposition. Environ Sci Technol37(10):2036–2042.

2. Worrall F, et al. (2004) Trends in dissolved organic carbon in UK rivers and lakes.Biogeochemistry 70(3):369–402.

3. Skjelkvåle BL, et al. (2005) Regional scale evidence for improvements in surface waterchemistry 1990-2001. Environ Pollut 137(1):165–176.

4. Monteith DT, et al. (2007) Dissolved organic carbon trends resulting from changes inatmospheric deposition chemistry. Nature 450(7169):537–540.

5. Cole JJ, et al. (2007) Plumbing the global carbon cycle: Integrating inland waters intothe terrestrial carbon budget. Ecosystems (N Y) 10(1):172–185.

6. Matilainen A, Vepsäläinen M, Sillanpää M (2010) Natural organic matter removal bycoagulation during drinking water treatment: A review. Adv Colloid Interface Sci159(2):189–197.

7. Ask J, et al. (2009) Terrestrial organic matter and light penetration: Effects on bac-terial and primary production in lakes. Limnol Oceanogr 54(6):2034–2040.

8. Driscoll CT, Fuller RD, Schecher WD (1989) The role of organic acids in the acidificationof surface waters in the Eastern U.S. Water Air Soil Pollut 43(1-2):21–40.

9. Macdonald RW, Harner T, Fyfe J (2005) Recent climate change in the Arctic and itsimpact on contaminant pathways and interpretation of temporal trend data. Sci TotalEnviron 342(1-3):5–86.

10. Freeman C, Evans CD, Monteith DT, Reynolds B, Fenner N (2001) Export of organiccarbon from peat soils. Nature 412(6849):785.

11. Ekström SM, et al. (2011) Effect of acid deposition on quantity and quality of dis-solved organic matter in soil-water. Environ Sci Technol 45(11):4733–4739.

12. Tranvik LJ, Jansson M (2002) Climate change—Terrestrial export of organic carbon.Nature 415(6874):861–862.

13. Hejzlar J, Dubrovský M, Buchtele J, R�uzicka M (2003) The apparent and potentialeffects of climate change on the inferred concentration of dissolved organic matter ina temperate stream (the Malse River, South Bohemia). Sci Total Environ 310(1-3):143–152.

14. Hongve D, Riise G, Kristiansen JF (2004) Increased colour and organic acid concen-trations in Norwegian forest lakes and drinking water—A result of increased pre-cipitation? Aquat Sci 66(2):231–238.

15. Fenner N, Freeman C (2011) Drought-induced carbon loss in peatlands. Nat Geosci4(12):895–900.

16. Evans CD, Monteith DT, Cooper DM (2005) Long-term increases in surface waterdissolved organic carbon: Observations, possible causes and environmental impacts.Environ Pollut 137(1):55–71.

17. Valinia S, Futter MN, Cosby BJ, Rosén P, Fölster J (2015) Simple models to estimatehistorical and recent changes of total organic carbon concentrations in lakes. EnvironSci Technol 49(1):386–394.

18. Armstrong A, et al. (2010) The impact of peatland drain-blocking on dissolved organiccarbon loss and discolouration of water; results from a national survey. J Hydrol(Amst) 381(1-2):112–120.

19. Yallop AR, Clutterbuck B, Thacker J (2010) Increases in humic dissolved organic carbonexport from upland peat catchments: The role of temperature, declining sulphurdeposition and changes in land management. Clim Res 45(1):43–56.

20. Tuvendal M, Elmqvist T (2011) Ecosystem services linking social and ecological sys-tems: River brownification and the response of downstream stakeholders. Ecol Soc16(4):21.

21. Freeman C, et al. (2004) Export of dissolved organic carbon from peatlands underelevated carbon dioxide levels. Nature 430(6996):195–198.

22. Findlay SEG (2005) Increased carbon transport in the Hudson River: Unexpectedconsequence of nitrogen deposition? Front Ecol Environ 3(3):133–137.

23. Rowe EC, et al. (2014) Predicting nitrogen and acidity effects on long-term dynamicsof dissolved organic matter. Environ Pollut 184:271–282.

24. Tipping R, et al. (2008) Prehistoric Pinus woodland dynamics in an upland landscape innorthern Scotland: The roles of climate change and human impact. Veg Hist Ar-chaeobot 17(3):251–267.

25. Bradshaw EG, Rasmussen P, Odgaard BV (2005) Mid- to late-Holocene land-usechange and lake development at Dallund So, Denmark: Synthesis of multiproxy data,linking land and lake. Holocene 15(8):1152–1162.

26. Taylor KJ, Potito AP, Beilman DW, Ghilardi B, O’Connell M (2013) Palaeolimnologicalimpacts of early prehistoric farming at Lough Dargan, County Sligo, Ireland.J Archaeol Sci 40(8):3212–3221.

27. Renberg I, Korsman T, Birks HJB (1993) Prehistoric increases in the pH of acid-sensitiveSwedish lakes caused by land-use changes. Nature 362(6423):824–827.

28. Rosén P, Bindler R, Korsman T, Mighall T, Bishop K (2011) The complementary powerof pH and lake-water organic carbon reconstructions for discerning the influences onsurface waters across decadal to millennial time scales. Biogeosciences 8(9):2717–2727.

29. Stenberg B, Rossel RAV, Mouazen AM, Wetterlind J (2010) Chapter five—Visible andnear infrared spectroscopy in soil science. Advances in Agronomy, ed Sparks DL(Elsevier Academic Press Inc., San Diego, CA), Vol 107, pp 163–215.

30. Foley WJ, et al. (1998) Ecological applications of near infrared reflectance spectros-copy—A tool for rapid, cost-effective prediction of the composition of plant andanimal tissues and aspects of animal performance. Oecologia 116(3):293–305.

31. Findlay S, Sinsabaugh RL (2003) Aquatic Ecosystems. Interactivity of Dissolved OrganicMatter (Elsevier Academic, Burlington, MA).

32. Rosén P (2005) Total organic carbon (TOC) of lake water during the Holocene inferredfrom lake sediments and near-infrared spectroscopy (NIRS) in eight lakes fromnorthern Sweden. Biogeochemistry 76(3):503–516.

33. Cunningham L, Bishop K, Mettävainio E, Rosén P (2011) Paleoecological evidence ofmajor declines in total organic carbon concentrations since the nineteenth century infour nemoboreal lakes. J Paleolimnol 45(4):507–518.

34. Larsson J (2012) The expansion and decline of a transhumance system in Sweden,1550-1920. Hist Agrar 56:11–39.

35. Ericsson S, Östlund L, Axelsson AL (2000) A forest of grazing and logging: De-forestation and reforestation history of a boreal landscape in central Sweden. NewFor 19(3):227–240.

36. Kortelainen P, et al. (2006) Controls on the export of C, N, P and Fe from undisturbedboreal catchments, Finland. Aquat Sci 68(4):453–468.

37. von Wachenfeldt E, Tranvik LJ (2008) Sedimentation in boreal lakes—The role offlocculation of allochthonous dissolved organic matter in the water column. Ecosys-tems (N Y) 11(5):803–814.

38. Roy PD, Caballero M, Lozano R, Smykatz-Klossd W (2008) Geochemistry of late qua-ternary sediments from Tecocomulco lake, central Mexico: Implication to chemicalweathering and provenance. Chem Erde 68(4):383–393.

39. Engstrom DR, Fritz SC, Almendinger JE, Juggins S (2000) Chemical and biologicaltrends during lake evolution in recently deglaciated terrain. Nature 408(6809):161–166.

40. Segerström U, Hörnberg G, Bradshaw R (1996) The 9000-year history of vegetationdevelopment and disturbance patterns of a swamp-forest in Dalarna, northernSweden. Holocene 6(1):37–48.

41. Giesecke T (2005) Holocene dynamics of the southern boreal forest in Sweden. Ho-locene 15(6):858–872.

42. Segerström U, Emanuelsson M (2002) Extensive forest grazing and hay-making onmires—Vegetation changes in south-central Sweden due to land use since Medievaltimes. Veg Hist Archaeobot 11(3):181–190.

43. Karlsson H, Emanuelsson M, Segerström U (2010) The history of a farm-shieling sys-tem in the central Swedish forest region. Veg Hist Archaeobot 19(2):103–119.

44. Ward SE, Bardgett RD, McNamara NP, Adamson JK, Ostle NJ (2007) Long-term con-sequences of grazing and burning on northern peatland carbon dynamics. Ecosystems(N Y) 10(7):1069–1083.

45. Aitkenhead JA, Hope D, Billett MF (1999) The relationship between dissolved organiccarbon in stream water and soil organic carbon pools at different spatial scales. Hy-drol Processes 13(8):1289–1302.

46. Worrall F, Armstrong A, Adamson JK (2007) The effects of burning and sheep-grazingon water table depth and soil water quality in a upland peat. J Hydrol (Amst) 339(1-2):1–14.

47. Fenner N, Freeman C, Worrall F (2009) Hydrological Controls on Dissolved OrganicCarbon Production and Release From UK Peatlands. Carbon Cycling in NorthernPeatlands, Geophysical Monograph Series, eds Baird AJ, et al. (American GeophysicalUnion, Washington, DC), Vol 184, pp 237–249.

48. Kalbitz K, Solinger S, Park JH, Michalzik B, Matzner E (2000) Controls on the dynamicsof dissolved organic matter in soils: A review. Soil Sci 165(4):277–304.

49. Clay GD, Worrall F, Fraser EDG (2009) Effects of managed burning upon dissolvedorganic carbon (DOC) in soil water and runoff water following a managed burn of aUK blanket bog. J Hydrol (Amst) 367(1-2):41–51.

50. Matskovsky VV, Helama S (2014) Testing long-term temperature reconstruction basedon maximum density chronologies obtained by reanalysis of tree-ring data sets fromnorthernmost Sweden and Finland. Clim Past 10(4):1473–1487.

51. Seppä H, Bjune AE, Telford RJ, Birks HJB, Veski S (2009) Last nine-thousand years oftemperature variability in Northern Europe. Clim Past 5(3):523–535.

52. Kaplan JO, et al. (2010) Holocene carbon emissions as a result of anthropogenic landcover change. Holocene 21(5):775–791.

53. Tolu J, Gerber L, Boily J-F, Bindler R (2015) High-throughput characterization ofsediment organic matter by pyrolysis-gas chromatography/mass spectrometry andmultivariate curve resolution: A promising analytical tool in (paleo)limnology. AnalChim Acta, 10.1016/j.aca.2015.03.043.

54. Meyer-Jacob C, et al. (2014) Independent measurement of biogenic silica in sedimentsby FTIR spectroscopy and PLS regression. J Paleolimnol 52(3):245–255.

55. Rydberg J (2014) Wavelength dispersive X-ray fluorescence spectroscopy as a fast,non-destructive and cost-effective analytical method for determining the geo-chemical composition of small loose-powder sediment samples. J Paleolimnol 52(3):265–276.

6584 | www.pnas.org/cgi/doi/10.1073/pnas.1501505112 Meyer-Jacob et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

1, 2

020