Biomisation of the Lake La Cocha pollen record Comparing an equatorial Colombian Andean Lateglacial-Holocene temperature reconstruction to ice core temperature proxies

(1)

Biomisation of the Lake La Cocha pollen record

Comparing an equatorial Colombian Andean Lateglacial-Holocene

temperature reconstruction to ice core temperature proxies

Bachelor these Earth Sciences

Serge Mooyman

1

st

supervisor: dr.ir. J. H. Boxel

2

nd

supervisor: prof.dr. H. Hooghiemstra

University of Amsterdam 2014 / 2015

(2)

Abstract

Knowledge of past climate dynamics and their drivers provides relevant information in understanding past changes as well as current and future climate changes. This research aims to reconstruct regional Lateglacial-Holocene ecosystem- and climate dynamics, and distinguish human impact on the vegetation in equatorial Andean southwest Colombia.

Biomes are usefull informants of regional climates. They can be distinguished by the plant functional types they contain. Biomes can be reconstructed from a sediment core by assigning the pollen grains it contains to the different plant functional types present in the regio. Human impact on the vegetation is reflected by pollen grains from agricultural herbs, pollen grains of herb taxa that typically accompany agriculture, and decreases in pollen grains of traditional logging trees.

Assignment of radiocarbon dated high resolution pollen data from the Lake La Cocha sediment core into plant functional types permitted downcore calculation of biome affinity scores. This allowed detailled reconstruction of the biome affinity dynamics over the last 14000 years BP. The resulting biome affinity dynamics were translated in climatic changes.

Páramo biomes dominated at the elevation of Lake La Cocha until 10000 years BP, indicating a colder climate. After 10000 years BP grass páramo affinity gradually decreased in favor of forest biomes, suggesting increasing temperatures. A mean annual temperature

reconstruction based on arboreal pollen percentages depicted a gradual temperature rise of 2.7 °C over the the last 14000 years BP. No sharp temperature raise could be distinguished at the transition from the last glacial to the Holocene. Comparison to climate dynamic of other regions showed mostly low resemblance. The hypothesis of higher tropical Holocene climate dynamics suggested by earlier research was confirmed, and indications of human disturbance on the vegetation were found after 2100 years BP.

Keywords: arboreal pollen, biomes, Lateglacial - Holocene climate change, downcore

reconstruction, vegetation dynamics, plant functional types, pollen taxa, sediment core, upper forest line.

(3)

Introduction

According to scientific consensus, current global warming and climate change is driven by anthropogenic emissions of greenhouse gasses and the decrease of important carbon sinks like tropical forests (Marchant et al., 2009; Houghton, 2009; IPCC 2013). In a complex system like the global climate, drivers of change can be difficult to distinguish. This is a main argument of skeptics on anthropogenic global warming, that is paralyzing an effective policy to prevent and mitigate enhanced anthropogenic climate change, with crucial consequences for sensitive ecosystems humanity depends on (Houghton, 2009). Enhanced mapping of climate change of the past can contribute to a better understanding and distinction of drivers of historical changes as well as current and future climate changes (Van Boxel et al., 2014), and additionally provide essential knowledge for biodiversity conservation strategies in a changing climate (Bush et al., 2007).

Whereas ice cores from Greenland and Antarctica have provided high resolution data that allowed reconstruction of the Holocene climate as relatively stable, much less is known about tropical Holocene montane forest dynamics. Ice cores from tropical glaciers depict a more dynamic postglacial climate (González-Carranza et al., 2012; IPCC 2001, p 92). However below the elevation of these glaciers climate change data are underrepresented (González-Carranza et al., 2012; IPCC 2001; Bush et al., 2007).

Pollen based vegetation reconstructions have demonstrated how ecosystems migrate to stay within their ecological range when climate boundaries are shifting. In mountainous areas this is expressed by altitudinal shifts of the region specific biomes (Rolefes, 2015; González-Carranza et al., 2012; Torres et al, 2006; Wille et al., 2001; Feeley et al., 2011). Biomisation is a method introduced by Prentice et al. in 1992, that allows reconstruction of ecosystem vegetation and their dynamics through time based on pollen analysis. Initially this technique was developed to test the results of biome reconstructions by vegetation models. Results using this method that have been compared to actual vegetation and modeling output, (e.g. by Marchant et al., 2002: Marchant et al., 2004;), have depicted high accuracy. However, biomisation proved to possess additional features of interest, of which the possibility to link vegetation dynamics to climate dynamics is of particular interest. As ecosystems possess clearly defined envelopes of

climatological tolerances, it is well possible to relate their dynamics to changes in temperature and precipitation regimes (Marchant et al, 2001). Another interesting feature is the possibility to recognize and date anthropogenic vegetation disturbance by the decrease and increase of certain pollen taxa. One example is the decrease of arboreal pollen taxa of trees that were popular for logging. Another example are increases in pollen taxa that are indicative of agricultural activity. Due to these features downcore biomisation qualifies for an objective investigation of vegetation response to climatic forcing and past human vegetation interactions (Marchant et al, 2001). Additionally results can be reproduced, since biomisation is based on calculation of empirical data, (Torres et al., 2006).

Lake sediments with fossil pollen and diatom are known for their capacities to provide good data for reconstruction of past changes in vegetation, environments and climate (González-Carranza et al., 2012). A sediment core that provides high resolution pollen and spores data was retrieved from Lake La Cocha in equatorial south west Colombian Andes at an elevation of 2780 meter above sea level (asl).This core contains suitable information for a downcore

(5)

reconstruction of the past vegetation and climate dynamics of the last 14000 years (yr) before present (BP: 1950 AD). Other sediment cores, like the Fuquene and Funza cores near Bogota (Colombia), often go back deeper in time, but also often present a coarser resolution, and therefore larger time steps (Torres et al., 2006). Additionally, locations near the altitudinal boundary between two different biome types offer the advantage of a clearly defined distinction which of these two biome types dominates at a certain point in time. Lake La Cocha is situated in the middle of the dynamic range of the last 14000 yr of the upper forest line (UFL), which is the natural boundary that divides the lower situated Andean forest from the higher elevated páramo.The biomes occurring in the study area are grass páramo (3600-4200 m asl), subpáramo (3200-3500 m asl), cool upper montane forest (between 2200 and

3200 m asl), warm lower montane (subandean) forest (1200-2300 m asl), and lowland forest (below 1200 m asl. Superpáramo is excluded, because elevations in the study area do not surpass 3650 m asl.The biomes are depicted in Figure 1 (after Wille et al., 2001), which includes a schematic representation of altitudinal biome migration since the Last Glacial

Maximum. Although the comparisment between present day and the Last Glacial Maximum also depicts how biomes change in vertical size under different climate regimes, this factor is less pronounced within the relatively moderate temperature changes during the epoque of research (the last ~14000 yr BP). Generally a vertical distribution of ~500 meter for grassparamo, 100– 200 meter for subparamo, and ~1200 meter of Andean forest is assumed (González-Carranza et al., 2012).

Figure 1: Visual impression of the biomes occurring in the area of study and an example of altitudinal biome migration under changes in climate regime (after Wille et al., 2001).

In this bachelor thesisbiomisation of the La Cocha core is applied in order to reconstruct regional Lateglacial-Holocene ecosystem dynamics and subsequently climate dynamics. The

(6)

results allow comparison to reconstructions of other climate proxies. A connected target is the distinction and dating of human impact on the local vegetation. This is reflected in the increase or decrease of pollen taxa that are indicative of human impact (Marchant et al., 2001; Torres et al., 2006; González-Carranza et al., 2012). The aim of this research is to reconstruct the dynamics of the different biomes in terms of affinity scores over the last 14 kyr (14000 years), and interpret the significance of the results for the regional climate. Two research questions were subtracted:

 Can we distinguish and date human impact on vegetation through time?

 Are there significant similarities and/ or differences with environmental dynamics shown by other climate proxies?

Results can address the hypothesis that tropical Holocene climate change has been more dynamic than climate change in the temperate regions (González-Carranza et al., 2012; IPCC, 2001), as well as provide insight in how global climate change might affect the considered biomes.

(7)

2 Materials and methods

2.1 Environmental setting of the study area

Lake La Cocha is located in the Guamués Basin near the equator at the eastern side of the Andes mountain range, in tropical South-West Colombia (01º06'N, 77º09W) at 2780 meter asl (Fig.2). The lake is surrounded by wetlands and elevations between 2800 and 3650 meters asl.

Figure 2: a. Location of Lake La Cocha, b. Combined vegetation and elevation map The catchment is bordered by a green line. A small green arrow indicates the extraction location of the core (Moreno Diaz 2004; Van Boxel et al., 2014).

2.2 Earlier research

A 12 meter long sediment core was taken from the wetlands on the north side of the lake (Fig. 2 b) using a 50 centimeter long Russian Corer (Van Boxel et al., 2014, Fig. 3a). It consists of lacustrine and peaty sequences, containing fossil pollen grains and spores (Fig. 3b). The upper 100 centimeter was not suitable for analysis due to superfluous water content, and is therefore left out of consideration.

(8)

In an earlier research conducted by González-Carranza et al. (2012), radiocarbon dating of the deeper 1100 centimeters of this sediment core depicted a time span ranging from 130 until 14079 calibrated 14C yr BP. Core samples were analyzed each 2 centimeter, resulting in 549 pollen samples with time steps of averagely ~27 yr. The core offers a high resolution that

provides good data for reconstruction of ecosystem dynamics (González-Carranza et al., 2012). By using higher resolution data the precision of the results can be improved, and a phenomenon like saw tooth graphing might be avoided. Saw tooth graphing can occur as a result of data sampling that is too coarse to smoothly represent system changes (Marchant et al., 2002). Ecosystems in Colombia have responded to environmental changes at a timescale of close to a thousand yr averagely over the last 6000 yr (Marchant et al., 2001).The analyzed pollen and spores were grouped in 8 ecological classes of which only sub-Andean forest, Andean forest, subpáramo and grass páramo indicated biomes. Only pollen grains from these 4 biome classes were included in the pollen sum. Pollen grains of water plants were excluded, because they reflect lake conditions. The pollen sum contained 350 pollen grains where possible. Most pollen grains could not be identified to the species level but until genus level (Fig. 3c). Generally this provides sufficient information for biomisation.

Russian corer Identification of pollen and spores Taxonomic .

ranking

Figure 3a Figure 3b Figure 3c

Figure 3a: Russian corer as used to extract the La Cocha sediment core (academic.emporia.edu). 3b. Pollen identification chard as used for pollen counting (palaeo-electronica.org). 3c. Hierarchy of the taxonomic

ranks (science.unsw. edu.au).

2.3 Method

The method used to apply biomisation on the Lake La Cocha core is based on the research Torres et al. (2006) did at the Funza 2 core, located roughly 500 kilometers northeast in the

(9)

Bogota basin (4°43’N, 74°13’W) at 2550 meter asl. Its facets are described in the following sections.

2.3.1 Plant Functional Types and biomes

Down core biomisation is a method that allows reconstruction of ecosystems and their dynamics through time. In the first step, pollen grain fossils from dated samples are assigned to plant functional types (PFTs). PFTs are groups of taxa that have common characteristics (Prentice et al., 1992; Marchant et al., 2009). There are two selection criteria for the PFT’s: climatological and/ or phenological conditions, and plant and/ or leave form. Table 1 (after Torres et al., 2006), provides an overview of the PFTs that can be distinguished in the area of study. A full overview and descriptive naming of the PTFs can be found in Appendix A.

Table 1: Code names and descriptive names of the PFTs, divided in two groups by the two different selection criteria: 1. Climatological and/ or phenological conditions, 2. Plant and/ or leave form. After Torres et al. (2006).

Most pollen grains could not be identified at species level, but until genus level only (González-Carranza et al., 2012). This often implies that one pollen taxon represents multiple species (Torres et al., 2006). These different species can belong to different PFTs. Therefore a taxon is often assigned to more than one PFT (Marchant et al., 2009).

In the second step of the biomisation process the PFTs are assigned to the biomes where the taxa they contain typically can be found. This step is summarized in table 2. PFTs can often be found in multiple biomes and biomes can contain multiple PFTs, as is illustrated in Figure 4 (Torres et al., 2006). Torres et al. (2006) distinguished the PFTs that occur in the area of research, as well as their assignment to the occurring biomes and position in the climatological and elevational spectra of the Colombian Andes (Fig. 4). The first part of the full code name of each PFT refers to the first selection criterium, and the last part to the second criterium of table 1.

(10)

Table 2. Assignment of the plant functional types (PFTs) to the biomes (Torres et al., 2006).

Positioning of biomes and plant functional types in the climates of the Colombian Andes

Figure 4: Scheme of the approximated distribution within two climate variables of the biomes and

contributing PFTs occurring in the Colombian Andes. The horizontal axis is divided in three humidity classes and the vertical axis represents mean annual temperature (MAT). The bold line depicts a cross section through the Andes at Bogotá latitude, depicting at which elevations the biomes and PFTs can be found (after Torres et al., 2006).

(11)

Table 3: The biomes occuring at the area of study, their code name, and their typical climatological conditions. After Torres et al. (2006).

2.3.2 Calculation of the biome affinity scores

In the third step of the biomisation process the biome affinity scores (BAFs) are calculated. BAFs are calculated based on the quantitative assignment of characterizing PFTs (Marchant et al, 2002). The affinity scores of the occurring biomes in Lake La Cocha over the given period in time have been calculated using software written by Van Boxel (BIOMISATION V05.b). The calculation performed by BIOMISATION V05.b is based on an algorithm first applied by Prentice et al. (1996).

[1]

Formula [1] gives the algorithm of Prentice et al. (1996) to calculate biome affinity. “As,b is the affinity score of

biome b in sample s. t is the taxon number. N is the total number of taxa. δt,b is equal to 1 if taxon t is a

member of biome b and equal to 0 otherwise. Pt,s is the pollen percentage of taxon t in sample s. Ɵt is a

threshold pollen percentage, commonly 0.5% for all taxa.” Torres et al. (2006).

A threshold of 0.5% of the pollen sum is set to avoid the effect of erroneous identifications and of pollen grains blowing in from far away.If the pollen percentage of a taxon in a sample (Pt,s) is 0.5 % or less of the pollen sum of that sample (s), the threshold neutralizes it. If the pollen percentage minus the threshold however is a positive number, the square root of that number is taken and subsequently multiplied by δt,b = 1 if taxon t is a member of biome (b), and by δt,b = 0 if taxon t is not a member of biome (b). This way each contribution to a BAF is assigned only to the biome it is associated with. Taking the square root increases the sensitivity to less abundant taxa (Prentice et al., 1996). All pollen percentages of taxa that surpass the threshold in a specific sample, contribute to the BAF of the biome they are associated with, in that sample. By summation, the total affinity scores of the biomes in the sample in question are calculated. By calculating the biome affinity scores of each subsequent pollen sample in time, down core biomisation reconstructs the dynamics of the different biome affinities at the core location. Details on the calculation process can be found in Prentice et al. (1996).

(12)

BIOMISATION V05.b requires 3 different text files as input in order to calculate the BAFs: 1) The pollen count data

2) The assignments of pollen taxa to the PFTs 3) The assignments of PFTs to the biomes

The data with sample depth, 14C age BP and pollen counts of the different taxa originate from the research conducted by González-Carranza et al. (2012). The input information that directs the BIOMISATION V05.b software in assigning pollen taxa to PFTs and PFT’s to biomes are based on the interpretation of Torres et al. (2006) as applied for biomisation of the relatively nearby Funza 2 core. Climate, vegetation, elevation and PFTs of Funza are comparable to Lake La Cocha. Minor deviations from the Torres et al. (2006) interpretation have been made under supervision of professor Hooghiemstra. These concerned assignment to PFTs of certain pollen taxa from the group that is not classified yet (this group can be found under PFT NON in Appendix B).

2.3.3 Interpolation

BIOMISATION V05.b was applied at the default minimum threshold of 100 pollen grains. Out of 549 samples 122 samples gave no results. This is because the sum of pollen grains that could be assigned to biomes was less than 100. Since data continuity and equidistant steps for sample depth can be required for statistical processing, linear interpolation from the values of the nearest neighboring data points has been applied to fill in the missing results. This and all other statistical analysis of the results of BIOMISATION V05.b have been performed in MATLAB 2015a.

2.3.4 Additional calculation

In order to improve results, a correction constant for the páramo affinity scores that was introduced by Rolefes (2015) has been applied. This correction constant is based on the

position of the UFL. Theoretically the ratio of COMI/CGSH should be 1 at the boundary between them, the UFL (Hooghiemstra et al., 2012). However, in Figure 5, where the calculated arboreal pollen percentages (AP%) are plotted against the COMI/CGSH ratio, the value of the regression line through the samples at the UFL, that corresponds with an AP% of ~40 (González-Carranza et al., 2012), is not 1, but 1.22. This implicates that the resulting páramo scores should be corrected by multiplication with 1.22 in order to be accurate.

In order to quantify relations between the different biomes, Pearson correlations between the BAFs were calculated. Tropical seasonal forest was not included, because it scores very low. A mean annual temperature reconstruction (MAT, in °C) was based on the AP%.A key tool in MAT reconstruction is the position of the UFL, corresponding with a MAT of ~9.5 °C and a AP% of 40 (van Boxel et al., 2014; González-Carranza et al., 2012; Groot et al., 2013; Bogotá-Angel et al., 2011). According to González-Carranza et al. (2012) the lowest and highest UFL in La Cocha, 2300 - 3550 meter, corresponded with respectively 10 and 60-70 % AP at lake level. This implies an increase or decrease of the elevation of the UFL of 100 meter with respectively an increase or decrease of ~ 5 % AP at lake level. By applying a lapse rate of -0.5 °C for every

(13)

+100 meter change in elevation under the local humid conditions (Van Boxel et al., 2014; Gonzalez-Carranza et al. 2012), MAT for all elevations can be reconstructed. This line of reasoning has been applied to reconstruct MAT dynamics by calculating the AP%. The MAT at La Cocha lake level (2780 meter asl) has been calculated using equation 2.

[2]

2.3.5 Comparing mean annual temperatures of Lake La Cocha to other regions

In order to compare the dynamics of the Lake La Cocha MAT reconstruction to dynamics of regional and global MAT, MAT proxies from ice cores from Greenland (GISP2), Antarctica (EPICA and Vostok) and the north Peruvian tropical Huascaran glacier have been graphed. The Huascaran tropical glacier is located roughly 1100 kilometers south from Lake La Cocha. A frequently used MAT proxy is the ∂18O isotope level (Fricke & O’Neill, 1999).

2.3.6 Pearson cross correlations of mean annual temperatures

In order to quantify similarity between the arboreal pollen based MAT reconstruction from Lake La Cocha and MAT reconstructions from the other regions, Pearson cross correlations were calculated with MAT proxies of Greenland (GISP2), Antarctic (EPICA and Vostok) and Huascaran ice cores. Since a cool phase before the end of the last ice age end was absent in the Lake La Cocha MAT reconstruction, cross correlations between Lake La Cocha and other proxies where calculated over te last 10 kyr. An additional aspect is the possibility of time lags between the signals of different temperature proxies. Jansen et al. (2013) state that upslope wind-blown pollen signals can be over a millennium ahead of the actual vegetation at páramo locations. They identified a migration lag between pollen grains (earlier) and biomarkers (later). This time lag reflects the difference between the non transported biomarker signal showing spot-dating, and the upslope wind-blown pollen grains (Jansen et al., 2013). When comparing different temperature proxies, the possibility of such a time lag has been taken into account. To achieve this, Pearson cross correlations have been considered including time lags ranging between -5 and 5 kyr.

2.3.7 Human indicators

Evidence of substantial human presence at higher elevations in the Andes has been found dating back untill ~3 kyr BP (Marchant et al., 2001; Torres et al., 2006). Marchant et al. (2001) defined an additional “degraded vegetation” PFT reflecting evidence for human disturbances on vegetation. This PFT contains region specific agricultural crops and herb taxa that are often found in the vicinity of agricultural fields. Examples of traditional crops are quinoa

(Chenopodiaceae), potato (Solanum), cassava (Manihot ) and maize (Zea mays). An example of herbs that are often found around agriculture is sorrel (Rumex). Another indicator of human impact on the vegetation is a decrease in those tree taxa that were traditionally used for timber.

(14)

In this research 2 PFTs were added indicating respectively agricultural activity and logging activity (PTFs HuIm and Logg in Appendix B and C). Results of the two were compared for matches. More elaborate literature on this topic can be found in Marchant et al. (2009).

An additional step that has been taken in order to distinguish the signal that indicates logging better from the general AP%, is to compare the logging signal with the dynamics of all arboreal pollen. Assuming that without human interference pollen of logging trees correlate well with arboreal pollen in general, a downfall in pollen of logging trees is a stronger indication of human activity when arboreal pollen oppose this trend, but weaker evidence when the AP% decreases as well. To detrend the logging signal with the arboreal pollen signal, the logging signal was divided by the ratio of the general AP% and its average.

(15)

3. Results

3.1 Correction factor for the páramo affinity scores

Calculation of the correction constant for páramo BAFs was done by the method Rolefes (2015). The regression line through the data has a value of 1.22 at an AP% of 40, yielding a multiplication factor for páramo biomes of 1.22.

Figure 5: The AP% against the COMI / CGSH ratio. The black line gives the regression through the data. The black dashed lines give the 95% confidence interval. The red dashed line depicts the AP% around the UFL. The green dashed line depicts the vaue of the regression at the UFL.

3.2 Biome affinity scores before and after correction for páramo

The calculated BAFs between 130-14079 14C yr BP were interpolated for depths without result, visualized (Fig. 6 ), corrected for páramo BAFs, and subsequently visualized once more (Fig. 7 ). Clearly visible is how the paramo BAFs are lifted 22% after appliance of the páramo correction factor.

(16)

Figure 6: BAFs against age before application of the correction factor for páramo. The yellow line is dashed because otherwise it would often cover the black line.

(17)

3.3 Pearson correlations between the biome affinity scores

Pearson correlations between the BAFs scores were taken in order to quantify relations between the different biomes (Fig 8).

Figure 8: Pearson correlations between the different BAFs. The diagonal depicts with histograms how BAFs are divided.

(18)

The most important correlations of the BAFs were graphed together in order to facilitate detailed inspection (Fig. 9). Tropical seasonal forest was excluded, because it scores very low. WEFO – COMI and COMI – CGSH are not depicted since the are almost identical to CEFO –WEFO and CEFO-CGSH.

Most important correlations between the BAFs

(19)

3.4 Mean annual temperature reconstruction

Figure 10: Lake level MAT reconstruction for La Cocha based on the arboreal pollen %. The red line depicts the regression of the MAT. The green dashed line depicts the MAT at the upper forest line.

Figure 10 represents a reconstruction of MAT through time of Lake La Cocha. The regression line indicates an increase of MAT of 2.7 °C over the full period. A clear and rapid temperatured increase between the end of the last glacial and the beginning of the Holocene, as present in other climate proxies like the Fuquene 2 record (van Geel & van der Hammen, 1973) and the Páramo de Frontino record (Velásquez & Hooghiemstra, 2013), is not detectable. The last 4.3 kyr BP are characterized by high dynamics with large amplitudes. The high

dynamics might be explained by a sample distance that nearly triples within several hundred year in age around 4.3 kyr BP.

(20)

3.5 MAT of Lake La Cocha compared to other regions

Figure 11: ∂18O isotope levels from the Greenland ice core GISP2 compared to the Lake La Cocha MAT reconstruction.

Figure 11 compares ∂18O isotope levels from the Greenland ice core GISP2 (Stuiver et al., 1993) with the Lake La Cocha MAT reconstruction. The data possess an ultra high resolution. The colder Younger Dryas between 12.7 - 11.7 kyr BP is clearly recogizable in GISP2. Also the 8.2 kyr cooling event is well recognizable. The 8.2 kyr event is described as a 161 yr cool and dry period of which clear evidence has been found in four ice cores in north, central and south Greenland. It was as a rapid cooling with a magnitude of 7.4 °C temperature change in

Greenland. However there is no strong evidence for effects outside the North Atlantic region (Thomas et al., 2007).

(21)

Figure 12: Comparison between MAT dynamics of Lake La Cocha and Antaric ice cores EPICA and Vostok.

Figure 12 compares MAT dynamics of Lake La Cocha with two Antarctic ice cores EPICA and Vostok. The Antarctic cores depict a clear distinction between the end of the last glacial and the Holocene (~ 12 kyr BP), where this is absent in the Lake La Cocha reconstruction. Source of the data is climatedata.info.

(22)

Figure 13: Dynamics of Lake La Cocha MAT compared to ∂18O isotope dynamics in an ice core from the Huascaran glacier (Peru, 09°07'S 77°36'W, elevation 6050 meter asl, Thompson et al., 1995).

In Figure 13 dynamics of Lake La Cocha MAT are compared to ∂18O isotope dynamics in an ice core from the tropical Huascaran glacier (Peru, 09°07'S 77°36'W, elevation 6050 m asl, Thompson et al., 1995). The Huascaran ∂18O record has more similarity with the Antarctic ∂18O record than with the the Greenland GISP2 ∂18O record (Ramirez et al., 2003). The Huascaran climate was warmest from 10500 to 8600 yr BP (Thompson et al., 2006).

Subsequently it cooled gradually, culminating with the Little Ice Age (200-500 yr BP). The last part of the graph depicts a strong increase in temperature (Thompson et al., 1995). Huascaran data where gathered by Thompson et al. (1995) and retrieved from climatedata.info.

(23)

3.6 Pearson cross correlations of mean annual temperatures

Figure 14: Cross correlations over the last 10 kyr between reconstructed MAT from lake La Cocha and other temperature reconstructions: a. with the Greenland GISP2 ice core, b. with the Antarctic EPICA ice core, c. with the Antarctic Vostoc core and d. with the tropical Huascaran glacier ice core (Peru).

Figure 14a depicts the cross correlations between reconstructed MAT from Lake La Cocha and the Greenland GISP2 ice core over the last 10 kyr. The maximum absolute cross

correlation is les than 0.15. Figure 14b depicts the cross correlations between reconstructed MAT from Lake La Cocha and the Antarctic EPICA ice core over the last 10 kyr. The maximum cross correlation is 0.42, corresponding with 18% explained variance. The Lake La Cocha pollen signal precedes the EPICA signal 1.2 kyr. This resembles the expected precedence of the pollen signal (Jansen et al., 2013). Figure 14c depicts low maximum cross correlation (0.20) between reconstructed MAT from Lake La Cocha and the Antarctic Vostok ice core over the last 10 kyr, corresponding with only 4 % explained variance. Additionally at this point the Lake La

(24)

Cocha pollen signal lags 600 yr behind the Vostok signal, where it could be expected to precede with a similar magnitude. Figure 14d depicts the cross correlations between reconstructed MAT from Lake La Cocha and ∂18O isotope from the tropical Huascaran glacier ice core (Peru) over the last 10 kyr. The graph depicts a high maximum absolute cross correlation (-0.51)

corresponding with over 25 % explained variance.

3.7 Human indicators

Figure 15: Human indicators: a. Affinity scores indicating agricultural activity over the full period. The full period is graphed in order to identify differences in dynamics before and after human presence would be likely, b. affinity scores indicating agricultural activity over the period in which human impact could be expected at the elevation of Lake La Cocha, c. affinity scores for traditional logging trees over the full period, and d. affinity scores for trees traditionally popular for logging in the period human impact can be expected. Agricultural indicators (Fig. 15b) are also graphed, in order to identify matches. A red dashed line means no match, a green dashed line is a match.

(25)

The first indicator of human activity is increase in affinity score for agricultural indicators. The second indicator of human activity is decrease in affinity score for traditional logging trees

Weinmannia, Podocarpus (conifers) and Quercus (oak). The figures on the right depict the period in which human activity would be likely. Both indicators are also graphed over the full periode to compare dynamics before and after human presence would be likely. Before ~3 kyr BP human presence is unlikely at Lake La Cocha elevation (Marchant et al., 2001; Torres et al., 2006).

3.7.1 The logging signal detrended with the arboreal pollen signal

Detrending the logging signal with the arboreal pollen signal assumes high correlation between pollen of logging trees and arboreal pollen without human interference. Both signals before 2.5 kyr BP are depicted in Figure 16a. The Pearson correlation is 0.71. The detrended result from 2.5 kyr BP is graphed in Figure 16 b.Agricultural indicators (Fig. 15b) are also graphed, in order to identify matches.

Figure 16 (left): Comparison of the logging signal with the overall arboreal pollen signal before human impact can be expected (2.5 kyr BP). Correlation is 0.71. In Figure 17 (right) the logging signal is detrended with the general arboreal pollen signal, in order to logging activity better. Including agricultural indicators (Fig. 15b), in order to identify matches.

4. Discussion

4.1 Biomisation of the la Cocha core

Comparison of Figures 6 and 7 depicts how the BAFs of the páramo biomes are lifted with 22% by applying the páramo correction factor. Not only the biomes with the highest calculated BAF were considered to interpret the final result (Fig. 7), but also the other biomes, since their dynamics can reflect relevant information about climatic changes (Marchant et al., 2002).

Figure 7 depicts biome dynamics at high resolution. In general both colder páramo biomes dominate at Lake La Cocha elevation (2780 meter asl) until 7.3 kyr BP. The shrub páramo scores highest, suggesting a colder period. After that it still dominates mostly, but is approached, equaled and sometimes even surpassed by cool mixed forest and even cool

(26)

evergreen forest, suggesting increasing temperatures. Around 8 kyr BP grass páramo is no longer the second best scoring biome. It is increasingly surpassed by cool mixed forest and even cool evergreen forest, except in the last 1.4 kyr BP, where the latter two depict a strong decrease that can not be easily explained otherwise than human influence in favor of logging and agriculture (taxa for logging and agricultural indication can be found in appendix B), since warm evergreen forest depicts a similar decrease, and the other biomes have no clear change.

Inspection of the details of Figure 7 shows that until 9 kyr BP BAFS for cool dry seasonal forest are increasing, indicating a dryer climate. This is supported by the first appearance of tropical evergreen forest in the second half of this period. Between 11.7 - 11.1 kyr BP warm evergreen forest clearly peaks, after an initially slow increase. This could be interpreted as increased precipitation and temperature, but all other forest biomes, especially cool mixed forest and cool evergreen forest are peaking as well in this period. Even the relatively marginally represented tropical seasonal forest is reacting for the first time. Since cool dry seasonal forest is also increasing at the same time, it is unlikely to simply suggest increased precipitation. Although the cooler biomes are also increasing, they increase less. Based on the MAT

reconstruction (Fig. 10) it seems reasonable to suggest all biomes have benefitted from a first relatively clear increase in temperature since the end of the last glacial, but that warm and moderately cool biomes profited more.

Something similar can be suggested in the period 10.5 to 8.5 kyr BP. All BAFs but the coldest, cool grass páramo, are increasing, especially until 9.5 kyr BP, dry as well as wet biomes. The MAT reconstruction (Fig. 10) confirms temperature increase in accordance to this description. Between 10 - 7.3 kyr BP shrub páramo is still scoring highest, but cool mixed forest and cool evergreen forest affinities are now competing with grass páramo affinity, indicating increased temperatures and after 9 kyr BP a wetter climate. The clear decrease in cool dry seasonal forest BAFs between 9 to 7 kyr BP confirm a less dry climate.

After 7 kyr BP BAFs for the grass páramo are generally decreasing, while warm evergreen forest scores higher. BAFs for cool non- drought indicators increase, suggesting that these biomes benefit at 2780 m asl. Between 6750 and 4900 yr BP there is a considerable increase of the BAFs of cool mixed forest and cool evergreen forest, and a decrease of them after that until 3.7 kyr BP. Shrub páramo seems to follow this trend. The absence of activity in the BAFs for tropical seasonal forest between 6750 and 4900 yr BP suggests a slightly wetter cooler period. Grass páramo clearly increases after 5.7 kyr BP, suggesting a cooler climate.

Although it seems logical to explain an increase in cool biomes as a cooling trend, the impression is that sometimes these biomes still seem to benefit from higher temperatures at the elevation of the core. The MAT reconstruction (Fig. 10) indeed depicts a warming trend between 6.7-5.7 kyr BP, and then cooling until 5 kyr BP. From 4.9 to 3.9 kyr BP there is a general

decrease in BAFs of shrub páramo, cool mixed forest and cool evergreen forest, but an overall increase in BAFs the coldest biome, grass páramo. Again this seems to suggest that these cool biomes do not react in the same way as grass páramo reacts to a climate trend at this elevation. The correlations between some cool biomes are indeed lower than one might expect. Pearson correlations (Fig. 8 and 9 ) between grass páramo and cool mixed forest/ cool evergreen forest are slightly below zero, while the correlation between grass páramo and shrub páramo is only 0.36. The correlation between the shrub páramo and the cool mixed and evergreen forest is

(27)

stronger, with 0.75 and 0.78 respectively. And although they are classified as cool biomes, increase in the latter three seems often rather an indication of warming than of cooling at Lake La Cocha elevation. This is supported by the MAT reconstruction (Fig. 10).

The BAFs from 3.9 to 2.2 kyr BP depict an increase in all four highest biomes (grass páramo, shrub páramo, cool mixed forest and cool evergreen forest), while the three lower biomes give little information, apart from relatively high scores for tropical seasonal forest. Since the increased scores of the four coolest biomes contradict temperature increase, the increased BAF of tropical seasonal forest should be interpreted as caused by a dryer climate in this period. The MAT reconstruction over this period (Fig. 10) depicts a general cooling over this period.

The period 2.2 to 1.3 kyr BP depicts a decreasing trend in the BAFs of the four highest biomes (grass páramo, shrub páramo, cool mixed forest and cool evergreen forest. Warm evergreen forest has its most significant peak in this period. This indicates a warmer and wetter climate. Human activity at higher elevations became relevant in this period in Colombia

(Marchant et al., 2001), implicating that the AP% the MAT reconstruction is based on and the BAFs might be affected. However, in this period the MAT reconstruction (Fig. 10) depicts increased temperatures, supporting the conclusion drawn from the biomes.

Between 1.3 - 0.3 kyr BP a decreasing trend is visible of all BAFs except cool dry seasonal forest and grass páramo. This is supported by the MAT reconstruction (Fig. 10). Whereas this would normally suggest a cooler and dryer climate, disturbance by human activities now prohibits such a conclusion. The last part of this period coincides with a more global climate event known as the Little Ice Age (200 - 500 yr BP, Thompson, 1995).

When we compare the visualizations of biomisation and especially the MAT reconstruction of Lake La Cocha to those of the other climate proxies, it becomes clear that the hypothesis of a more dynamic Holocene in the neotropics compared to temperate regions is supported by the data. The marking points of changes in BAF trends are generally supported by those of the MAT reconstruction.

Figures 8 and 9 depict the Pearson correlations between the BAFs. The correlation between cool evergreen forest and cool mixed forest is 0.98. This can be explained since the taxa they ompared to contain are the same, except that cool mixed forest contains Podocarpus and cool evergreen forest does not. Correlations between COMI and other biomes therefore are very similar to correlations between CEFO and these other biomes. But also the correlations between cool evergreen forest and warm evergreen forest, cool evergreen forest and shrub páramo, cool mixed forest and warm evergreen forest, and cool mixed forest and shrub páramo are high (Fig.8). Other higher correlations are cool dry seasonal forest and cool grass shrub páramo, shrub páramo and warm evergeen forest, and grass páramo and shrub páramo (Fig.8).

4.2 Reconstruction of mean annual temperatures

Regression through the MAT reconstruction depicts a raise of MAT of 2.7 °C over the last 14 kyr BP. The Lake La Cocha MAT reconstruction resembles most to Antartic EPICA MAT. The profile of Lake La Cocha MAT change however deviates considerably from other MAT

(28)

be recognized. During the Younger Dryas, midway through the transition from the last glacial period to the present interglacial (11650 -12850 BP), the climate returned to almost full glacial conditions for about 1000 yr. Since most evidence for the Younger Dryas indeed comes from regions in and around the North Atlantic, it is associated with this region (Rodbell,2000). There has been much debate about the presence or absence of the Younger Dryas return to glacial circumstances in the neotropics (Bush et al., 2007). According to Bush et al. (2007)

paleoclimate records appear to support an Antarctic-style deglaciation in the Andes south of ~10 S, while further north the characteristic climatic dynamics of the North Atlantic and Greenland seem to be reflected in the data. Based on this assumption the Lake La Cocha climate could be expected to resemble the Greenland reconstruction to a certain extend. However, cross correlation is low even in the last 10 kyr BP. Although there is a depression in the MAT reconstruction in this period, it is much less pronounced than even in the Peruvian Huascaran data. Even other dips in the Lake La Cocha temperature record are more pronounced. Additionally a relatively large increase in MAT of several °C at the onset of the Holocene within a few hundred years can not be distinguished in the Lake La Cocha MAT reconstruction while such is clearly recognizable in all other proxies, even that of the relatively nearby Huascaran glacier core.

Remarkable is the comparison with MAT reconstruction of the tropical Huascaran glacier core. There is a maximum absolute cross correlation of -0.51 with a time lag close to zero. Bush et al. (2007) also identified high variability of climatological trends with latitude in the tropical Andes. Within this region evidence for opposing climate trends has been found within relatively minor distances. Even locations that depicted similar climate trends over a certain period have shown evidence of opposing trends during another period (Bush et al., 2007). This could justify the high negative cross correlation found when comparing Lake La Cocha MAT to Huascaran, suggesting a reversed climate trend with an explained variance over 25%. However, the comparison with all ice core based MAT reconstructions of other regions, except maybe the Antartic EPICA ice core, depicts such low cross correlations, even when taken over the last 10 kyr, that some doubts about the suitability of AP% for MAT reconstruction seem justified in this case.

4.3 Human impact

Evidence of human presence during the last millennia of the Holocene has been reported from many sites in Colombia (Marchant et al., 2001). Human presence at higher elevations in the Andes has been found dating back until ~3000 yr BP (Marchant et al., 2001; Torres et al., 2006). In order to identify changes in dynamics, the affinity scores indicating agricultural activity where taken over the full period (Fig.15a). A logical step is to look for contemporaneous logging activity. In Figure 15d both human indicators are compared over the last 2.5 kyr BP. The only direct match is found at 963 yr BP, as agricultural indicators peak and logging trees dip.

The BAFs of mixed forest, cool evergreen forest and warm evergreen forest depict a sharp downfall in the last 1.4 kyr BP. The latter two depict a strong decrease that can not be easily explained otherwise than human influence in favor of logging and agriculture. Maybe even more remarkable are two revival peaks of the same signals within relatively short time (750 to 500

(29)

and 350 to 200 yr BP). The second revival peak could be caused by a strong decrease of human activity due to outbreak of plagues introduced by European immigrants, as suggested by Brandts (2013). The first one however can not, since there where no Europeans in South

America before the end of the 15th century.

Another explaination could be that the downfalls in the forest biomes in the last 1.4 kyr BP are caused by two separate fire events, that might possibly be caused by humans in favor of agriculture. This is supported by the detrended logging signal, that seems to suggest that the two downfalls and recoveries in this period concern AP in general rather than logging trees only. When comparing the detrended affinity scores for traditional logging trees (Fig. 16b) during the period in which human impact becomes relevant at higher Andean elevations to the non

detrended signal (Fig. 15d), there is a clearer and more consistent downfall after 2.1 kyr BP. The recovery after the zero value at 500 yr BP is not gradually but sudden, implicating that the recovery follows a similar trend as the AP signal. This is supported by similar trends in the MAT reconstruction, that has a linear relation with the AP signal. Therefore it seems likely that the absolute minumum at 500 yr BP is not caused by logging but by a more general event, like fire. Additionally it suggests that the AP% is not a fit indicator for the MAT reconstruction in this period. The suggestion of forest fires however in not supported by the charcoal record of Lake La Cocha by González-Carranza et al. (2012).

5. Conclusion

The hypothesis of a more dynamic Holocene climate in the neotropics is supported by the results of biomisation and the MAT reconstruction. Increase in BAFs for shrub páramo, cool mixed forest and cool evergreen forest often seem to indicate warming rather than cooling at the elevation of Lake La Cocha. Marking points of changing trends in BAFS are generally supported by changes in the MAT reconstruction. Lake La Cocha MAT resembles most to Antarctic EPICA MAT, when only the last 10 kyr are considered for better results. However the profile deviates considerably from all other MAT reconstructions. The high negative correlation with the relatively nearby Huascaran MAT reconstruction is remarkable. Additionally Lake La Cocha MAT lacks a clear MAT raise between the Lateglacial and the Holocene and a recognizable Younger Dryas. Although literature confirms the possibility of opposing climate tendencies within relatively short distance in the Andes, the question is how accurately the AP% reconstructs the MAT in the Lake La Cocha case. The fact that BAFS and MAT reconstruction seem in

accordance can not refute the doubt, since they are partially based on the same data. The combined indicators of human impact, agriculture and the detrended logging signal, suggest human impact since ~2.1 kyr BP. This is in accordance with reports from literature. The match around 963 yr BP between the logging signal and the agricultural indicators strengthens the evidence for human impact at Lake La Cocha. The logging signal indicates human activity after 1.6 kyr BP. Detrending the logging signal by the general arboreal pollen signal gives results that are in better accordance with reports on human activity in the Andes at higher elevations from previous studies, as only then it depicts a consistent downfall after 2.1 kyr BP.

(30)

Exceptional downfalls in the detrended logging signal the last millennium BP followed by fast recovery are remarable. However, the suggestion that these events might depict destruction and recovery after fires that possibly have been used by humans in favor of agricultural land or settlement, is not supported by the charcoal record of Lake La Cocha.

6. Acknowledgements

I would like to express my sincere gratitude towards my first supervisor, dr.ir. John van Boxel for facilitating my research by providing the necessary data as well as the software he wrote to process these data. Additionally John supplied me with suggestions how to progress, encouraged me in my research, and elaborately shared his expertise on the field of climatology and paleo ecology. I would also like to thank professor Henry Hooghiemstra, my second

supervisor, for kindly investing his time and expertise in answering my questions and supplying me with additional information and literature, even when located in Japan. Finally, I would like to express my appreciation for facilitating my research and supplying me with relevant information through the work done by Zaire González-Carranza et al. (2012) and Vladimir Torres et al. (2006). It is due to the efforts of these scientists that I have been granted the possibility to execute my research.

(31)

7. References

Bogotá-Angel, R. G., Groot, M. H. M., Hooghiemstra, H., Lourens, L. J., Van der Linden, M., & Berrio, J. C. (2011). Rapid climate change from north Andean Lake Fúquene pollen records driven by obliquity: implications for a basin-wide biostratigraphic zonation for the last 284 ka. Quaternary Science Reviews,30(23), 3321-3337.

Brandts, B.E. (2013). Biomisatie van de La Cocha kern. BSc Thesis Earth Sciences. University of Amsterdam, Faculteit der Natuurwetenschappen, Wiskunde en Informatica, 19 pp. Bush, M. B., Hanselman, J. A., & Hooghiemstra, H. (2007). Andean montane forests and

climate change. In Tropical rainforest responses to climatic change (pp. 33-54). Springer Berlin Heidelberg.

Feeley, K. J., Silman, M. R., Bush, M. B., Farfan, W., Cabrera, K. G., Malhi, Y., et al. (2011). Upslope migration of Andean trees. Journal of Biogeography, 38(4), 783-791.

Fricke, H. C., & O'Neil, J. R. (1999). The correlation between 18 O/16 O ratios of meteoric lateral and surface temperature: its use in investigating terrestrial climate change over geologic time. Earth and Planetary Science Letters,170(3), 181-196.

González-Carranza, Z., Hooghiemstra, H., & Vélez, M.I. (2012). Major altitudinal shifts in Andean vegetation on the Amazonian flank show temporary loss of biota in the Holocene. Holocene 22 (11), 1227–1241.

Groot, M. H., Hooghiemstra, H., Berrio, J. C., & Giraldo, C. (2013). North Andean

environmental and climatic change at orbital to submillennial time-scales: Vegetation, water levels and sedimentary regimes from Lake Fúquene 130–27ka. Review of Palaeobotany and Palynology, 197, 186-204.

Hooghiemstra, H., Berrio, J. C., Groot, M. H., Bogotá-Angel, R. G., Olivera, M. M., & González-Carranza, Z. (2012). The dynamic history of the upper forest line ecotone in the northern Andes. In Myster R.W. (Ed.), Ecotones between forest and grassland, Springer, New York, pp. 229-246.

Houghton, J. T. (2009). Global warming: the complete briefing. Cambridge University Press. IPCC (2001). Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the

Third Assessment Report of the Intergovernmental Panel on Climate Change [Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, X., Maskell, K., & Johnson, C.A. (Eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 881 pp.

IPCC (2013). Summary for Policymakers. In: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the

Intergovernmental Panel on Climate Change [Stocker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex & P.M. Midgley (Eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Jansen, B., De Boer, E. J., Cleef, A. M., Hooghiemstra, H., Moscol-Olivera, M., Tonneijck,

F. H., et al. (2013). Reconstruction of late Holocene forest dynamics in northern Ecuador from biomarkers and pollen in soil cores. Palaeogeography, Palaeoclimatology,

Palaeoecology, 386, 607-619.

Marchant, R., Boom, A., Behling, H., Hooghiemstra, H., Melief, B., Geel, B. V., et al. (2004). Colombian vegetation at the Last Glacial Maximum: a comparison of model‐and pollen‐based biome reconstructions. Journal of Quaternary Science, 19(7), 721-732.

(32)

Marchant, R., Behling, H., Berrio, J. C., Cleef, A., Duivenvoorden, J., Hooghiemstra, H., et al. (2001). Mid-to Late-Holocene pollen-based biome reconstructions for

Colombia. Quaternary Science Reviews, 20(12), 1289-1308.

Marchant, R., Boom, A., & Hooghiemstra, H. (2002). Pollen-based biome reconstructions for the past 450 000 yr from the Funza-2 core, Colombia: comparisons with model-based vegetation reconstructions. Palaeogeography, Palaeoclimatology, Palaeoecology, 177(1), 29-45.

Marchant, R., Cleef, A., Harrison, S. P., Hooghiemstra, H., Markgraf, V., Van Boxel, J. H., et al. (2009). Pollen-based biome reconstructions for Latin America at 0, 6000 and 18 000 radiocarbon years ago. Climate of the Past 5, 725-767.

Prentice, I. C., Cramer, W., Harrison, S. P., Leemans, R., Monserud, R. A., & Solomon, A. M. (1992). Special paper: a global biome model based on plant physiology and

dominance, soil properties and climate. Journal of biogeography 19, 117-134.

Ramirez, E., Hoffmann, G., Taupin, J. D., Francou, B., Ribstein, P., Caillon, N., et al. (2003). A new Andean deep ice core from Nevado Illimani (6350 m), Bolivia. Earth and

Planetary Science Letters, 212(3), 337-350.

Rodbell, D. T. (2000). PALEOCLIMATE: The Younger Dryas: Cold, Cold Everywhere? Science,

290(5490), 285-286.

Rolefes, S. M. F. (2015). Periodicity of altitidinal vegetation shift in the Colobian Andes. Spectral analysis if biome scores from the Fúquene pollen record during 284-27 ka reveals vegetation response to climate change. BSc Thesis Earth Sciences. University of Amsterdam, Faculteit der Natuurwetenschappen, Wiskunde en Informatica, 28 pp. Grootes, P.M., Stuiver, M., White, J.W.C., Johnsen, S.J., & Jouzel., J. (1993). Comparison of

oxygen isotope records from the GISP2 and GRIP Greenland ice cores. Nature 366, 552-554.

Thomas, E. R., Wolff, E. W., Mulvaney, R., Steffensen, J. P., Johnsen, S. J., Arrowsmith, C., et al. (2007). The 8.2 ka event from Greenland ice cores. Quaternary Science Reviews, 26(1), 70-81.

Thompson, L. G., Mosley-Thompson, E., Brecher, H., Davis, M., León, B., Les, D., et al. (2006). Abrupt tropical climate change: Past and present. Proceedings of the National

Academy of Sciences, 103(28), 10536-10543.

Thompson, L. G., Mosley-Thompson, E., Davis, M. E., Lin, P. N., Henderson, K. A., Cole-Dai, J., et al. (1995). Late glacial stage and Holocene tropical ice core records from Huascaran, Peru. Science, 269(5220), 46-50.

Torres, V., Hooghiemstra, H., Van Boxel, J.H., & Cleef, A.M. (2006). Biomisation-based evolution of North Andean Ecosystems from late Pliocene to latest Pleistocene time. Ch. 7 in: Torres, V., Pliocene-Pleistocene evolution of flora, vegetation and climate: A

palynological and sedimentological study of a 586-m core from the Bogota Basin, Colombia. Ph.D. Thesis Universiteit van Amsterdam, 181 pp.

Van Boxel, J.H., González-Carranza, Z., Hooghiemstra, H., Bierkens, M., & Vélez, M.I. (2014). Reconstructing past precipitation from lake levels and inverse modelling for Andean Lake La Cocha. Journal of Paleolimnology, 51 (1), 63-77.

Van Geel, B., & Van der Hammen, T. (1973). Upper Quaternary vegetational and climatic sequence of the Fuquene area (Eastern Cordillera, Colombia). Palaeogeography, Palaeoclimatology, Palaeoecology, 14(1), 9-92.

(33)

Velásquez-R, C. A., & Hooghiemstra, H. (2013). Pollen-based 17-kyr forest dynamics and climate change from the Western Cordillera of Colombia; no-analogue associations and temporarily lost biomes. Review of Palaeobotany and Palynology, 194, 38-49.

Wille, M., Hooghiemstra, H., Behling, H., van der Borg, K., & Negret, A. J. (2001).

Environmental change in the Colombian subandean forest belt from 8 pollen records: the last 50 kyr. Vegetation History and Archaeobotany, 10(2), 61-77.

(34)

8. Appendix A

Acron ym Old acr. Description Te-Tf tx Tree ferns

Te-Bt te2 Tropical evergreen broadleaved trees Td-Bt tr2 Dry tropical deciduous (raingreen) trees Tr-Fh tf Tropical forbs and herbs

We-Bt wte Warm temperate evergreen broadleaved trees and shrubs

Wc-Fh

tef Warm or cool temperate forbs and herbs

Ce-Ct ctc2 Cool temperate coniferous trees

Ce-Bt wte1 Cool temperate evergreen broadleaved trees and shrubs Ce-Xt wte4 Cool temperate evergreen sclerophyll trees and shrubs Dt-Xt Dry temperate sclerophyll trees and shrubs

Cd-Bt ts1 Cool temperate deciduous (summergreen or raingreen) trees and shrubs Al-Ds aa Alpine dwarf shrubs

Al-Fh af Alpine forbs and herbs Al-Cp cf Alpine cushion plants

Grass g Grasses

Table 4 provides an descriptive overview of the PFTs that can be distinguished in the area of study. The second column lists the acronyms as used in earlier publications.

(35)

Appendix B

TdBt Apocynaceae Bignoniaceae Cordia Croton Fabaceae Malpighiaceae

Malvaceae Psychotria

TeTf Anthurium Cyathea Dicksonia Lophosoria

TeBt Alchornea Anacardiaceae Apocynaceae Bignoniaceae Cecropia Clusia

Cordia Eugenia Fabaceae Ilex Malvaceae Mauritia/Mauritiella

Melastomataceae Meliaceae Myrtaceae Palmae Psychotria

Proteaceae Protium Sapium Urticaceae/Moraceae Vismia

TrFh Alternanthera Borreria Bromeliaceae Fabaceae Malvaceae Pilea

Sagittaria Sellaginella Thelypteris

WeBt Acalypha Alchornea Anacardiaceae Apocynaceae Cecropia Clusia

Cordia Croton Drimys Eugenia Fabaceae Fuchsia

Hedyosmum Heliocarpus Hyeronima Ilex Juglans Malvaceae

Melastomataceae Meliaceae Palmae Panopsis Proteaceae

Psychotria Quercus Salix Sapium Styloceras Symplocos

Urticaceae/Moraceae Viburnum Weinmannia

CdBt Myrteola Solanaceae

WcFh Alternanthera Amaranthaceae/Chenopodiaceae Borreria .

Brassicaceae Bromeliaceae Calceolaria Caryophyllaceae Cruciferae=Bras.

Fabaceae Geranium Gunnera Hypolepis Iridaceae

. Lachemilla Lathyrus Lupinus Malvaceae Muehlenbeckia

Pilea Plantago Puya Ranunculaceae Relbunium

Rosaceae Rosaceae-l Rosaceae_ Sagittaria Salvia Scrophulariaceae

Sellaginella Thalictrum Thelypteris Urticaceae/Moraceae Vicia

CeBt Alnus Anacardiaceae Bocconia Calceolaria Clethra Clusia Cordia

daphnopsis Drimys Ericaceae Escallonia Fuchsia Hedyosmum Ilex

Juglans Melastomataceae Miconia Oreopanax Palmae .

Quercus Rubus Salix Styloceras Symplocos Vallea

. Viburnum Vismia

. CeXt Aragoa Asteraceae Asteraceae_sub._Tub. Clethra Croton

Dodonaea Ericaceae Eugenia Gaiadendron

Hesperomeles Hypericum Melastomataceae Miconia

Monnina Myrica Myrsine Myrtaceae Polylepis/acaena

Proteaceae Rosaceae Rosaceae-l Rosaceae_ Satureja

Weinmannia

CeCt Podocarpus

Gras Poaceae

AlDs Aragoa Arcytophyllum Calceolaria Clethra

Ericaceae Gaiadendron Gentianaceae Hesperomeles Hypericum

Melastomataceae Miconia Monnina Myrica Satureja

(36)

AlFh Apiaceae Asteraceae Asteraceae_sub._Tub. Bartsia

Brassicaceae Bromeliaceae Calceolaria Caryophyllaceae Compositae=Aste.

Cruciferae=Bras. Draba Fabaceae Gentiana Gentianaceae

Geranium Hypolepis Iridaceae Jamesonia Lachemilla Lobelia_type

Lupinus Lycopodium Lysipomia Malvaceae Montia Muehlenbeckia Ottoa

Papaelanthus_/Eriocaulaceae plantago Puya Ranunculaceae Ranunculus

Relbunium rhizocephalum Rosaceae Rosaceae-l Rosaceae_ Salvia

Sellaginella Sisirynchum Umbelliferae=Apia. Valeriana Vicia

Lycopodium_foveolatum

AlCp Caryophyllaceae Plantago

NON Amarillidaceae Araliaceae Arecaceae Azolla Botryocuccus

Caesalpina Campanulaceae Cardulovica Cedrela Celtis Centropogon

Cestrum Cleidion Clusia-type/Clusiaceae

Convolvulaceae/Flacourt Coriaria Cuphea Cyperaceae Cybianthus

Debaraya Didimopanax Euphorbiaceae Evolvulus_(dry) Flacourtiaceae

Hydrocotyle Iriartea Isoetes Labiatae Liliaceae Loasea

Ludwigia Macrocarpea Maytenus Mimosa-casta Mimosa_cf._Pudica

Myriophyllum Myrsinaceae Myrtaceae Myrteola Nertera

Nothophagus Papaelanthus_/Eriocaulaceae Papillionaceae Passifloraceae

Pedastrium Phytolacca Polygonum Rhamnus Rubiaceae Sericotheca- .

argentea Snella Stylosanthes

Ugni Verbenaceae Zygnema

Tr1 Anacardiaceae Apocynaceae Bignoniaceae Fabaceae

Malvaceae

HuIm Allium Amaranthaceae Chenopodiaceae Amaranthaceae/Chenopodiaceae

Ambrosia Eucalyptus Mauritia Phaseolus Pinus Rumex

Schinus Zea mays

Logg Quercus Podocarpus Weinmannia

Table 5: Overview of the assignment of taxa to PFTs as applied in the biomisation of Lake La Cocha. Deviations from the Torres et al. (2006) interpretation have been made under

supervision of professor Hooghiemstra. These concerned assignment to PFTs of certain pollen taxa from the group that is not classified and assignment of taxa to the human indicators HuIm and Logg.

(37)

Appendix C

Biome PFT PFT PFT PFT PFT

CDSF DtXt, WcFh, Gras

TSFO DtXt, TdBt, TrFh

WEFO WeBt, TeBt, TeTf, TrFh

CEFO CeBt, CeXt, WcFh

COMI CeBt, CeXt, CeCt, WcFh, CdBt

CGSH AlDs, AlFh, Gras, CdBt

CGPR AlFh, AlCp, Gras

HuIm HuIm

Logg Logg

Table 6: Overview of the PFTs assigned to the different biomes in the biomisation of the Lake La Cocha core. In this research two biomes and equally named PFTs were added indicating respectively agricultural activity (HuIm) and logging activity (logg).

Biomisation of the Lake La Cocha pollen record Comparing an equatorial Colombian Andean Lateglacial-Holocene temperature reconstruction to ice core temperature proxies