Past human impact on vegetation in western Amazonia

Past human impact on vegetation in

western Amazonia

A phytolith reconstruction of the vegetation reconstruction

surrounding Lake Ayauch

_Ecuador

Britte Heijink 5/7/17 Supervisor: Dr. Crystal McMichael.

Abstract

A scientific debate is present about the extent and severity of the influence pre-Columbian people had on the Amazonian rainforest. Due to the long lifespan of trees, changes made by the pre-Columbian people can still be visible in the modern rainforest. This research investigates the effects humans had on the vegetation composition

surrounding Lake Ayauchi _{over the past 2000 years. Phytoliths were used to reconstruct} past vegetation. Phytoliths can expand pollen records since they can indicate agriculture, canopy openings, and palm abundances. Lake Kumpak has had very limited human influence but has the same environmental conditions as Lake Ayauchi_{. By comparing the} phytolith record of Lake Ayauchi _{to that of Lake Kumpak, the effect of humans on the} vegetation was assessed. The hypotheses that human presence increased palm and Poaceae percentages were investigated since many palm species produce edible parts or building materials and Poaceae can indicate canopy openings. Furthermore, it was

hypothesized that the difference in vegetation composition between Ayauchi _{and Kumpak} would increase when there was a bigger human influence at Ayauchi_{. Humans did}

increase the total grass percentages and the percentages of useful palm species. Humans possibly influenced the way environmental conditions affected the vegetation. No

evidence was found of a population collapse of the indigenous peoples as a consequence of the European arrival.

Introduction

People have influenced their surrounding environments all over the world. The pre-Columbian Amazonian rainforest was home to at least 8 million people, who all impacted their surrounding landscapes (Clement et al., 2015; Denevan, 2014). Their influence is however highly debated. It is possible that large parts of Amazonia were domesticated, but the radius of human influence around a settlement was probably limited (C. Levis et al., 2017; C. H. McMichael et al., 2012, 2014). Data on Amazonian plant communities is limited and possibly biased towards areas that are likely to have ancient human impacts (Feeley, 2015; C. N. H. McMichael, Matthews-Bird, Farfan-Rios, & Feeley, 2017).

Knowledge of human impact on vegetation composition is relevant to the conservation and research of the Amazonian forest. Trees in Amazonia can have a lifespan of a few hundred years, indicating that the current vegetation might still be in recovery from or is changed by human influences made in the pre-Columbian era (Clement & Junqueira, 2010; Junqueira, Shepard, & Clement, 2010; Carolina Levis et al., 2012). Species that were useful to the indigenous people are now five times more likely to be a

hyperdominant species than expected by chance (C. Levis et al., 2017; ter Steege et al., 2013). Humans distributed these useful species outside their original range, resulting in wild populations and (semi-) domesticated populations (C. Levis et al., 2017). The resulting anthropogenic forests still exist (Junqueira et al., 2010; Carolina Levis et al., 2012) (Junqueira et al., 2010; Carolina Levis et al., 2012).

The human population of Amazonia has fluctuated heavily. For the first 8000 years of human occupation in South-America, population numbers remained low. Agriculture has a long history in the Amazonian Basin. Humans have been domesticating plants since 8000 BP in Amazonia (Piperno, 2011; Purugganan & Fuller, 2009). Exponential growth in cultural hotspots started 5,500 years ago (Goldberg, Mychajliw, & Hadly, 2016). Evidence of 6900 years old maize was found at Lake Sauce and Lake Ayauchi _{(M. B. Bush et al.,} 2016a; Mark B. Bush & Colinvaux, 1988; Mark B. Bush, Piperno, & Colinvaux, 1989). After the Spanish invasion and conquest, starting with the discovery of the Americas by Columbus, indigenous populations collapsed, mostly due to war and sickness (Dobyns, 1966). Many human settlements were abandoned, inducing the illusion for early

naturalist explorers like Von Humboldt that the Amazonian rainforest was a pristine place (Hulme, Youngs, & Whitehead, 2002).

Phytoliths can be used as a proxy to detect cultivation and vegetation composition in paleoecological archives. They can complement pollen data because, in contrast to pollen, phytoliths can remain in sediments and soils for thousands to possibly millions of years (Morcote-Ríos, Bernal, & Raz, 2016; Piperno, 2006). Phytoliths show a very local signal and can be used to infer past vegetation or agricultural activities (Morcote-Ríos et al., 2016; Piperno, 2006).

The aim of this research is to assess to what extent humans have impacted the vegetation composition surrounding Lake Ayauchi_{, using phytoliths as a proxy. The} corresponding research question is: ‘What is the influence of humans on the vegetation

composition surrounding Lake Ayauchi _{for the past 2000 years?’. These past 2000 years}

contain information about the influence of the native population on its peak level as well as the collapse of this population due to the arrival of the Spanish. By comparing Lake Ayauchi _{to a nearby lake, Lake Kumpak, human-induced changes can be detected.} Climatic conditions are equal for both lakes because the lakes are located 20 km from each other. Recent work suggested that human influence around Lake Kumpak has been very limited (Palmeira, 2016; Vogel, 2016). Therefore, the only difference between the lakes is the difference in human influence.

The following hypotheses are discussed in this study. The first hypothesis is that

greater when human impact at Ayauchi _{is greater (See Figure 3). The number of maize} phytoliths will serve as a proxy for human impact. This hypothesis expects that the forest is not very resilient and that most human-induced change in the vegetation composition is permanent. It is expected that after the population collapse, human-induced changes will be reduced. The vegetation composition of Ayauchi _{and Kumpak will become similar if} the forest is highly resilient (See figure 1, blue line). The vegetation composition remains similar even though maize increases again in recent times. This is because there is maize agriculture and a human settlement at Kumpak since the 1960s (Palmeira, 2016). If the forest is not very resilient, I expect that the vegetation composition of Ayauchi _and Kumpak will remain different (See figure 1, red line).

Figure1. Conceptual figure of hypothesis 1. DCA values show no change in vegetation composition at Kumpak

(straight line). Dashed line at Ayauchi _{represents a scenario where humans did not have a significant effect on} the vegetation composition. The red line Ayauchi _{indicates that change in vegetation composition correlates} with the amount of human impact. The red line in the figure on the right represents the amount of maize through time, declining after the Spanish invasion and increasing again in modern time. The blue line in the middle figure represents the decline in difference between Kumpak and Ayauchi _{after the Spanish invasion.}

The second hypothesis is that the presence of humans, indicated by the percentage of maize phytoliths, correlates with an increase in palm and Poaceae phytoliths (See Figure 2). Palm abundance is expected to increase during human presence because many palm species are useful to people. Grasses are likely to increase during human presence because humans made clearings for agriculture. Palms are a natural component of the rainforest and are therefore expected to have a higher percentage than Poaceae species. The percentage of Poaceae is not expected to become zero because they occur in natural canopy openings.

Figure 2. Conceptual figure of hypothesis 2. Palm and Poaceae phytolith percentages are expected to increase

with an increase in human indicators (maize phytoliths). The blue line represents palm percentages, the green line represents Poaceae percentages.

Methods

Site description

Lake Ayauchi_{(3°2'42.93"S, 78°2'4.32"W) is located in southeast Ecuador, near the} border of Peru (See Figure 3). Ayauchi _{is surrounded by wet lowland forest and situated} at an elevation of 245 meters above sea level (masl). The site is located in close

proximity to Rio Santiago, which makes it a preferred site for pre-Columbian peoples (See Figure 4) (C. H. McMichael et al., 2012).

The sediment core was obtained in June 2016 by Dr. C.N.H. McMichael, using a universal corer. The universal corer was used to extract the sediments in 0.5 cm depth increments in the field to prevent sloshing or mixing of stratigraphic layers. The entire core was 82.5 cm in length. According to a preliminary age-depth model (See Appendix I, Äkesson, unpublished data), the core corresponds to roughly 2170 years.

Figure 3. Map indicating locations of Lake Ayauchi _{and Lake Kumpak. The red symbol points to Lake Ayauch}i_, the blue symbol points to Lake Kumpak. Figure modified from Google Maps.

Figure 4. Overview of Lake Kumpak and Lake Ayauchi_{. White dots are human settlements. Figure}

Collaboration

To answer the main research question and achieve a complete view of the history of Lake Ayauchi_{, a collaboration with Simon Noten, Eliane Bakker, and Christine Äkesson. The} role of this research within the overall research project is to analyze the phytolith record from the past 1000 years. Simon Noten will do a phytolith analysis of Ayauchi from 1000

BP until 2000 BP and Christine Äkesson is working on a pollen analysis covering the past 2000 years for both Lake Ayauchi and Lake Kumpak. Eliane Bakker has done a charcoal

analysis of Lake Ayauchi covering the past 2000 years. For comparison with Lake

Kumpak, previous research results from Mona Palmeira and Johnny Vogel were used (Palmeira, 2016; Vogel, 2016).

Sample preparation

Samples were prepared by Annemarie Philip, laboratory manager of the research group of Paleo-Ecology and Landscape Ecology at the University of Amsterdam. 140.000

microparticles with a diameter of 15.29 μm were added at the start of the procedure. The microparticles are added to simplify measurements of the size of the phytoliths and to determine the concentration of the phytoliths. Comparison of the phytolith concentrations between different samples provides insight into the phytolith production of plants during a specific period and therefore could indicate plant abundance. However, since tropical systems usually have a very high productivity, the phytolith concentration in this study most likely indicates deposition rates (C. McMichael, personal communication, May 5, 2017). Samples were treated with 33% H2O2 (hydrogen peroxide) to remove organic materials. In addition, the samples were treated with KMnO4 (potassium permanganate), eliminating any remaining organic material left.HCl 10 % (hydrochloric acid) was used to catalyze the reaction with H2O2 and to remove any lime in the sample. To separate

phytoliths from heavier material they are floated. CHBr3 (bromoform) with a volumetric density of 2.3 g/cm3_{was used in combination with a centrifuge (1.15 min, 4600 rpm) to} decant the samples. Naphrax was used as a mounting medium (A. Philip, personal communication, April 18, 2017).

In this study, the upper 34-47 cm of the 90 cm long core has been analyzed. This corresponds to roughly 500 years. 13 samples were analyzed at a 1.0 cm interval. The samples from the depth of 46 and 47 cm have been counted in both this research as well as in the research from Simon Noten. Double-counting ensured data quality and

accuracy. 250 phytoliths have been counted on every slide. The first 33 cm of the core contained too few phytoliths to count and were therefore discarded as uncountable.

Identification

Phytoliths were used to identify changes in the vegetation composition. Phytoliths are silica bodies that are produced by vascular plants, mainly in the leaves (Morcote-Ríos, Bernal, & Raz, 2016; Piperno, 2006). Phytoliths can be used to identify plants at family, genus or in some cases species levels (Morcote-Ríos et al., 2016; Piperno, 2006).

Phytoliths could be used to determine cultivation of maize, indicated by the presence of the characteristic maize phytoliths (Piperno, 2006). Phytoliths can differentiate savanna from forest communities (Bremond, Alexandre, Hély, & Guiot, 2005). The presence of early successional taxa and grasses can also indicate canopy openings (Piperno, 2006). Recent research by Morcote-Ríos et al., (2016) identified eight types of palm phytoliths that provide diagnostic information. These subtypes can help distinguish between palms on an ecological level and between palms that were useful or non-useful for the

indigenous peoples (Morcote-Ríos et al., 2016).

Identification will be done using Dolores R. Piperno’s book ‘Phytoliths: A comprehensive

guide for archaeologists and paleoecologists', complemented with a paper on palms by

Morcote-Ríos et al. (2016) and the phytolith-database from ArcheoSciencei_{. An Axiophot} photomicroscope with polarizing glass was used to distinguish phytoliths. Using an

objective lens with the magnification of 40x, a total magnification of 400x was achieved. This magnification was sufficient for the identification. A count sheet was used,

identifying phytoliths into 30 distinct categories (See Appendix II). These categories can be divided into three major groups: grasses, palms and other arboreal phytoliths.

Upon the identification of every different kind of phytolith, a clear picture was made as a reference for further identification. These pictures were also used to compare the

identification procedures with Simon Noten. Phytoliths that could not be identified were photographed, their location was noted, and they were given a description and name (See Appendix III).

Statistical analysis

RStudio (version 0.99.491) was used for statistical analysis. A Detrended

Correspondence Analysis (DCA) was performed using the ‘vegan’ package in RStudio (Oksanen et al., 2014). A DCA was used to order samples along an environmental gradient and shows the variation on those gradients in standard deviations from the community average. A DCA is an ordination method on which the gradient with the highest variation is shown on the first axis and the second highest variation on the second axis and so on (Palmer, n.d.; Gauch, 1982). Only phytolith morphotypes that had reached two percent or higher in at least once sample was included in the DCA. This method was used to ensure that very rare morphotypes did not skew the DCA.

Pearson product-moment correlation was used to determine the correlation between Zea

mays and other phytolith morphotypes. An α-value of 0.05 was used to determine

significance. R-package ‘ggplot2’ was used to create correlation plots and regression lines.

Results

30 phytolith morphotypes were found and divided into three categories, arboreal, palm, and grasses (See Appendix). Palms were excluded from the arboreal category. Six unknown/unidentifiable morphotypes were found (See Appendix …).

Stratigraphic diagram

Figure 5 shows the phytolith reconstruction of the past 2000 years. The diagram is divided in a post-Columbian and a pre-Columbian era.

Arboreal

Throughout the past 2000 years, arboreal phytoliths are the biggest group,

compromising 45% and 70% of the total phytoliths. Small rugose spheres are the most common arboreal phytoliths and increase around 650 yrs BP, but do not fluctuate heavily. Large rugose spheres increase around the same time as small rugose spheres but fluctuate more heavily throughout the reconstruction. Decorated spheres slowly decrease from 900 yrs BP and are at their lowest at the European arrival, but start to increase again after that. Heliconia has low abundances percentages and is not always present. At the start of the post-Columbian era, arboreal phytoliths start to increase. Over the whole 2000 years, arboreal phytoliths decrease when maize increases and increase when maize decreases (See Figure 5).

Maize

Zea mays percentages were lower in the post-Columbian era than in the pre-Columbian

era, with a sharp decrease after the transition from the pre-Columbian to post-Columbian zone. However, there is peak present during the transition from pre- to post-Columbian. (See Figure 5).

Grasses

Grasses are the second-biggest group in the entire reconstruction. They are less

except wavy trapezoids, fluctuate heavily, and most morphotypes, with the exception of other crosses, rondels and wavy trapezoids, have a dip between 1350 and 1600 yrs BP. Wavy trapezoids slowly start to decrease at 1150 yrs BP. The ratio between grass and non-grass (arboreal and palm) declines after 600 yrs BP (See Figure 5).

Palms

Palms increase towards the end of the pre-Columbian era. Palms are more abundant in the post-Columbian era than in the pre-Columbian era, expect for the globular echinate symmetrical palm phytoliths. The conical morphotype is most the most abundant palm type. The globular echinate elongate is sporadically abundant in low abundances between 2000 and 700 yrs BP. From 700 yrs BP to more recent times, it is much more abundant. It seems like there is more fluctuation in the amount of palms in the post-Columbian era than in the pre-Columbian era. However, the sampling resolution from 1000-2000 yrs BP is lower than in the post-Columbian era, thereby smoothing out possible fluctuations (See Figure 5).

Figure 5. Phytolith reconstruction of Lake Ayauchi _{showing the most abundant or important phytolith}

morphotypes. Phytoliths are depicted in percentages. Light gray shows post-Columbian era (post 409 yrs BP),

Detrended Correspondence Analysis

A DCA was performed to test the first hypothesis (See Figure 6 & 7). On the left side of axis 1, only samples older than 700 yrs BP are present (See Figure 6 and Appendix …). On the right side of axis 1, only samples younger than 750 yrs BP are present. The species on the negative side of axis 1 mostly represent phytoliths that are indicative of openness, like heliconia, maize, wavy trapezoids, tall and squat saddles (See Figure 7). On the positive side of axis 1, forest phytoliths like palm phytoliths and the small and large rugose sphere are present.

On the extreme negative side of axis 2, sites with a relatively high percentage of arboreal phytoliths (1;5;8;25). The morphotype globular echinate symmetrical is on the positive most extreme side of axis 2. Other phytoliths on the positive side of axis 2 are short rondels, conical palm, maize, and heliconia. On the other, positive side of axis 2, samples have a low percentage of arboreal phytoliths (9;13;27). There is only one phytolith on the extreme negative side of this axis, which is the large rugose sphere.

Sample Depth Age (in yrs BP)

1 34 168.1 2 36 175.0 3 37 217.0 4 38 258.7 5 39 300.1 6 40 341.2 7 41 381.9 8 42 421.9 9 43 461.9 10 44 501.8 11 45 541.7 12 46 581.4 13 47 661.5 14 48 701.8 15 49 741.8 16 50 781.6 17 51 822.6 18 52 863.3 19 53 904.1 20 54 944.8 21 55 985.1 22 56 1025.2 23 59 1145.8 24 61 1225.3 25 65 1386.9 26 70 1588.9 27 75 1790.8 28 80 1992.8

Table 1. Sample numbers with corresponding depth and age

The values for DCA axis 1 start to increase around 950 yrs BP. This increase continues in the post-Columbian zone. Axis 2 fluctuates more through time than axis 1. Its highest value is reached at the transition from pre-Columbian to post-Columbian. The values decrease after the transition. DCA axis 1 through time looks similar to the total palm percentages through time. DCA axis 2 looks similar to the total grass percentages over time.

Figure 6. Detrended Correspondence Analysis showing sample numbers. Numbers shown in the figure

correspond to different samples. Samples numbers are ordered from young to old. The youngest sample is number 1, the oldest sample is number 28.

Correlations

To test the second hypothesis that the presence of human indicators increases the palm and Poaceae percentages, a Pearson correlation test was executed for maize with the total palm percentage and the total grass percentage. A significant positive correlation was found between Zea mays and the total amount of grass phytoliths (See Figure 8). There was no significant relationship between Zea mays and the total palm percentage. However, a positive correlation between Zea mays and conical palm (morphotypes G and H from Morcote-Ríos et al., 2016) and between Zea mays and globular echinate

symmetrical palm (morphotype A from Morcote-Ríos et al., 2016) was present (See figure 9 & 10). Lastly, a negative correlation was found between Zea mays and small rugose spheres and between the total arboreal percentages (See Figure 11 & 12).

Figure 8. Scatterplot of the percentage maize and the percentage of the total grass phytoliths. Details of the Pearson

Figure 9. Scatterplot of the percentage of maize and the percentage of conical palm phytoliths. Details of the

Figure 10. Scatterplot of the percentage of maize and the percentage of globular echinate symmetrical palm phytoliths.

Details of the Pearson correlation test are included in the figure

Figure 11. Scatterplot of the percentage of maize and percentage of the total arboreal phytoliths. Details of the

Figure 12. Scatterplot of the percentage of maize and the percentage of small rugose spheres. Details of the Pearson

correlation test are included in the figure.

Discussion

The main research question of this research is: What is the influence of humans on the

vegetation composition surrounding Lake Ayauchi _{for the past 2000 years?’.The phytolith}

reconstruction indicated that humans increased the abundance of the total number of grass species and useful palm species, such as Iriartea, Bactridinae, and Geonomeae. Human presence decreases the abundance of arboreal species. People possibly have an influence in how climatic factors affect the vegetation.

Vegetation response

Maize is an indicator of human presence (Piperno, 2006). Multiple significant correlations between maize and other phytolith morphotypes indicate that humans had an influence on the vegetation surrounding Lake Ayauchi_{. As predicted in the first hypothesis, a} significant correlation between maize, a human indicator, and the total of Poaceae phytoliths exists. The increase in grasses at human presence could be explained by man-made clearings and human induced fires. Humans cleared land to practice agriculture and to build settlements (Clement et al., 2015). These land clearings would mean a decrease in arboreal phytoliths. A negative correlation between arboreal phytoliths and maize exists, indicating that people removed trees for agriculture.

Forests also comprise of palms. To clear land, palms have to be removed as well, but multiple palm species are useful to humans. Multiple species produce edible parts or can be used as building material.It has already been hypothesized that indigenous peoples enhance the numbers of useful palms. There is no correlation between the total of palm morphotypes, but positive relationship exists between maize and conical palm phytoliths and maize and globular echinate symmetrical phytoliths. The correlations indicate that humans had an effect on the surrounding vegetation.

Conical palm phytoliths are mostly produced by three tribes of palms, namely the Iriartea, Bactridineae, and Geonomeae (sub-)tribes (Morcote-Ríos et al., 2016). Multiple species in each (sub-)tribe can be classified as useful species. Globular echinate

symmetrical palms are produced by a variety of palm species, the majority of which have a human use (Morcote-Ríos et al., 2016). The positive correlation of these phytolith types with maize gives an indication that humans increased the total numbers of these useful species.

Figure 13. Phytolith reconstruction of Lake Kumpak and Lake Ayauchi_{. Summary graph of the total arboreal, total palm and}

total grass percentages of Lake Ayauchi _{and Lake Kumpak, maize percentage and charcoal abundance of Lake Ayauchi.}

Crosses indicate where samples for charcoal analysis were taken. DCA axis 1 from both Lake Ayauchi _{and Lake Kumpak are}

shown as well.

Charcoal

The presence of charcoal analyzed by Bakker (2017) coincides with relative higher abundance in grass abundances, around 1750 yrs BP, 1200 yrs BP, and 600 yrs BP (See Figure 13). However, between 450 yrs BP and 250 yrs BP, charcoal is present, but grass abundances decline. This period is preceded by the arrival of the Europeans (409 yrs BP).Charcoal is a human indicator in western Amazonia, due to the lack of naturally occurring fires (Mark B Bush et al., 2007). The increase in charcoal abundance after the European arrival does not coincide with a population decline as a result of the European arrival. It is yet unclear what causes this increase in charcoal.

Kumpak

Lake Kumpak and Ayauchi _{are similar in their climatic conditions. A phytolith}

reconstruction of Lake Kumpak has been made by Palmeira (2016) and Vogel (2016). By comparing the DCA from Lake Ayauchi _{to the DCA of Lake Kumpak, the human influence} on Ayauchi _{can be assessed. In figure 13, the values of DCA axis 1 were plotted in time} for both Lake Kumpak and Lake Ayauchi. A similar trend is for the DCA values is visible. However, the values for Lake Kumpak are considerably higher than the values for Ayauchi_{. A possible explanation for this is that DCA axis 1 for both lakes is driven by the} same environmental gradient, most likely a climatic factor, but that humans influenced the way the environmental gradient manifested itself on the vegetation composition surrounding Lake Ayauchi. There does not seem to be a greater difference between the values for DCA axis 1 for both lakes when maize is present at Ayauchi_{. This probably} indicates that human population size did not affect the way that humans had an influence on the effect of the environmental gradient represented by DCA axis 1.

However, the phytolith identification for Kumpak and Ayauchi _{were different. The main} difference lies in the identification of palm phytoliths. For Ayauchi_{, eight different kinds of} palm phytoliths were recognized. For Kumpak, only a distinction between two different kinds was reached. For better comparison, the identification level should be similar.

Influence Spaniards

Since maize and the total grass phytoliths are positively correlated, almost no grass phytoliths were found at Lake Kumpak (Palmeira, 2016; Vogel, 2016), and maize is an indicator for human presence, it is expected that grass abundances would decrease after the population crash caused by the European arrival. However, it is clear that grass abundances already start to decline at 550 yrs BP, thus before the European arrival. This could indicate that a population decline was already present before the Spanish conquest. Another explanation might be that a cultural shift in the way the indigenous people

cultivated their lands and thereby possibly changing their effect on the surrounding vegetation (M. B. Bush et al., 2016b).

Conclusion

The continuous presence of maize phytoliths throughout the last 2000 years indicates that humans were constantly occupying the land surrounding Lake Ayauchi_{. Humans} likely increased the total amount of grasses and useful palm species. They possibly changed the effect of climatic change had on the surrounding vegetation composition of Lake Ayauchi _{throughout the time period.}

The difference between the vegetation composition of Kumpak and Ayauchi _{can be} explained by the human presence at Lake Ayauchi_{. It has been proposed that the} interaction between nature and the human influence has caused the exceptionally high biodiversity of the Amazonian rainforest. Humans distributed useful species, resulting in wild populations and (semi-) domesticated populations, thereby influencing alpha-diversity (Levis et al., 2017). This could indicate that the alpha-alpha-diversity at Lake Ayauchi is higher than at Lake Kumpak. Further field research could confirm or deny this. In

summary, humans had a significant influence on the vegetation composition surrounding Lake Ayauchi_.

Acknowledgements

I would like to thank Crystal McMichael for being my supervisor, answering all of my questions, and for helping me with the phytolith identification. Secondly, I would like to thank William Gosling for being my second supervisor. Lastly, I wish to thank Simon Noten and Lorenzo Turk for the help during the phytolith identification.

References

Bakker, E. (2017). Fire history at Lake Ayauchi. University of Amsterdam

Bremond, L., Alexandre, A., Hély, C., & Guiot, J. (2005). A phytolith index as a proxy of tree cover density in tropical areas: Calibration with Leaf Area Index along a forest-savanna transect in southeastern Cameroon. Global and Planetary Change, 45(4), 277–293. https://doi.org/10.1016/j.gloplacha.2004.09.002

Bush, M. B., & Colinvaux, P. A. (1988). A 7000-year pollen record from the Amazon lowlands, Ecuador. Vegetatio, 76(3), 141–154. https://doi.org/10.1007/BF00045475 Bush, M. B., Correa-Metrio, A., McMichael, C. H., Sully, S., Shadik, C. R., Valencia, B. G.,

… Overpeck, J. T. (2016a). A 6900-year history of landscape modification by humans in lowland Amazonia. Quaternary Science Reviews, 141(June), 52–64. https://doi.org/10.1016/j.quascirev.2016.03.022

Bush, M. B., Correa-Metrio, A., McMichael, C. H., Sully, S., Shadik, C. R., Valencia, B. G., … Overpeck, J. T. (2016b). A 6900-year history of landscape modification by

humans in lowland Amazonia. Quaternary Science Reviews, 141, 52–64. https://doi.org/10.1016/j.quascirev.2016.03.022

Bush, M. B., Piperno, D. R., & Colinvaux, P. A. (1989). A 6,000 year history of Amazonian maize cultivation. Nature, 340(6231), 303–305.

Bush, M. B., Silman, M. R., de Toledo, M. B., Listopad, C., Gosling, W. D., Williams, C., … Krisel, C. (2007). Holocene fire and occupation in Amazonia: records from two lake districts. Philosophical Transactions of the Royal Society of London. Series B,

Biological Sciences, 362(1478), 209–18. https://doi.org/10.1098/rstb.2006.1980

Clement, C. R., Denevan, W. M., Heckenberger, M. J., Junqueira, A. B., Neves, E. G., Teixeira, W. G., & Woods, W. I. (2015). The domestication of Amazonia before European conquest. Proceedings. Biological Sciences / The Royal Society,

282(1812), 20150813. https://doi.org/10.1098/rspb.2015.0813

Clement, C. R., & Junqueira, A. B. (2010). Between a Pristine Myth and an Impoverished Future. Biotropica, 42(5), 534–536.

https://doi.org/10.1111/j.1744-7429.2010.00674.x

Denevan, W. M. (2014). Estimating Amazonian Indian numbers in 1492. Journal of Latin

American Geography, 13(2), 207–221. https://doi.org/10.1353/lag.2014.0036

Dobyns, H.F. (1966) Estimating Aboriginal American Population I: An Appraisal of Techniques with a New Hemispheric Estimate. Current Anthropology, 7, 395-416. Feeley, K. (2015). Are we filling the data void? An assessment of the amount and extent

of plant collection records and census data available for tropical South America.

PLoS ONE, 10(4). https://doi.org/10.1371/journal.pone.0125629

Gauch, J.H.G. (1982) Multivariate Analysis in Community Ecology. Cambridge University Press, Cambridge.

Goldberg, A., Mychajliw, A. M., & Hadly, E. A. (2016). Post-invasion demography of prehistoric humans in South America. Nature, 532(7598), 232–235.

https://doi.org/10.1038/nature17176

Hulme, P., Youngs, T., & Whitehead, N. L. (2002). South America / Amazonia: the forest of marvels, 122–138.

Junqueira, A. B., Shepard, G. H., & Clement, C. R. (2010). Secondary forests on

anthropogenic soils in Brazilian Amazonia conserve agrobiodiversity. Biodiversity and

Conservation, 19(7), 1933–1961. https://doi.org/10.1007/s10531-010-9813-1

Levis, C., Costa, F. R. C., Bongers, F., Pena-Cardos, M., Clement, C. R., Junqueira, A. B., … Ter Steege, H. (2017). Persistent effects of pre-Columbian plant domestication on

Amazonian forest composition, 931(March), 925–931.

Levis, C., de Souza, P. F., Schietti, J., Emilio, T., Pinto, J. L. P. da V., Clement, C. R., & Costa, F. R. C. (2012). Historical Human Footprint on Modern Tree Species

Composition in the Purus-Madeira Interfluve, Central Amazonia. PLoS ONE, 7(11). https://doi.org/10.1371/journal.pone.0048559

McMichael, C. H., Bush, M. B., Piperno, D. R., Silman, M. R., Zimmerman, a. R., & Anderson, C. (2012). Spatial and temporal scales of pre-Columbian disturbance associated with western Amazonian lakes. The Holocene, 22(2), 131–141. https://doi.org/10.1177/0959683611414932

McMichael, C. H., Palace, M. W., Bush, M. B., Braswell, B., Hagen, S., Neves, E. G., … Czarnecki, C. (2014). Predicting pre-Columbian anthropogenic soils in Amazonia.

Proceedings of the Royal Society, (January), 2–9.

https://doi.org/10.1098/rspb.2013.2475

McMichael, C. N. H., Matthews-Bird, F., Farfan-Rios, W., & Feeley, K. J. (2017). Ancient human disturbances may be skewing our understanding of Amazonian forests.

Proceedings of the National Academy of Sciences, 201614577.

https://doi.org/10.1073/PNAS.1614577114

Morcote-Ríos, G., Bernal, R., & Raz, L. (2016). Phytoliths as a tool for archaeobotanical, palaeobotanical and palaeoecological studies in Amazonian palms. Botanical Journal

of the Linnean Society, 182(2), 348–360. https://doi.org/10.1111/boj.12438

Noten, S. (2017). Pre-Columbian influence of an Amazonian lake before population

collapse. University of Amsterdam

Oksanen, J., Blanchet, F. G., Kindt, R., Legendre, P., Minchin, P. R., Hara, R. B. O., … Stevens, M. H. H. (2014). Package “ vegan .”

Palmeira, M. (2016). Post-colonial vegetation change in an undisturbed Amazonian

forest. University of Amsterdam.

Piperno, D. R. (2006). Phytoliths: a comprehensive guide for archaeologists and

paleoecologists. Rowman Altamira.

Piperno, D. R. (2011). The Origins of Plant Cultivation and Domestication in the New World Tropics. Current Anthropology, 52(S4), S453–S470.

https://doi.org/10.1086/659998

Purugganan, M. D., & Fuller, D. Q. (2009). The nature of selection during plant domestication. Nature, 457(February), 843–848.

https://doi.org/10.1038/nature07895

ter Steege, H., Pitman, N. C. A., Sabatier, D., Baraloto, C., Salomão, R. P., Guevara, J. E., … Silman, M. R. (2013). Hyperdominance in the Amazonian tree flora. Science,

342(6156), 1243092. https://doi.org/10.1126/science.1243092

Vogel, J. D. (2016). TESTING THE EFFECTS OF HUMAN IMPACT AND CLIMATE CHANGE

ON THE VEGETATION OF LAKE KUMPAK , ECUADOR 600-2000 BP. University of

Appendices

Appendix I Age-Depth model Ayauchi

Preliminary ages Ayauchi

depth

wmean

min

max

median

0.5

-60

1.5

-53.0882

2.5

-46.1765

3.5

-39.2647

4.5

-32.3529

5.5

-25.4412

6.5

-18.5294

7.5

-11.6176

8.5

-4.70588

9.5

2.205882

10.5

9.117647

11.5

16.02941

12.5

22.94118

13.5

29.85294

14.5

36.76471

15.5

43.67647

16.5

50.58824

17.5

57.5

18.5

64.41176

19.5

71.32353

20.5

78.23529

21.5

85.14706

22.5

92.05882

23.5

98.97059

24.5

105.8824

25.5

112.7941

26.5

119.7059

27.5

126.6176

28.5

133.5294

29.5

140.4412

30.5

147.3529

31.5

154.2647

32.5

161.1765

33.5

168.0882

34.5

175

12.3

305.5

187.7

35.5

217

35.8

352.2

222.1

36.5

258.7

57.3

426.7

257.1

37.5

300.1

78.5

521.6

291.8

38.5

341.2

99.3

625.6

325.1

39.5

381.9 140.8

695.3

361.2

40.5

421.9 185.9

726

403.7

41.5

461.9 212.9

767.9

445.8

42.5

501.8 234.7

829.6

485.1

43.5

541.7 255.5

889.7

523.6

44.5

581.4 294.1

938.1

564.4

45.5

621.2 340.4

965.2

606.3

46.5

661.5 377.4

1006.7

649.6

47.5

701.8 407.2

1055.4

691.9

48.5

741.8 434.9

1116.4

732.9

49.5

781.6 479.7

1163.6

770.4

50.5

822.6

533

1192.2

808.4

51.5

863.3 573.9

1227.1

848.6

52.5

904.1 609.4

1262.8

891.1

53.5

944.8 638.1

1305.2

935.1

54.5

985.1

692

1343.1

973.3

55.5

1025.2 758.6

1364.3

1008.4

56.5

1065.8 810.4

1385.9

1043.1

57.5

1105.8

846

1413.6

1083.2

58.5

1145.8 872.1

1474.7

1122.2

59.5

1184.9 896.6

1537.2

1159.4

60.5

1225.296

61.5

1265.692

62.5

1306.088

63.5

1346.484

64.5

1386.88

65.5

1427.276

66.5

1467.672

67.5

1508.068

68.5

1548.464

69.5

1588.86

70.5

1629.256

71.5

1669.652

72.5

1710.048

73.5

1750.444

74.5

1790.84

75.5

1831.236

76.5

1871.632

77.5

1912.028

78.5

1952.424

79.5

1992.82

80.5

2033.216

81.5

2073.612

82.5

2114.008

Extrapolated values

Appendix II Countsheet

Phyotlith Counts

number:

Analyst/date analyzed:

Non-grass types habitat type

large rugo 15-20um Forest

sm rugo ~10um F cranate spheres F decorated F Marantaceae F Annonaceae F Heliconia* F

Gobu EchiSymmetri Palm

GlobuEchi 13.7um P SmGlobEchiElo P BigGloEchiShortProj 22um P Reniform Echi 7.8 um P GloEchiLongAcPro P palm-conical P GlobuEchiAsym P other arboreal F Moonface * (disturbed)

Subtotals:

bilobate OF = Open forest

BiloLongShaft OF

tall saddle-bilobate OF/CF

tall saddle OF/CF

other bamboo OF/CF

squat saddle OF

collapsed saddle OF/CF

2 Spiked Irregular OF/CF

cross-Bam OF/CF cross-Pani OF rondels CF = Closed Forest short rondels OF long rondels CF Ehrh Left Ehrh mid Ehrh Right Zea mays

_*

Subtotals:

Microsphere Markers

Totals:

Grass/ non grass

Notes:

Appendix III Unknown phytoliths

Figure 14. Unknown phytolith 1

Figure 16. Unknown phytolith 3

Figure 19. Unknown phytolith 6