Introduction

Lake-sediment records are critical archives of palaeoenvironmental information in the neotropics. The records preserve palaeoecological and palaeoclimatic information on disturbance and landscape development through changes in a lake’s material flux and hydrology. The analysis of these preserved materials leads to insights into past vegetation assemblages, fire activity, and human impacts. Palaeoenvironmental data from lake sediment provide insight into the temporal and spatial record of vegetation assemblages and fire frequency and complement and extend archaeological records (Cohen 2003; Horn 2007; Roberts 2014; Horn et al. 2018; Johanson et al. 2019, 2020). Within Costa Rica, proxy records from lake sediments have shown changes in vegetation, fire, and climate during the late Holocene linked to population decline, Spanish Conquest, and climate variability, including multidecadal droughts (Clement and Horn 2001; Anchukaitis and Horn 2005; Horn 2006; Taylor et al. 2013a; Wu et al. 2017; Johanson et al. 2019; Kerr et al. 2020; Yanuskiewicz et al. 2022). In this work we use pollen grains and microscopic charcoal fragments from a lake in southern Pacific Costa Rica to create vegetation and fire records spanning the past 400 years.

Pollen grains serve as a proxy for change in a landscape’s vegetation, and provide a means for assessing the presence and scale of human activity around a lake (e.g. Bush et al. 1992; Bennett and Willis 2001). Maize (Zea mays ssp. mays) is a key indicator for prehistoric agriculture in this region and was domesticated from Balsas teosinte (Zea mays ssp. parviglumis) in Mexico’s Central Balsas River Valley in ca. 9,500–8,800 cal bp (Piperno et al. 2009). Pollen grains of maize in Costa Rican lake sediments range from 63 to 90 μm in diameter, based on measurements of pollen residues mounted in silicone oil (Horn 2006). The large size of the grains makes it possible to scan for them at low power and use them as a marker for past agriculture even when rare in pollen counts. However, Taylor et al. (2013b) found that maize pollen is not necessarily a good indicator of the scale of maize cultivation. Maize pollen is typically only found in lacustrine records where agriculture occurs close to a lake shore, as the large and heavy pollen grains have been documented to travel less than 60 m from the parent plant (Raynor et al. 1972; Jarosz et al. 2005; Horn 2006). This is both a limitation and an asset in the interpretation as the proxy indicates highly local agriculture and human activity.

The analysis of preserved charcoal fragments provides insight into burning patterns and fire activity, contributes to interpretations of past agricultural practices, and assists in understanding drought extent, temporally and spatially (e.g. Whitlock and Larson 2001; Power et al. 2008; Nevle et al. 2011). Macroscopic charcoal typically reflects fire activity local to a lake, however, microscopic charcoal has been found to travel an extended distance through the atmosphere (Clark 1988; Carcaillet et al. 2002; Peters and Higuera 2007). Microscopic charcoal can be transported from fires as far as 20 to 100 km away from the lakeshore (Clark 1988; Peters and Higuera 2007; Duffin et al. 2008; Nelson et al. 2012). Consequently, microscopic charcoal records are understood to represent extra-local, as well as local, fire activity. Macroscopic sized charcoal is typically found deposited by fires within 5 km of a lake (Peters and Higuera 2007; Duffin et al. 2008; Nelson et al. 2012). It travels shorter distances due to its larger size, as larger fragments are not suspended by normal surface-wind speeds, and thus not commonly moved long distances by wind (Clark 1988; League and Horn 2000). Tandem analysis of microscopic and macroscopic charcoal records recovered from the same core has been used to help differentiate extra-local from local fire activity (Colombaroli et al. 2008; Duffin et al. 2008; Nelson et al. 2012).

Climate and interannual precipitation dynamics are influenced by multiple large-scale forces within Central America (Yanuskiewicz et al. 2022). Interplay of these drivers resulted in eras of warmer and cooler climates and affected vegetation assemblages, fire regimes, and moisture availability in the environment. The Little Ice Age (LIA, 550−100 cal bp/1400–1850 ce) climate event impacted Central America as a whole for a period lasting almost 500 years (Haug et al. 2001). Within the Central American Isthmus, the El Niño-Southern Oscillation (ENSO) is one of the most significant controls on precipitation and is responsible for periods of environmental stress on an inter-annual scale (Krishnaswamy et al. 2001). Millennial-scale movements of the Intertropical Convergence Zone (ITCZ) affect variation in regional precipitation on a sub-annual and inter-annual level (Portig 1965; Krishnaswamy et al. 2001).

Within Central America, the LIA generally manifested cooler and more arid conditions (Lane et al. 2011; Bhattacharya and Chiang 2014; Wu et al. 2017). While the LIA dried and cooled the neotropics and much of the Northern Hemisphere as a whole, regional small-scale response shows variation. Evidence of drought during the LIA was found in studies by Haug et al. 2001; Johanson et al. 2019, 2020; Lane et al. 2011; Sachs et al. 2009; Taylor et al. 2020; Wu et al. 2017 and Yanuskiewicz et al. 2022. Regional drying has been identified during the LIA at multiple lakes in southern Pacific Costa Rica, but with variations on the timing and duration. At Laguna Zoncho, Wu et al. (2017) identified a period of cool conditions lasting from 370 to 90 cal bp through the analysis of sub-fossil chironomids, and the n-alkane δ2 H data from Taylor et al. (2020) identified that vegetation around Laguna Zoncho experienced moisture stress. Compound specific stable isotope analyses at Laguna Los Mangos by Yanuskiewicz et al. (2022) found that precipitation varied over the length of the LIA. It is likely that the LIA had an uneven spatial impact on precipitation in Central America, based on site locations relative to oceanic and atmospheric drivers and to variables such as terrain and microclimates.

The El Niño Southern Oscillation strongly affects southern Pacific Costa Rica. There is evidence of additional environmental stress present in the region throughout the LIA from severe ENSO events in the Pacific (Haberle 2000). In El Niño years, the northeast trade winds increase in strength and velocity in the Caribbean which leads to increased rainfall on Costa Rica’s Caribbean slope and a subsequent rain shadow effect on the Pacific slope (Vargas and Trejos 1994).

Additionally, variations in the seasonal positions of ITCZ have a strong influence on regional climate variation in the neotropics (Haug et al. 2001). Haug et al. (2001) published a key record of variation in ITCZ dynamics inferred from titanium and iron concentrations in laminated marine sediments from the Cariaco Basin, offshore of Venezuela. These indicators were related to rainfall in northern South America linked to shifts in the mean latitudinal position of the ITCZ. Due to its influence in the region, the effects of this climate driver should be considered when examining signals of climate patterns in central and southern Costa Rica, including the LIA. Conroy et al. (2008) found that during the LIA, the Walker circulation weakened which led to a southern displacement of the ITCZ and brought cooler air and lower sea surface temperatures with it, in turn weakening the monsoon in Central America during that time.

Shortly after the onset of the LIA, the Spanish Conquest began in Costa Rica in 1502 ce with the arrival of Cristopher Columbus in present day Limón (Fernández Guardia 1913; Nunley 1960). Conquest extended throughout the 16th century but reached its apogee in 1561 when the Spanish reached the Central Valley region of what is today Costa Rica (Fernández Guardia 1913). This contact spread conflict, cultural influence, and disease throughout the country. The loss of life in the New World as a whole is estimated to be 90–95% of the population (Dobyns 1966; Lovell and Lutz 1995). This drastic decline in population would have affected agriculture and land clearance within established communities. A sudden, large-scale cessation of land clearing and burning, followed by natural regeneration of forests on abandoned fields, could have influenced global carbon cycles and decreased carbon dioxide levels, as proposed by Dull et al. 2010. The population collapse of the Americas, known as the Great Dying of the Indigenous Peoples of the Americas (Wolf 1982; Koch et al. 2019), is regarded by some researchers as a possible driving force for the LIA (Dull et al. 2010; Koch et al. 2019). Alternatively, Mann et al. 2009 argued that the LIA was solely the result of natural climatic variation, while Power et al. 2013 interpreted the decline in fire frequency in the neotropics around the time of Contact as driven by the climate changes and not by changes in population.

Palaeoenvironmental analyses at some sites in Costa Rica have shown evidence of a decline in agriculture and fire followed by forest regeneration near the time of the Conquest (Horn 2006; Horn and Haberyan 2016). These shifts have been interpreted to indicate population decline or near-complete site abandonment related to mortality from disease or direct conflict, or to the movement of populations to new sites (Horn 2006; Horn and Haberyan 2016). However, other studies have identified declines in agriculture and population that began before European contact (Anchukaitis and Horn 2005; Taylor et al. 2013a; Kerr et al. 2020), driven by drought.

In this study, we present the results of pollen and charcoal analyses of sediments recovered from Laguna Carse, Costa Rica, to further examine evidence of population decline, landscape modification, and anthropogenic land use as affected by the LIA and Spanish Conquest. We focus on (1) reconstructing vegetation and fire history in the Laguna Carse area over the past 400 years; (2) detecting periods of agricultural activity; (3) comparing fire trends evident in the microscopic charcoal record with previous macroscopic charcoal data; and (4) comparing our records with historical activity in the area and with climate shifts.

Study area

Environmental setting

Laguna Carse (9.0669°, -83.3472°) is a small (1.5 ha), moderately shallow (5 m) lake located at 365 m elevation in the southern Pacific lowlands of Costa Rica (Fig. 1). The lake is located in the Boruca District of the canton of Buenos Aires, where the El General Valley meets the lower slopes of the Fila Costeña, in an area of Upper Miocene and Pliocene plutonic rocks and breccias (Alvarado et al. 2009a; Horn et al. 2018). Maps in the online climate atlas of Costa Rica (Instituto Meteorológico Nacional de Costa Rica 2023) indicate that the lake receives 2,000–3,000 mm precipitation yearly, mainly between May and November. Landslides are common in the area (Alvarado et al. 2009a, b), and the formation of the basin that the lake occupies is attributed to landslides (Horn et al. 2018). The lake is classified in the tropical moist forest life zone in the Holdridge (1967) system (Tosi 1969); modern vegetation consists mainly of pastures and agricultural fields, with some remnant or secondary forest.

Fig. 1
figure 1

Location of Laguna Carse and other lake sites sampled for research within southern Pacific Costa Rica. Bathymetric map of Laguna Carse in lower left corner (adapted from Johanson et al. 2020)

Full size image

Cultural setting

Laguna Carse is situated within the Diquís subregion of the Gran Chiriquí archaeological region, which covers southern Pacific Costa Rica and western Panama. The archaeology of the Diquís subregion is divided into several major cultural periods that document changes from early, transitional periods to a complex hierarchical society (Dobyns 1966). Evidence of maize agriculture in the Diquís subregion dates as far back as ca. 3,360 cal bp, based on pollen analysis at the site of Laguna Los Mangos, ca. 13 km W of Laguna Carse (Johanson et al. 2019). The basal date of 350 cal bp on the Laguna Carse core indicates that the lake formed close to the end of the Chiriquí cultural period, which began ca. 1,150 cal bp (800 ce) and persisted until Spanish Conquest. The Chiriquí period was marked by high population levels and a complex hierarchical society supported by maize agriculture (Corrales 2000). Historic records from 1562 to 1563 ce from the Diquís region note that inhabitants relied on the large-scale agriculture of maize, beans, and fruit at the time (Fernández Guardia 1913).

In 253 cal bp (1697 ce), the Spanish relocated members of the Térraba indigenous community from settlements on the Caribbean slope of the Cordillera de Talamanca to the Diquís Valley near Laguna Carse (Pittier 1941; Stone 1962). Currently, the lake is within the modern Térraba-Boruca Indigenous Reserve and close to Bijagual, Ceibon, and Doboncragua. More than 3,000 indigenous people reside in the Puntarenas Province (OIRSA 2005). Six pre-Columbian archaeological sites dating pre-1500 ce (450 cal bp) are documented near Laguna Carse (Horn et al. 2018).

Previous work at Laguna Carse

Laguna Carse was cored in 2014 (Johanson et al. 2020), at its deepest point (water depth 5 m), as determined by bathymetric survey. A Colinvaux-Vohnout locking piston corer (Colinvaux et al. 1999) was used from an anchored floating platform to extract consecutive, meter-long sediment core sections. The first core drive began 51 cm below the mud-water interface and recovery continued until they reached an impenetrable rocky layer interpreted to be a landslide deposit. Initial limnological observations at the time of coring showed that Laguna Carse was circum-neutral in pH, not stratified, and had a surface temperature of 31 °C.

The Carse core sections were transported in their original aluminium coring tubes to the Laboratory of Palaeoenvironmental Research at the University of Tennessee, where they were stored in a cold room (4–5 °C). Core sections were opened and photographed, and sediment texture, stratigraphy, and colour were recorded. The sediment profile is relatively consistent and most of the core consists of silty sediment with some intervals showing fine laminations. The base of the profile consists of white (Munsell 2.5 YR 8/1), dense, fine mineral sediment (Johanson et al. 2020; Fig. 2). Johanson et al. 2020 developed a chronology for the record based on five AMS dates on plant macrofossils and an identified historic horizon in the sediments using the Bacon age-modeling package for R (Blaauw and Christen 2011). We used the same chronology in our study. The 714 cm core has a modelled basal date of 350 cal bp (1600 ce). The record spans the second half of the LIA through the present, in a post-Columbian context. The anomalously high sedimentation rate (1.8 cm/yr) found by Johanson et al. (2020) at Laguna Carse may have been caused by infilling from repeated landslides in the lake watershed and by post-Conquest forest clearance and agricultural activity. With its high rate of sediment accumulation, the Carse core has a higher temporal resolution than cores from other lakes in the southern Pacific region of Costa Rica (Clement and Horn 2001; Anchukaitis and Horn 2005; Taylor et al. 2015; Johanson et al. 2020). Johanson et al. (2020) examined sediment composition, stable carbon and nitrogen isotope ratios (δ13COM; δ15NOM), and macroscopic charcoal in the Carse sediments. C:N ratios between 3.8 and 10.9 indicated that the lake received a combination of aquatic and terrestrial inputs over the course of its record. Stable carbon isotope ratios suggested intervals of maize agriculture (Lane et al. 2004), and the near continuous presence of macroscopic charcoal was attributed to the use of fire in both indigenous and modern agriculture. Johanson et al. (2020) interpreted peaks in the Carse charcoal record as indicators of increased fire activity driven by a drier climate and subsequent increased wildfire activity. The researchers noted an extended period of drought beginning at ca. 340 cal bp marked by a broad peak in macroscopic charcoal influx that was consistent with the inferred regional position of the ITCZ at that time (Haug et al. 2001).

Fig. 2
figure 2

Sediment core sections recovered from Laguna Carse in July 2014 by Johanson et al. (2020). At top, drive depths are displayed including water depth and the height of the coring platform. Numbered locations represent locations of material sampled for radiocarbon dating

Full size image

Materials and methods

Pollen analysis

We processed 23 samples from the Laguna Carse core for pollen and microscopic charcoal analysis. Sampling was done at regular 16 cm intervals with a sample volume of 0.5 cm3. Pollen processing procedure followed standard methods (KOH, HCl, HF, acetolysis, safranin stain) with Lycopodium spores (Stockmarr 1971) added as a control. We mounted the pollen residues on microscope slides using silicone oil. Slides were examined at 400x magnification and counted to a minimum of 250 pollen grains per level, excluding spores and indeterminate pollen. We calculated pollen percentages as percent of total pollen, excluding indeterminate type pollen. To determine the presence or absence of maize pollen (Zea mays ssp. mays) at each level, we scanned 5 slides from each level at low (100x) magnification.

We identified pollen grains to the lowest taxonomic level possible using published photographs and keys (Kapp et al. 2000; Bush and Weng 2007). Maize pollen was differentiated from other grass pollen by size (grain diameter > 63 μm, Horn 2006). Asteraceae pollen was separated by spine length as ‘high spine’ and ‘low spine’. Within the Euphorbiaceae, we separately identified the genera Alchornea, Cnidoscolus, Acalypha, and Sapium. From the clade of families including Cannabaceae, Moraceae, Ulmaceae, and Urticaceae (formerly in the order Urticales), we separated and specifically identified Ficus, Cecropia, Trema-type, and Celtis-type and grouped others by number of pores (diporate, triporate, and polyporate). We identified Ficus as oblong grains with pores at their ends, as other Ficus morphologies present similarly to other diporate Urticales (Clement and Horn 2001).

Microscopic charcoal analysis

We examined microscopic charcoal on the same slides used for pollen analysis. Sediment was sieved with a 125 μm mesh sieve during our processing procedure. Using point counting methods modified from Clark 1982, we quantified charcoal areas at 400x magnification. We counted only completely opaque, black, and angular charcoal fragments.

Results

Pollen

We identified 61 pollen taxa in the Carse sediments. The most abundant throughout the record were Ficus (10.5%), ‘low spine’ Asteraceae (7.2%), and Piper (7.3%). Zea mays ssp. mays pollen recovery was low; no grains were found at any level in regular counts, and additional scans only confirmed one grain at one level. Possible broken fragments of maize pollen were noted rarely, but could not be confirmed. Montane and premontane types (including Alnus and Weinmannia) accounted for 16.3% of the total pollen. Pollen preservation over the length of the core was generally good. Indeterminate pollen grains averaged 3.1% of total pollen overall. Pollen preservation was notably poorer from ca. 175 to 80 cal bp.

The oldest sediments in the core, from ca. 330 to 280 cal bp, are characterized by high percentages of tree and shrub pollen and low percentages of herbaceous pollen (Figs. 3 and 4). This dominance by woody taxa indicates a forested landscape that continued until ca. 280 cal bp, when Poaceae and other herbaceous taxa came to dominate the pollen assemblage. At ca. 225 cal bp, woody taxa increase through ca. 100 cal bp during a period of forest recovery and culminate in the record’s lowest levels of herbaceous pollen types that is roughly coincident with the end of the LIA. Herbaceous pollen increases from ca. 100 until ca. 50 cal bp and reaches high levels similar to those found in the earliest section of the record. At ca. 50 cal bp, herbaceous pollen decreases until ca. 25 cal bp (Fig. 4). Herbaceous taxa increase and stay high from ca. 25 cal bp through modern times.

Fig. 3
figure 3

Laguna Carse pollen diagram showing relative abundance of selected taxa as % of total pollen plotted against calibrated age in cal bp with depth shown. The end of the LIA (ca. 100 cal bp) is displayed as solid blue line and historical resettlement of the Térraba people as red dashed line (253 cal bp/1697 ce). Evidence of Zea mays ssp. mays pollen is shown as an orange bar where a grain was found

Full size image
Fig. 4
figure 4

Laguna Carse microscopic charcoal area influx and macroscopic charcoal particle influx from Johanson et al. (2020) with pollen composition plotted against calibrated age in cal bp with depth shown. The blue line displays the end of the LIA (ca. 100 cal bp). Historical resettlement of the Térraba people is shown as red dashed line (253 cal bp/1697 ce) and evidence of Zea mays ssp. mays pollen as yellow dashed line. Herbaceous pollen types (light grey) include Amaranthaceae, Asteraceae, Cyperaceae and Poaceae. Arboreal pollen types (dark gray) include Alchornea (Euphorbiaceae), Anacardiaceae, Burseraceae, Cecropia, Ficus, Moraceae (Maclura and Helicostylis), Piper, Weinmannia and Quercus. Grouped pollen by type excludes some taxa due to the ambiguity of the clade and are plotted as % of total pollen at a given level

Full size image

Microscopic charcoal

The highest influx of microscopic charcoal occurs in the deepest section of the core at ca. 325 cal bp (Fig. 4). Influx decreases from this peak until ca. 250 cal bp, where the second lowest influx value in the core is reached. Following this short period of lower fire activity, influx values increase and consistently remain above the level at ca. 250 cal bp through the end of the LIA (ca. 100 cal bp). During this period from ca. 250 through ca. 100 cal bp, influx levels fluctuate, with two increases occurring at ca. 225 and ca. 140 cal bp. A marked decrease in microscopic charcoal influx begins at ca. 145 and continues until ca. 105 cal bp, when the lowest influx value in the core is recorded. Following that low, an abrupt increase in influx occurs at ca. 90 cal bp. Levels decrease through ca. 50 cal bp and then remain consistent through modern times, with a recent increase in charcoal influx at the top of the core, ca. -30 cal bp (Fig. 4).

Discussion

The relatively high-resolution palaeoenvironmental reconstructions from Laguna Carse provide insight into vegetation changes, fire activity, and timing of human settlement patterns after Spanish Conquest in the southern Pacific region of Costa Rica. The Laguna Carse pollen record indicates a history of lowland forest cover. The pollen record primarily reflects taxa growing within a few km of the lake, as Bush et al. 2000 found the downslope movement of high-elevation pollen types to be insignificant in research on modern pollen in Costa Rica.

Lake formation and initial exploitation

The Laguna Carse sediment core is the youngest of the cores collected in the western portion of the Diquís subregion. The sediment record spans approximately 400 years over 7.14 m, resulting in a high-resolution record compared to other records in the region. The archaeological evidence of human activity in the region predates the formation of Laguna Carse (Stone 1977). Maize pollen has been found within the Diquís Subregion at Laguna Los Mangos as early as ca. 3,360 cal bp (Johanson et al. 2019). Based on the radiocarbon-derived age model, the Carse sediments studied in this work reflect a landscape already altered by human impacts.

In the southern Pacific region of Costa Rica, proxy evidence of anthropogenic activity is commonly found in lake sediments almost immediately following the formation of a lake (Clement and Horn 2001; Anchukaitis and Horn 2005; Taylor et al. 2013a, b; Johanson et al. 2019; Kerr et al. 2020). Laguna Carse would likely have become a useful and exploitable resource for people inhabiting nearby settlements shortly after its formation. However, our records from Carse do not support this inference. Instead, from evidence of vegetation change and burning patterns, we infer a delay of ca. 100 years following lake formation before the earliest evidence of human use occurs in our records. Following lake formation, the pollen shows a forest-dominated system that changes after ca. 100 years when herbaceous taxa increase. The δ13COM and macroscopic charcoal analyses produced by Johanson et al. 2020 indicated that the area surrounding the lake was sparsely populated until ca. 250−225 cal bp. The pollen and previous analyses indicate an unmanaged system surrounding the lake with initial agriculture occurring after ca. 100 years (ca. 250−225 cal bp).

Impacts of Spanish Conquest within the region began before lake formation, and the extreme population decline from the resulting spread of conflict and disease throughout the area may be a driving factor in the inferred delay (Dobyns 1966; Wolf 1982; Lovell and Lutz 1995; Koch et al. 2019). The drop in regional population levels at this magnitude likely impacted agricultural output and spurred abandonment or restructuring of established local communities. Lacustrine sediment records from nearby Laguna Santa Elena showed human habitation near the lake prior to the Spanish Conquest and evidence of a temporary abandonment of the site coincident with the timing of Spanish arrival into the area and the onset of the Great Dying (Anchukaitis and Horn 2005; Kerr et al. 2020); however, the records also indicated an interval of site abandonment prior to Conquest. Although archaeological evidence in the region predates both lake formation and Spanish contact, due to these factors, it is likely that the surrounding area was sparsely populated at the time Laguna Carse formed and for this reason, the lake was not immediately exploited.

Agricultural signals and the onset of anthropogenic activity within the watershed

Between ca. 250 and 225 cal bp, anthropogenic activity began within the watershed. Signals of human activity associated with agriculture include high relative percentages of herbaceous pollen taxa (e.g. Asteraceae, Amaranthaceae, and Cyperaceae) and increases in microscopic charcoal influx (Figs. 3 and 4). These changes indicate fires set by people and increased herbaceous plant cover following forest clearance. At ca. 225 cal bp, an increase occurs in both microscopic charcoal and herbaceous pollen that is further evidence of agricultural activity and human occupation. Analyses of macroscopic charcoal and δ13COM by Johanson et al. 2020 also suggest that the area surrounding Laguna Carse had low population levels until ca. 250−225 cal bp. Increases in Cyperaceae pollen at this time could relate to growth of a fringing marsh area associated with increased soil delivery to the lake edge from agricultural land clearance (Bush 2002).

The timing of the initial human activity at ca. 225 cal bp corresponds with the historically documented colonial-period resettlement of members of the Térraba indigenous community to the Diquís Valley in 253 cal bp (1697 ce; Pittier 1941; Stone 1962). The increase in microscopic charcoal influx and herbaceous pollen at this time may indicate initial land clearance in preparation for agriculture and community establishment.

The first signals of agriculture in our pollen and microscopic records are linked to the historic date of settlement in the watershed. Macroscopic charcoal data from Johanson et al. 2020 show a slight delay (ca. 25 years) in fire activity following the historic settlement date before a sustained period of roughly constant fire activity that lasts approximately 100 years, from ca. 225 cal bp through the end of the LIA. The Carse δ13C record indicates increased agricultural activity at this time as well (Johanson et al. 2020; Fig. 3). It illustrates a replacement of C3 by C4 vegetation with a shift to more negative δ13C values at ca. 200 and ca. 170 cal bp. While evidence of maize pollen is absent in the pollen record until ca. 125 cal bp, other proxy signals strongly indicate that anthropogenic activity was occurring near the lake during the post-resettlement period.

The low recovery rate of Zea mays ssp. mays pollen in the Carse core could indicate that agricultural activity in the area was not occurring around the lake shore. Maize pollen has been shown to travel less than 60 m from a parent plant, due to its high sedimentation velocities, 18–30 cm/s (Raynor et al. 1972; Di-Giovanni et al. 1995; Jarosz et al. 2005; Horn 2006). For this reason, maize pollen will typically only be found in a lake-sediment core when the plant was cultivated close to the lake shore. Pollen of other cultivated crops, such as manioc, rarely preserves in neotropical sediments, or is indistinguishable from pollen of non-cultivated, wild varieties (e.g. legumes), and cannot be used in lieu of maize as a reliable indicator of agricultural activity in a pollen record (Clement and Horn 2001). Noting the travel limitation of maize pollen grains, the paucity of maize pollen in our record during the post-resettlement period indicates that maize was likely not grown on the lakeshore or immediate vicinity. However, from the pollen, charcoal, and previously published δ13C trends indicating maize agriculture (Johanson et al. 2020), we are confident that forest clearance and agricultural activities were occurring in the Laguna Carse watershed.

Following the end of the LIA and transitioning into modern times, our records indicate a possible change in land use patterns that likely represents the developmental progression of agricultural techniques and the implementation of new burning regimes and land management strategies. The microscopic charcoal influx from ca. 60 cal bp through the top of the core shows these modern agricultural burning practices. During this period, we see previously elevated δ13C levels sustained during the last 100 years of the LIA (ca. 200−100 cal bp) decrease sharply at ca. 60 years after the LIA ends (Fig. 3; Johanson et al. 2020). This δ13C trend aligns with an increase in macroscopic charcoal influx at the same time (Fig. 4; Johanson et al. 2020). The organic matter content also shows an increase following the end of the LIA, with a high initial rise congruent with the δ13C drop (Johanson et al. 2020). Subsequent to this, the record shows an increase in fire activity that occurs in tandem with increasing dominance of disturbance taxa and δ13C (Fig. 4). These conditions are likely a result of modern land management strategies due to changes in agricultural practices during the 20th century.

Comparison of microscopic and macroscopic charcoal records at Laguna Carse

Comparison of the microscopic charcoal record to the published macroscopic charcoal record from Laguna Carse (Johanson et al. 2020) shows relatively similar patterns of fire activity reflected in both data sets with differences present in the later portion of the LIA (Fig. 4). The differences in sample density between the records limit some comparative interpretations. Our microscopic charcoal analysis included 23 sample levels, while our macroscopic charcoal data are from 135 sample levels. The macroscopic record from Laguna Carse (Johanson et al. 2020) has a 1-year resolution from ca. 330 to 250 cal bp and a 4-year resolution for the more recent segment of the record from ca. 250 cal bp through modern times. The microscopic record has a wider sampling density and an average 15-year resolution, lower than the macroscopic charcoal record but higher than many tropical microscopic charcoal records (Clement and Horn 2001; Anchukaitis and Horn 2005; Johanson et al. 2019).

Both charcoal records show increased levels of fire activity that generally decrease from the base of the core through ca. 250 cal bp. This fire activity during a time of low population density in the region may represent wildfires during a period of drier conditions. At the time of the Térraba resettlement at 253 cal bp, microscopic charcoal influx shows an increasing trend that is also reflected in the macroscopic record, which increases at ca. 225 cal bp. The small difference in timing between increases in microscopic and macroscopic charcoal influx may relate to the differed sample intervals between the microscopic and macroscopic charcoal records. It could also reflect differential transport of smaller and larger charcoal particles, with smaller particles transported by wind reaching the core site sooner than larger particles (Clark 1988; Carcaillet et al. 2002).

During the last 100 years of the LIA (ca. 200−100 cal bp), the macroscopic and microscopic records both show an increase in influx roughly following the date of the historic Térraba resettlement. The macroscopic record displays a decline that starts at ca. 190 cal bp with low levels sustained through ca. 90 cal bp, but the microscopic record does not reflect this decline. However, while the microscopic record during this time period remains elevated, it does show fluctuations that follow the same general pattern of the increase-decrease-increase as the macroscopic record (Fig. 4). This indicates that fire was occurring consistently in the area away from the lake, and outside the transport range of most macroscopic charcoal, as microscopic records reflect charcoal from wider areas.

During the 20th century, from ca. 60 cal bp through modern times, the microscopic record shows a period of consistent influx, while the macroscopic record shows more variation. The macroscopic record indicates multiple high spikes in influx interspersed with periods of very low influx, specifically seen from ca. 45 to ca. 20 cal bp. This may be evidence of irregular small-scale burning near the lake that is depositing charcoal directly into the lake sediments coupled with a lessening in general fire activity in the region due to modern land management.

Climate at Laguna Carse during the LIA

Laguna Carse formed during the LIA and this climate event encompasses over two-thirds of the record. Pollen and charcoal vary over the LIA. Microscopic charcoal influx is highest in the earliest sediments of the core and generally decreased toward the end of the LIA. Macroscopic charcoal also trends downward in this period (Fig. 4; Johanson et al. 2020). From the core base until ca. 250 cal bp, microscopic charcoal area influx remains high and represents a period of persistent, likely climate-controlled, fire activity (Fig. 4). Generally, the proxy evidence indicates an overall drier environment surrounding Laguna Carse from lake formation at ca. 350 through ca. 250 cal bp. This inferred dry period is roughly coincident with charcoal peaks seen at Laguna Danta (12 km away from Laguna Carse; Johanson et al. 2020), where the macroscopic charcoal record shows similar fire activity at this time, from ca. 315 to 250 cal bp, suggesting that drier conditions were likely present over an area that encompassed both lakes. The sustained fire activity during the earlier part of the LIA may not represent agricultural fires during this time of low population density, but instead wildfires during drought intervals. A nearby record at Laguna Los Mangos (Johanson et al. 2019) documents a record of pre-agricultural fire activity, which supports the interpretation of climate-driven fire events during the early LIA at Laguna Carse.

Over the last 150 years of the LIA (ca. 250−100 cal bp), our pollen record indicates a slow increase in tree and shrub taxa that corresponds with a compelling decrease in macroscopic charcoal influx (Figs. 3 and 4). This period of forest recovery and diminished macroscopic charcoal influx may be driven by variability in LIA conditions within the region or reductions in anthropogenic fire. However, this signal is influenced by human action and not likely driven solely by a shift away from drought conditions in the region, as evidence of human-environment interaction and sustained fire activity in the wider region occurs at this time.

After the LIA ends, the Carse records show evidence of sustained forest recovery through a shift towards arboreal pollen taxa dominance together with lower charcoal influx. The decline in δ13C at ca. 150 cal bp is interpreted to represent decreasing agricultural activity coupled with growth in forest coverage (Fig. 3; Johanson et al. 2020). This may indicate either a shift in regional climate as the LIA ends or may be attributed to reduced anthropogenic activity. Overall, climate conditions at Laguna Carse likely varied throughout the LIA and were not uniformly cool and dry, but more likely consisted of extended droughts interspersed with intervals of higher precipitation that changed to a more moderate climate.

Conclusions

Our records from Laguna Carse show the history of a lake whose basin formed during a period of drought ca. 350 cal bp shortly following Spanish Conquest in the region (Fernández Guardia 1913; Nunley 1960). There was likely a century-long delay in occupation or significant anthropogenic exploitation following lake formation due to low regional population levels at the time. Initial settlement likely occurred following the colonial resettlement of members of the Térraba indigenous community into the Diquís Valley away from communities on the Caribbean slope of the Cordillera de Talamanca in 253 cal bp (Pittier 1941; Stone 1962). Our pollen and microscopic charcoal records show evidence of anthropogenic land clearance and onset of agricultural activity at ca. 225 cal bp that is roughly congruent with the historical settlement date. The Laguna Carse records indicate intervals of drought during the LIA at the time of lake basin’s formation, ca. 350 cal bp, that persisted until ca. 250 cal bp, when agriculture and anthropogenic activities began. The vegetation of the landscape surrounding Laguna Carse is comprised of lowland flora typical of the region and the pollen record indicates shifts in vegetation over time from a period with higher herbaceous taxa coverage (ca. 325 to 220 cal bp) to a forest-dominated system (ca. 220 to 50 cal bp). A low abundance of maize pollen found in the core indicates that agricultural fields in the watershed were likely not located adjacent to the lakeshore.