TESTING THE EFFECTS OF
HUMAN IMPACT AND
CLIMATE CHANGE ON THE
VEGETATION OF LAKE
KUMPAK, ECUADOR
600-2000 BP
Author: J.D. Vogel
Supervisor: mw. dr. C.N.H. McMichael Second supervisor: dr. ir. E.E. van Loon University of Amsterdam
Subject: Testing the effects of human impact and climate change on the vegetation of Lake Kumpak, Ecuador 600-2000 BP
Course: Bachelorproject aardwetenschappen Word count: 5807
2
Abstract
A sedimentary record from a unique Western Amazonian lake provided a phytolith record reflecting changes in vegetation between 600 and 2000 BP. Piston cores were collected from Lake Kumpak, a 19.5 meter deep lake at 330 meter a.s.l. in the aseasonal tropical rainforest of Ecuador. The phytolith reconstruction was compared with a charcoal record and a stable isotope record. Charcoal, a proxy for human-induced fire, was not found in the studied time window. Climate seemed to have some effect on the phytolith record, but this effect was not consistent enough to be significant. Evidence for a wetter phase between 1000-1400 BP was derived from the increased palm abundance during this phase. This observation conflicted with earlier research. Arboreal trees showed a declining trend over the time window. This might reflect a slower response to a climate which became gradually wetter over the last 4000 years. Grasses showed no consistent relation with climate but appeared to be less abundant in the wet phase proposed in this paper. Without any evidence for human disturbance, Lake Kumpak offers a unique possibility to reconstruct a high-resolution paleoecological history of an undisturbed rainforest.
Introduction
Western Amazonia is considered one of the major tropical wilderness areas of the world (Mittermeier, Myers, Thomsen, da Fonseca, & Olivieri, 1998). The region has attracted a lot of attention from researchers over the last decades. Researchers have been using the area as a test bed for several ecological theories (Hubbell, 2001; Whittaker, R.H.,Levin, S.A., Boot, 1973). Some of these theories assume that the forest ecosystem is in equilibrium. However, this assumption has developed into a topic of scientific debate. Researchers have suggested that human impact has been considerable prior to the arrival of Columbus to the Americas referring to Amazonia as a manufactured landscape (Clement et al., 2015; Heckenberger et al., 2003). Evidence for human impact in some parts of Amazonia is quite
convincing, while for other parts of the Amazonian rainforest, their presence has not been verified (Levis et al., 2012; McMichael et al., 2012b; McMichael et al., 2015). This research focuses on Lake Kumpak, Ecuador. From this lake a phytolith reconstruction ranging from 600 BP to 2000 BP is presented.
3 The alpha diversity of Western Amazonia is amongst the highest measured on Earth (Pitman, 2000). Possible explanations are diverse and often look at different spatial scales (Willis et al., 2002). The Intermediate Disturbance Hypothesis is one of the major ideas in ecology put forward to explain
differences in biodiversity at a landscape scale, basically stating that biodiversity is highest at intermediate levels of disturbance (Connell, 1978). Human impact has been suggested as one of the possible
mechanisms creating ‘intermediate disturbances’ and may therefore improve biodiversity (Barlow, Gardner, Lees, Parry, & Peres, 2012; Bongers, Poorter, Hawthorne, & Sheil, 2009; Heckenberger, Russell, Toney, & Schmidt, 2007). Ideas how ecosystems recover from disturbances are fundamental in the way that we think about the resilience of tropical rainforests and what role human may have played in creating this very diverse ecosystem. In this research potential human impact on the Lake Kumpak watershed is discussed using Human Index Scores as a tool for comparative analysis (McMichael et al., 2015). In addition to that, two groups of vegetation, palms and grasses, are compared with a charcoal record and a climate proxy to investigate vegetation responses to these two predictor variables.
4 Palms have been chosen in this study because relatively many palm species are considered to be
hyperdominant species (ter Steege et al., 2013). Two potential mechanisms to explain this property of the palm family are put forward. Because many palm species are useful to humans (Zambrana et al., 2007), certain palm species might have been promoted over time (Balée, 2013). For example by sparing palm species when a forest is cleared with fire (Byg & Balslev, 2006). Because natural fires in the aseasonal rainforest of Western Amazonia are considered to be extremely rare, charcoal found in lake sediments is a strong proxy for human-induced fire (Bush et al., 2007a). By comparing the amount of charcoal with relative palm phytolith abundance the effect of human-induced fire on palm abundance is tested. The expectation is that relative abundance of palm phytoliths increases after any peaks in charcoal abundance. Another hypothesis is that palm abundance increased as an effect of increased moisture availability in the Amazon rainforest. This hypothesis is tested by comparing a stable isotope record (van Breukelen, Vonhof, Hellstrom, Wester, & Kroon, 2008). Both hypotheses are visualized in figure 2.
Figure 2. Left: hypothetical situation were the % of palm phytoliths increases after a peak in charcoal abundance. Right: hypothetical situation were palm phytoliths gradually increase over time as stable isotope levels drop.
Please note that all values in this figure have no meaning and are only conceptual visualizations
Grasses and bamboo are considered early pioneer species in the ecological succession of tropical rainforests (Araújo et al., 2005; McMichael et al., 2013) Certain areas of the Amazonian rainforest are dominated by bamboo stands. Basically two hypotheses are put forward in this paper to explain bamboo dominated forests. The first hypothesis proposes that human induced fire has been responsible for the spread of bamboo forests. This would mean that the fraction of bamboo phytoliths would increase shortly after a fire event which is obviously true as well for other grasses and herbs. The second hypothesis is that bamboo spreads after periods of drought (McMichael et al. 2013). This is true as well for grasses as drought leads to higher tree mortality opening up the forest floor for successional species (Barlow, Peres, Lagan, & Haugaasen, 2002). Because grasses and bamboo are predicted to behave similarly, the
5 Figure 3. Left: hypothetical situation were the % of bamboo/grass phytoliths increases after a peak in charcoal abundance. Right: hypothetical situation were grass phytoliths gradually decrease over time as stable isotope levels drop.
Please note that all values in this figure have no meaning and are only conceptual visualizations.
Together, the four hypotheses test which situation fits the vegetation history better. Can human impact explain changes in the vegetation of the Lake Kumpak watershed or does climate prove to be a better predictor? As Western Amazonia is considered as one of the major wilderness areas, protection of this ecosystem seems desired. But what if human disturbance has been a component shaping this highly diverse ecosystem? In such a case overprotecting the rainforest might be harmful to the ecosystem. On the other hand, if we conclude that human impact over time had a positive impact on the functioning of the Amazon rainforest and we decide that the tropical rainforest is depended on some human
disturbance, we might devastate the rainforest (Bush & Silman, 2007). Therefore, the extent and effect of human disturbance must be understood properly. This thesis aims to improve our understanding of this.
Methods
Methods: Site description
Lake Kumpak (2°50'9.92" S, 77°57'43.85"W) is a closed volcanic basin in the province of Morono-Santiago, Ecuador. The lake has a surface elevation of 330 m a.s.l, has a surface area of 1.29 km2 and a maximum depth of 19.5 meter (Colinvaux, Miller, Liu, Steinitz-Kannan, & Frost, 1985). The lake only receives water from narrow inlet streams entering on all sides transporting young quartz-rich volcanic material from the catchment area to the lake bottom. Especially during high-energy events such as strong precipitation events sediment transport is considerable. The absence of outlets suggests that the water level of Lake Kumpak responds to changes in evapotranspiration (Colinvaux et al., 1985; Liu & Colinvaux, 1989) Lake Kumpak is located 25 kilometer North of Lake Ayauchi were a high to moderate impact by humans has been reported (Bush, Piperno, & Colinvaux, 1989; McMichael et al., 2012). Lake Ayauchi is known to be the site where the earliest use of maize agriculture in the Amazon basin has been reported (Bush et al., 1989). Earlier research has also located a single maize pollen at 1550 BP in Lake Kumpak (Liu & Colinvaux, 1989). The original sediment core revealed that Lake Kumpak, due to high sedimentation rates, offered a high-resolution sedimentary archive spanning over 5215 years. Lake Kumpak was cored in 2014 by Crystal
6 McMichael and Mark Bush using a Colinvaux-Vohnout piston corer. The original sediment core has
accidently been destroyed.
Climate data from remote sensing show that Lake Kumpak and Lake Ayauchi receive on average 3600-3700 mm precipitation per year, have a mean annual air temperature of 21-24 degrees Celsius and no prolonged dry season (figure 4). The area is located on the edge of the zone of influence of the South American Summer Monsoon which explains the lack of a dry season (Vuille et al., 2012)
The aseasonal rainforests of Ecuador are known to have high values for alpha and gamma diversity and low values for beta diversity (Pitman, 2000). Ecuadorian rainforest are dominated by trees, treelets, lianas and shrubs complemented by a weak herbaceous and epiphytic community (Pitman, 2000). The forest canopy is generally around 30 meter tall, sometimes broken by tree stems reaching 20 meter above the canopy. Endemism has been reported to be as low as 2-3% (Balslev, 1988).
Figure 4: Three different climate maps derived from Remote-Sensing data. The maps show the difference in climate for the two lakes sites (Kumpak and Ayauchi) and the speleothem record Cueva del Tigre. Left: Annual mean temperature, middle: Annual precipitation and right: Seasonality measured in coefficients of variation (Deblauwe et al., 2016).
Methods: Core sampling
In total phytoliths were counted for 54 samples covering a time window from 536 BP to 2018 BP corresponding to 3.95 meter of sediment. Between 536 and 979 BP a sample was taken every 5
centimeter. Between 979 and 2018 BP a sample was taken every 10 centimeters. Per slide 200 phytoliths were counted.
Methods: Phytolith extraction
Phytoliths were extracted for each sample. From the soil, clay was removed using pyrophosphate (P2O7
4-). The remaining soil was sieved through nested 212 micron sieves. Organic matter, humic acids and carbonates were removed from the samples by boiling the samples in 30% hydrogen peroxide (H2O2)
solutions. Phytoliths were separated from the remaining sediment using a heavy liquid, bromoform (CHBR3), having a specify density of 2.9 g/mL. The phytolihts were mounted on microscope slides using
Naphrax as mounting medium.
Methods: Phytolith identification
Phytoliths were identified using a Differential Interference Contrast microscope (DIC) with a magnification of 40. Based on phytolith morphotypes, phytoliths were grouped into arboreal trees, palms and
7 diagnostic phytolith morphotypes, namely as spiked, spherical phytoliths or as hat shaped, conical
phytoliths (Piperno, 2006). Bamboo can be separated from other grass types as bamboo constructs saddle-shaped phytoliths (McMichael et al, 2013). In this research, bamboo and other grasses are
evaluated together because grasses and bamboo are assumed to respond similar to the hypotheses. More information on phytoliths including pictures can be found in the appendix.
Methods: Human Index Scores
Human Index Scores were presented by McMichael et al. (2015) as a tool for comparative analysis between different sites in Amazonia. The calculation is based upon direct and indirect proxies for human activity. Direct proxies were archeological or palynological artifacts left by former inhabitants of an area. The artifacts taken into account are: anthropogenic soils (TP) and agricultural pollen such as maize (AG). These two artifacts are considered to be direct proxies for human activity and were therefore assigned a weight of 3. Indirect proxies used were charcoal and phytolith data. Presence of charcoal is assigned a weight of 2 because evidence of humans using fire to clear a forest is convincing but charcoal presence can also be attributed to regional transportation or natural fires (Whitlock & Larsen, 2002). Changes in the phytolith composition are assigned a weight of 1. This is caused by the assumption that many factors besides human activity may affect the phytolith composition over time.
Methods: Stable isotope record
Vegetation change was compared with a stable isotope record from Cueva del Tigre Perdido in the Peruvian district San Martín. The cave is located at an altitude of circa 1000 m. a.s.l. (van Breukelen et al. 2008). Cueva del Tigre Perdido is located 360 km SSE of Lake Kumpak. Climate might change abruptly over such distances, but climatic oscillations known to affect Southern American climate are thought to have a similar effect on both sites (Flantua et al., 2016). Low isotope values represent wetter and warmer climates while high isotope values represent cooler and drier climates.
Statistical procedures: phytoliths
In all statistical tests applied in this thesis separated tests ran for both sampling distances identified by the words ‘begin’ or ‘end’. 200 phytoliths were counted per slide in order to reduce the statistical impact of a couple of potential miscounts. Any outliners were recounted and adjusted if the outliner was not
reproducible.
Human Index Scores derived from the phytolith record were obtained by calculating values for two indicators. These indicators are adapted from McMichael et al. (2015). The first indicator (trend) is indicative for the direction of change over the timeframe. This value is obtained by subtracting the relative percentage per group for the uppermost sample from the relative percentage per group for the lowermost sample. The second indicator (delta) is indicative for the maximum variation within the relative phytolith groups.
𝑇𝑅𝐸𝑁𝐷 = 𝑥𝑇=536𝐵𝑃− 𝑥𝑇=2017𝐵𝑃 𝐷𝐸𝐿𝑇𝐴 = 𝑥𝑀𝐴𝑋− 𝑥𝑀𝐼𝑁
Statistical procedures: Interpolation of data
Interpolation was required because the predictor values were sampled at different time intervals. All predictors were interpolated to obtain values corresponding to the age of the phytolith samples. As the age of the phytolith samples were derived from an age-depth model, the phytolith ages were also
8 interpolated increasing uncertainty in an unquantified way. The stable isotope ratios obtained by van Breukelen et al. (2008) were interpolated using a spline interpolation method (see appendix). The original stable isotope ratios were obtained from a high resolution sampling with time steps of roughly 15 years. Such a high-resolution sampling reduces the chance of obtaining flawed data by interpolating. Because the age of the stable isotope sampling points was determined using U-Th radiometric dating, the calculated age of these data points were assumed to be accurate.
Statistical procedures: Spearman correlation coefficients
The statistical relevance of the predictor values on the phytolith assemblage were calculated using a spearman correlation coefficient. This is a nonparametric function assuming a monotonic relation
between the predictor and response variable without assuming linearity. With a Spearman correlation the null hypothesis stating that predictor and response variable are independent was tested against the alternative hypothesis that a consistent relation exists between predictor and response variable. Because the spearman correlation coefficient is sensitive to large jumps in the data, the data was normalized by calculating a best-fitted 9th order polynomial fitting through the response variables. The values obtained
from the 9th order polynomial were subtracted from the original values. Spearman correlation coefficients
were calculated for these normalized values. A 9th order polynomial was chosen because lower order polynomials were not able to calculate a best-fit line through the data were variance was small such as grasses. The choice for a 9th order polynomial instead of a lower order polynomial did yield similar results than lower order polynomials.
Results
Human Index Scores
The calculations for the Human Index Scores were based on McMichael et al. (2015). In figure 5 the Human Index Score calculated for Lake Kumpak is compared with other Amazonian sites. Lake Kumpak is one of the least impacted sites accounting to this calculation although phytolith-derived values (dESH, dP and tAR) seem higher than phytolith-derived values for similarly impacted sites.
9
Phytoliths compared with a stable isotope record
Figure 6: The three phytolith groups plotted together with the stable isotope record. Some graphs are further differentiated in the appendix.
Figure 6 shows the relation between the three main phytolith groups and stable isotope levels. Stable isotope levels are indicated by the blue line while phytoliths are indicated by the red line. The left axis corresponds to isotope values and the right axis to phytolith percentages. Palm phytolith seem to be more abundant towards to present while arboreal trees show a downward trend. No clear trend was observed for the grass phytoliths. Climate seems to have a clear effect on palm phytolith between 1400 and 1100 BP.
Spearman coefficients between phytoliths and 18O ratios and
phytoliths
Spearman correlation coefficients were calculated to evaluate whether a consistent reaction existed between the predictor variable (stable isotope ratios) and response variable (phytolith percentage). In the appendix, all p-values and correlation coefficients per phytolith group are given. None of the phytolith groups responded significantly to the stable isotope record. Spherical palms (begin) and grasses (end) responded most consistent to the predictor variable (Palms: RHO= 0.14, P= 0.47, Grasses: RHO = 0.17, P=0.41). See appendix for all values.
10
Discussion
Human index scores
Charcoal was absent throughout the timeframe analyzed in this paper (Huisman, 2016). And no evidence for agriculture was found in the phytolith record. The single maize pollen grain (1550 BP) reported by Colinvaux et al. (1985) could not be verified in the phytolith record. This maize pollen grain is therefore left out of consideration in the Human Index Score. Further palynological research should confirm the presence or absence of maize pollen.
The baseline for an undisturbed phytolith record is thought to be Cocha Cashu (McMicheal et al. 2015). Comparing Lake Kumpak with the baseline shows that the phytolith derived Human Index Scores (dP, dESH and tAR) are higher than the baseline values. Phytolith-derived Human Index Scores are indirect measures of human impact meaning that other natural mechanisms may affect vegetation as well. Climate is an important abiotic factor influencing plant distribution and diversity and will be evaluated later in this paper (Pearson & Dawson, 2003).
Sampling resolution might affect phytolith-derived Human Index Scores. Samples from Lake Kumpak are obtained from a high-resolution sediment core covering nearly 6000 years of rainforest history in 19.6 meter. For Lake Ayauchi over 7000 years of sedimentation history is captured by only 2.86 meter of sediment (McMichael et al. 2012a). The possibility to sample at a high resolution at Lake Kumpak increases the chance of capturing a stochastic event. One outliner (936BP) which was reproducible directly influenced the phytolith-derived Human Index Scores.
Impact of human-induced fire on Lake Kumpak: Hypothesis 1 and 3
Due to the absence of charcoal throughout the Lake Kumpak record, the first and third hypothesis could not be tested (Huisman, 2016; Heining, 2016). Natural fires are very rare in aseasonal tropical rainforests (Nepstad et al., 2004). Finding charcoal in lake sediment is therefore a strong indicator for human presence (Bush & Silman, 2007; Bush et al. 2007).
Many palms are considered as useful species because palms can be used for many purposes (Zambrana et al. 2007). Useful species might be promoted over time as their presence is favored over the presence of less useful species (Balée, 2013). However research so far showed that human impact on for instance palm species Iriartea is not always consistent (Bush & McMichael, 2016). Long-term land management of grass landscapes in Venezuela using fire resulted for instances in patches of Mauritia Flexuosa palm forest which were spared consistently over time (Rull & Montoya, 2014). Byg & Balslev (2006) found that modern indigenous tribes spare palms when a forest patch is cleared because of the useful properties of palms. Other research reported that entire patches of rainforest –including the useful palm Iriartea
deltoidea- were denuded by European settlers (Clark, Clark, Sandoval, & Castro, 1995). Promotion of
palms by human has also been shown in a phytolith reconstruction for Teotoniô were the difference between the highest amount of palm phytoliths in a sample and the lowest was 60 (McMichael et al. 2015). However, without any indicator that humans had any impact on Lake Kumpak it is impossible to
11 test whether humans affected the palm phytolith record.
Figure 7. Left: Mauritia Flexuosa, a true niche-specialist on very wet soils in Amazonia reaching very high abundances locally. Right: Iriartea deltoidea, one of the most common tree species in Western Amazonia from the Palm family
Successional grasses and herbs are known to dominate a denuded forest patch for 10-20 years (Araujo, 2005). Therefore grasses and herbs are expected to increase after evidence of human-induced fire. However, many details on specifically grass species are not given in most literature. Most research discussing succession of aseasonal tropical rainforest focusses on gap succession by bamboo or tree species (Whitmore, 1984). As fire is absent in the sedimentary record of Lake Kumpak, fire can’t explain presence or persistence of bamboo forests. Other hypotheses explaining bamboo abundance are based upon soil properties, drought adaptations and bamboo being a strong successional species (McMichael et al. 2013 and references therein). Soil properties were not incorporated in this research while succession of gaps might be testable in further research looking at erosion rates and presence of successional species such as Cecropia in the pollen record. A potential effect of climate on bamboo forests will be tested in the following section.
Hypothesis 2: Palm abundance increases when stable isotope levels
are lower
This paper suggests no significant relation between stable isotope ratios and relative palm phytolith abundance. The correlation between spherical shaped palm and the stable isotope record was however amongst the most significant relations found (see appendix). A weak positive RHO-value of 0.14 was found for spherical palms with a p-value of 0.47. This would indicate that palm phytoliths are more likely to be abundant when isotope levels are low. In ecology a single predictor can never explain all variance in the data. Further research incorporating more predictor variables might be able to show whether the van Breukelen et al. (2008) stable isotope record is able to explain more of the variance in the phytolith data than other predictor variables. Beside an analytical approach to answer whether this hypothesis might be valid, the relation will be compared visually in figure 8.
12 Figure 8. Palm phytoliths compared to the van Breukelen et al. (2008) isotope record. Modified image: stars numbered 1 to 9 indicate selected peaks. Vertical black lines indicate the three major phases in the phytolith record potentially explained by isotope values. First phase is 530 to 980 BP, second phase is 980-1400 BP and third phase is 1400-2000 BP.
In general the abundance of palm phytoliths compared to the other phytolith groups seems to increase throughout the entire record which is not matched by a clear trend in the isotope record. In the first phase of the record (530-980 BP) there seems to be no consistent relation between the location of the isotope peaks (1 and 2) and the palm phytoliths. The isotope level drop around 930 BP matched by a strong palm phytolith outliner is clearly visible. A similar response in the grass phytolith reconstruction is discussed in the following section. The peaks in the second phase (980-1400 BP) align with the phytolith record relatively well. There seems to be a pronounced effect of climate on palm phytolith abundance during this phase. The idea that high (drier) isotope values align with high relative abundance of palm phytoliths is however unexpected. Peak 7 is the lowest level in the isotope record and marks the beginning of a general increase in palm abundance which gives the impression that this wetter phase in the isotope record helped the palm community to stabilize. This idea may be tested by analyzing the pollen reconstruction. In the third phase, the two peaks in the isotope record have no clear influence on palm phytoliths and palm phytolith abundance is low in general.
One interesting observation is that a dry phase at Lake Kumpak has been proposed between 700-1200 BP based on the absence of Iriartea in the pollen record (Liu & Colinvaux, 1989). This observation was mentioned to be an out-of-phase climate shift as a wet phase was proposed elsewhere during the same time window. Evidence for a wet phase is found at several sites over large spatial distances (Bush et al., 2016; Colinvaux, Frost, Frost, Liu, & Steinitz-Kannan, 1988; Thompson et al., 2013) The deepest drop in the Cueva del Tigre isotope record coincides with the onset of an increase in relative palm phytoliths and the onset of Paul Colinvaux’s wet phase. Without speculating too much, such observations might feed the
13 idea that these events were related. This study claims that a wet phase might have been present at Lake Kumpak between 1200-700 BP as palms in general increased in abundance between 1500 and 800 BP. This would mean that the climate of Lake Kumpak did not behave out of phase with climate in the North of Ecuador. More proxies such as diatom analysis and pollen reconstructions are required to build up evidence for either the ideas presented in this paper or the original ideas of Liu & Colinvaux (1989).
hypothesis 4: Grass abundance increase when stable isotope levels
are higher
Figure 9. Grass phytoliths compared to the van Breukelen et al. (2008) isotope record. Modified image: stars numbered 1 to 7 indicate selected peaks. Vertical black lines indicate the three major phases in the phytolith record potentially explained by isotope values. First phase is 530 to 980 BP, second phase is 980-1400 BP and third phase is 1400-2000 BP.
The fourth hypothesis states that relative abundance of grass phytoliths is high when isotope values are high. The general impression of phytolith record is that the first phase starts with very low values. Relative grass abundance increases until the isotope drop at 930 BP. The high abundance of grass while isotope values dropped was unexpected. Because peak 3 of the isotope record triggers both a rapid increase in palm abundance and in grass abundance, the sedimentary record was checked to ensure that no slumping or landslides had affected the phytolith record. Greyscale values around the 365th centimeter of the
sedimentary record were not significantly different than greyscale values elsewhere. During the second phase (980-1400 BP) the relative abundance of grass phytoliths is relatively low. These lower values coincide with the proposed wet phase in Northern Ecuador and the proposed dry phase of Lake Kumpak. Following the logic of hypothesis four, the low abundance in grass during the second phase is in favor of wetter circumstances around Lake Kumpak. The third phase shows two aligning peaks (6 and 7) and a general increase in the relative abundance of grass phytoliths. In general the phytolith record follows the isotope record relatively well during this phase.
14
Hypotheses 2 and 4: general objections
The first problem is the stable isotope data derived from stalagmites from Cueva del Tigre Perdido, Peru may show a signal which was not present at Lake Kumpak. Climatic maps based on remote sensing by Deblauwe et al. (2016) show that current climate is somewhat similar at Cueva del Tigre Perdido and Lake Kumpak. There are however differences and these differences might have been exaggerated in the past. Vuille et al. (2012) show that the South American Summer Monsoon (SSAM) is for instance stronger in Peru while Flantua et al. (2016) show that climate modes can behave different over spatial scales of a few hundreds of kilometer. The usefulness of the van Breukelen et al. (2008) record for Lake Kumpak might be assessed in further research comparing the Cueva del Tigre Perdido isotope record with the Santiago Cave isotope record analyzed in (Mosblech et al., 2012). It would also be useful to see whether using the Santiago Cave isotope record yields better results.
A general problem with comparing isotope levels with vegetation is that ecosystem responses to climate changes might show a temporal lag. Bush & McMichael (2016) show by analyzing Iriartea pollen data that this hyperdominant species perhaps only achieved this status since the late Holocene (past 3000 years). In general the abundance of this species increases gradually as moisture availability increases over time. However, this doesn’t imply that such a species shows no temporal dynamism. The species often increases or decreases in the pollen record without any direct indictor for such behavior. In this paper, some peaks in the isotope record were aligned with peaks in the phytolith record, but this might not be a good way of analyzing phytolith data. Even though peaks may align, other factors could have a certain effect leading to an apparent relation which is only a result of chance. The broad trends in the phytolith record are assumed to be useful, but cross-validation with the pollen record and a local proxy for paleoclimate should be assessed to obtain a more complete picture.
Relevance to modern ecology
The vegetation in Lake Kumpak watershed seems to be a tropical rainforests undisturbed by humans. This is an important finding because humans were definitely present around Lake Ayauchi. Given the fact that Lake Kumpak is located within a day hiking of Lake Ayauchi, one would expected that indigenous people living around Lake Ayauchi knew that Lake Kumpak existed. There is however no indication that these people have used Lake Kumpak, not even temporarily. The zone of disturbance by indigenous people living around Lake Ayauchi is therefore considered smaller than 25 km which adds strength to earlier formulated ideas suggesting a zone of disturbance for individual settlements of respectively 3-5 km and 5-10 km (Bush & Silman, 2007; Denevan, 1996). This opposes the view of Amazonia as a cultural parkland (Heckenberger et al., 2003). This research shows that some forests in Amazonia have not been impacted by humans opposed to what other researchers state (Clement & Junqueira, 2010). This research
strengthens the hypothesis of Denevan (1996) that human habitation is more likely in seasonal settings along major rivers and these results add validity to models predicting the likelihood of finding indicators for human impact such as Terra Preta (McMichael et al., 2014).
Without any indicators for human occupation of Lake Kumpak, disturbances within the Lake Kumpak watershed are natural. Lake Kumpak is unique within Western Amazonia as Lake Kumpak has been undisturbed by riverine influences and undisturbed by humans. Leaving ecologists with a high-resolution sediment core providing 6000 years of rainforest history. Comparing 6000 years of undisturbed rainforest with 6000 years of disturbed rainforest in a very similar climatic setting will improve our understanding of past human impact on the Amazonian forest.
15
Recommendations for further research
Isotopes and human-induced fire are only two proxies while many more proxies can be used to
reconstruct the history of Lake Kumpak. More proxies will improve the reconstruction of the ecological history of Lake Kumpak. A short list of potential projects is provided.
- Further separation of the phytolith record. Recently information became available discussing palm phytoliths at a species level (Rios-Morcote et al., 2016). Conical and spherical palm behavior might be analyzed to see whether some trends are visible in this. By using predictions for dominance by ter Steege et al. (2013) a model might be built to predict palm species composition from a phytolith record.
- Validation or rejection of the hypothesized dry phase between 1220 and 700 +/- 100 years by Paul Colinvaux and colleagues. As the preciseness of paleoecological research improves over time, we might be able to evaluate the original hypothesis. In this thesis, indicators were found which reject the original hypothesis. But more validation is required. For instance by a pollen reconstruction.
- Validation of the single Maize pollen grain found in the original research.
- More research and knowledge is required to validate the Cueva del Tigre Perdido isotope record as a useful proxy for climate reconstructions near Lake Kumpak. Perhaps Santiago cave will provide a late Holocene stalagmite record in the future.
- Because Lake Kumpak is surrounded by steep slopes, landslides might have occurred in the Lake Kumpak watershed. These landslides should be visible in the greyscale record. Greyscale data can be found in the data file belonging to this thesis. Evaluating the effect of different sedimentation regimes on the phytolith record could be interesting.
Conclusion
This research aimed to test whether human impact and climatic variations had an effect on the vegetation of Lake Kumpak between 2000 and 600 BP. Human disturbance was found to be absent because no charcoal was found in the stratigraphic column. Human Index Scores showed that Lake Kumpak is as affected by human disturbance as Cocha Cashu. Phytolith-derived Human Index Scores were higher than observed at Cocha Cashu which can either be an artifact of the amount of samples obtained increasing the chance of outliners or a reflection of other disturbances. A significant effect of climate on the phytolith composition could not be found, but climate may have some effect. Palm phytoliths started to increase around 1400 BP shortly after warmest and wettest part of the stable isotope record and seemed to follow the stable isotope record remarkably well between 1400-1100 BP. During the same period grass abundance was relatively low. Together these observations doubt the original Liu & Colinvaux (1989) hypothesis of a dry phase at Lake Kumpak from 1220 to 700 BP +/- 100 years. Arboreal phytoliths showed a decreasing trend throughout the record which might reflect a slower response to a tropical rainforest which became gradually wetter over the last 4000 years. The presence of Lake Ayauchi close to Lake Kumpak adds weight to the discussion on the extent of pre-1492 human impact on the Amazonian rainforest. As Lake Ayauchi is definitely used by humans and Lake Kumpak is not, we can state that the zone of influence by indigenous people was localized. There is no evidence that people, even though they lived within 25 kilometers, used Lake Kumpak, not even temporarily. Because both lakes are located in similar climatic settings, studies comparing Lake Kumpak and Lake Ayauchi can be done to detect the development of a rainforest with human presence and with human absence.
16
Acknowledgements
I want to thank Crystal McMichael for being my supervisor and for giving me the chance to do a bachelor project in this field of research which has interested me for years. I was already ruling out the possibility to ever be able to work in this field of research. Emiel van Loon for helping me out with statistics and for being second supervisor. Will Gosling for being so polite to answer my questions while Crystal was in the field. Suzette Flantua who introduced me to the Latin America Pollen Database (LAPD) and for being my discussion partner regarding paleoenvironmental reconstructions and paleoclimate of Andean and Amazonian basin (although I didn’t find that much place for paleoenvironmental reconstructions in the final report). Finally the entire group of paleoecology students and in special my Lake Kumpak colleagues for simply being there.
17
Appendix
Phytolith identification
Phytoliths were identified using the handbook of Piperno (2006) and specialist knowledge of Crystal McMichael. Separating phytoliths can still be challenging. For this reason 200 phytoliths were counted per slide, this reduces the impact of any miscounts.
The following arboreal phytoliths were distinguished. Large rugose sphere, small rugose sphere, oblong capitate, cranate spheres, decorated, small echinate, large echinate and oblong echinate. From this arboreal group, small and large echinates and oblong capitate were not found. These three phytolith morphotypes were initially counted as oblong echinate due to inexperience. In order to maintain consistency, these phytoliths were not differentiated. The difference between cranate spheres, large echinate and spherical palms is not much and the main criteria used was size. The following classes were used. Cranate sphere >25 micron, large echinate 15-25 micron, spherical palms <15 micron. Besides that, spherical palms are less symmetric than large echinate and cranate spheres. The class spherical palm was only used when the phytoliths were spikey enough. Research published during the process of phytolith identification by Morcote-Rios et al. (2016) showed that spherical palms actually have more shaped than thought before. It was too late to correct for this in an appropriate way. Because this paper deals rather with comparing relative abundance of species groups, consistency in the counting process is more important than the exact percentage of a phytolith group.
Figure 10. 1. Small rugose sphere, large rugose sphere looks similar, but is larger. Divide is +/- 15 micron. 2. Decorated, small black dot in the middle is a property of this morphotype., 3. Small echinate rugose, note that these are easily misinterpret as spherical palms. Spikes are not distinguishable enough, 4. Oblong echinate, main criteria is an extension in the horizontal direction, 5. Large echinate, these resemble palms a lot and might even be a palm looking at new criteria of Rios-Morcote et al. 2016. 6. Cranate Sphere of 30 micron. Photos 1-5 courtesy of Crystal McMichael, 6 by author.
18 Two palm phytolith morphotypes were differentiated being conical palms and spherical palms. Conical palms are relatively easily recognized. A small discussion on spherical palm phytoliths was presented in the previous paragraph.
Figure 11. Upper three pictures are spherical palms. The left picture shows an example of a symmetric spherical palm while the right picture shows an example of a spikey spherical palm. Lowest two pictures are hat shaped or conical palms. The left is seen from the side and the right in seen from above. All photos are courtesy of Gasper Rios-Morcote and colleges.
Grass phytoliths are bilobates, tall-saddle bilobates, tall saddles, squat saddles, collapsed saddles, cross-bodies, rondels and heliconia. Of these, tall-saddle bilobates, tall saddles, squat saddles and collapsed saddles are bamboo phytoliths.
Figure 11. Collection of grass phytoliths. 1. Cross-body, 2. Tall-saddle bilobate, 3. Collapsed saddle, 4. Heliconia (indicator for gap opening, 5. Squat saddle, 6. Squat saddle or rondel, 7. Rondel
Photo 1. Courtesy of author, 2. Dolores Piperno, 3. Crystal McMichael and 4. (phytolith 5,6 and 7) are courtesy of Roussouw et al. (2009) retrieved from http://www.scielo.org.za/scielo.php?script=sci_arttext&pid=S0038-23532009000300016
19
20
Code: Interpolation of Stable isotope data
% All values in the Isotope record being zero are located locZero = find(Tigre_Perdido(:,2)==0)
A = Tigre_Perdido(:,2)
% Instead of zero, the average between the value above and below zero is taken
A(locZero)=(A(locZero-1)+A(locZero+1))/2 Tigre_Perdido(:,2)=A
% Step 2. Interpolate for AgeBP
TPint = interp1(Tigre_Perdido(:,1),Tigre_Perdido(:,2),AgeBP,'spline')
%First value is NAN replace NAN manually (=-7)
TPint(1) = -7
TPint = [AgeBP,TPint]
Code: Spearman correlation, example for spherical palms against
isotope data
% Define x and y. X is the predictor and y the response variable. In this code the data is split because of the difference in time interval after the 29th value.
x = TPint(1:29,2); %Can be replaced by another predictor variable y = FSphere(1:29); % Can be replaced by another response variable % Fit a 9th order polynomial through the data
[p,S,mu] = polyfit(x,y,9); y2 = polyval(p,x,S,mu);
% subtract the 9th order polynomial from the original data, normalizing the
data in such a way that only obvious peaks remain visible. normal = y - y2;
X = [x, normal];
[c,p] = corr(X,'type','Spearman')
% c = Spearman RHO, p = p-value indicating significance.
Results: Spearman correlation coefficients for stable isotope levels
Type Location Spearmans Rho P-value
Grass(total) Begin -0.05 0.78
Grass (total) End 0.17 0.41
Bamboo Begin -0.02 0.91
Bamboo End 0.09 0.67
Palms (total) Begin 0.09 0.64
Palms (total) End -0.03 0.89
Palms (Spherical) Begin 0.14 0.47
Palms (Spherical) End) 0.02 0.93
Palms (Conical) Begin 0.02 0.94
Palms (Conical) End 0.00 0.98
Arboreal Begin -0.10 0.59
Arboreal End -0.05 0.83
21
References
Araújo, M. M., Tucker, J. M., Vasconcelos, S. S., Zarin, D. J., Oliveira, W., Sampaio, P. D., … Miranda, I. (2005). Padrão e processo sucessionais em florestas secundárias de diferentes idades na amazônia oriental.
Ciência Florestal, 15(4), 343–357.
Balée, W. (2013). Cultural forests of the Amazon: a historical ecology of people and their landscapes. Tuscaloosa: The University of Alabama Press.
Balslev, H. (1988). Distribution Patterns of Ecuadorean Plant Species. Source: Taxon, 37(3), 567–577. Retrieved from http://www.jstor.org/stable/1221100
Barlow, J., Gardner, T. A., Lees, A. C., Parry, L., & Peres, C. A. (2012). How pristine are tropical forests? An ecological perspective on the pre-Columbian human footprint in Amazonia and implications for contemporary conservation. Biological Conservation, 151(1), 45–49.
http://doi.org/10.1016/j.biocon.2011.10.013
Barlow, J., Peres, C. A., Lagan, B. O., & Haugaasen, T. (2002). Large tree mortality and the decline of forest biomass following Amazonian wildfires. Ecology Letters, 6(1), 6–8. http://doi.org/10.1046/j.1461-0248.2003.00394.x
Bongers, F., Poorter, L., Hawthorne, W. D., & Sheil, D. (2009). The intermediate disturbance hypothesis applies to tropical forests, but disturbance contributes little to tree diversity. Ecology Letters, 12(8), 798–805. http://doi.org/10.1111/j.1461-0248.2009.01329.x
Bush, M. B., Correa-Metrio, A., McMichael, C. H., Sully, S., Shadik, C. R., Valencia, B. G., … Overpeck, J. T. (2016). A 6900-year history of landscape modification by humans in lowland Amazonia. Quaternary Science
Reviews, 141, 52–64. http://doi.org/10.1016/j.quascirev.2016.03.022
Bush, M. B., & McMichael, C. N. H. (2016). Holocene variability of an Amazonian hyperdominant. Journal of
Ecology. http://doi.org/10.1111/1365-2745.12600
Bush, M. B., Piperno, D. R., & Colinvaux, P. A. (1989). A 6,000 year history of Amazonian maize cultivation.
Nature, 340(6231), 303–305. http://doi.org/10.1038/340303a0
Bush, M. B., & Silman, M. R. (2007). Amazonian exploitation revisited: ecological asymmetry and the policy pendulum. Frontiers in Ecology and the Environment, 5(9), 457–465. http://doi.org/10.1890/070018 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.
http://doi.org/10.1098/rstb.2006.1980
Byg, A., & Balslev, H. (2006). Palms in Indigenous and Settler Communities in Southeastern Ecuador: Farmers’ Perceptions and Cultivation Practices. Agroforestry Systems, 67(2), 147–158.
http://doi.org/10.1007/s10457-005-1704-1
Clark, D. A., Clark, D. B., Sandoval, R. M., & Vinicio Castro, M. C. (1995). Edaphic and Human Effects on Landscape-Scale Distributions of Tropical Rain Forest Palms. Ecology, 76(8), 2581–2594. Retrieved from http://links.jstor.org/sici?sici=0012-9658%28199512%2976%3A8%3C2581%3AEAHEOL%3E2.0.CO%3B2-C Clement, C. R., Denevan, W. M., Heckenberger, M. J., Junqueira, A. B., Neves, E. G., Teixeira, W. G., … Watling, J.
(2015). The domestication of Amazonia before European conquest. Proceedings. Biological Sciences / The
Royal Society, 282(1812), 20150813. http://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. http://doi.org/10.1111/j.1744-7429.2010.00674.x
Colinvaux, P. A., Frost, M., Frost, I., Liu, K.-B., & Steinitz-Kannan, M. (1988). Three pollen diagrams of forest disturbance in the western amazon basin. Review of Palaeobotany and Palynology, 55(1-3), 73–81. http://doi.org/10.1016/0034-6667(88)90054-1
22 lakes and hydraulic disturbance in the upper Amazon Basin. Nature, 313(5997), 42–45.
http://doi.org/10.1038/313042a0
Connell, J. H. (1978). Diversity in tropical rain forests and coral reefs. Science (New York, N.Y.), 199(4335), 1302–10. http://doi.org/10.1126/science.199.4335.1302
Deblauwe, V., Droissart, V., Bose, R., Sonké, B., Blach-Overgaard, A., Svenning, J.-C., … Couvreur, T. L. P. (2016). Remotely sensed temperature and precipitation data improve species distribution modelling in the tropics. Global Ecology and Biogeography, 25(4), 443–454. http://doi.org/10.1111/geb.12426
Denevan, W. M. (1996). A Bluff Model of Riverine Settlement in Prehistoric Amazonia. Annals of the Association
of American Geographers, 86(4), 654–681. http://doi.org/10.1111/j.1467-8306.1996.tb01771.x
Flantua, S. G. A., Hooghiemstra, H., Vuille, M., Behling, H., Carson, J. F., Gosling, W. D., … González-Arango, C. (2016). Climate variability and human impact in South America during the last 2000 years: synthesis and perspectives from pollen records. Climate of the Past, 12(2), 483–523. http://doi.org/10.5194/cp-12-483-2016
Heckenberger, M. J., Kuikuro, A., Kuikuro, U. T., Russell, J. C., Schmidt, M., Fausto, C., & Franchetto, B. (2003). Amazonia 1492: pristine forest or cultural parkland? Science (New York, N.Y.), 301(5640), 1710–4. http://doi.org/10.1126/science.1086112
Heckenberger, M. J., Russell, J. C., Toney, J. R., & Schmidt, M. J. (2007). The legacy of cultural landscapes in the Brazilian Amazon: implications for biodiversity. Philosophical Transactions of the Royal Society of London.
Series B, Biological Sciences, 362(1478), 197–208. http://doi.org/10.1098/rstb.2006.1979
Hubbell, S. P. (2001). The Unified Neutral Theory of Biodiversity and Biogeography. New Jersey : Princeton University Press.
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), e48559. http://doi.org/10.1371/journal.pone.0048559
Liu, B., Colinvaux, P. (1989). A 5200-Year History of Amazon Rain Forest, 15(2), 231–248.
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. http://doi.org/10.1177/0959683611414932
McMichael, C. H., Bush, M. B., Silman, M. R., Piperno, D. R., Raczka, M., Lobato, L. C., … Palace, M. (2013). Historical fire and bamboo dynamics in western Amazonia. Journal of Biogeography, 40(2), 299–309. http://doi.org/10.1111/jbi.12002
McMichael, C. H., Palace, M. W., Bush, M. B., Braswell, B., Hagen, S., Neves, E. G., … Hart, J. (2014). Predicting pre-Columbian anthropogenic soils in Amazonia. Proceedings. Biological Sciences / The Royal Society,
281(1777), 20132475. http://doi.org/10.1098/rspb.2013.2475
McMichael, C. H., Piperno, D. R., Bush, M. B., Silman, M. R., Zimmerman, A. R., Raczka, M. F., & Lobato, L. C. (2012). Sparse pre-Columbian human habitation in western Amazonia. Science (New York, N.Y.),
336(6087), 1429–31. http://doi.org/10.1126/science.1219982
McMichael, C. H., Piperno, D. R., Neves, E. G., Bush, M. B., Almeida, F. O., Mongeló, G., & Eyjolfsdottir, M. B. (2015). Phytolith Assemblages Along a Gradient of Ancient Human Disturbance in Western Amazonia.
Frontiers in Ecology and Evolution, 3(December), 1–15. http://doi.org/10.3389/fevo.2015.00141
Mittermeier, R. A., Myers, N., Thomsen, J. B., da Fonseca, G. A. B., & Olivieri, S. (1998). Biodiversity Hotspots and Major Tropical Wilderness Areas: Approaches to Setting Conservation Priorities. Conservation
Biology, 12(3), 516–520. http://doi.org/10.1046/j.1523-1739.1998.012003516.x
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.
23 Mosblech, N. A. S., Bush, M. B., Gosling, W. D., Hodell, D., Thomas, L., van Calsteren, P., … van Woesik, R.
(2012). North Atlantic forcing of Amazonian precipitation during the last ice age. Nature Geoscience,
5(11), 817–820. http://doi.org/10.1038/ngeo1588
Nepstad, D., Lefebvre, P., Lopes da Silva, U., Tomasella, J., Schlesinger, P., Solorzano, L., … Guerreira Benito, J. (2004). Amazon drought and its implications for forest flammability and tree growth: a basin-wide analysis. Global Change Biology, 10(5), 704–717. http://doi.org/10.1111/j.1529-8817.2003.00772.x Pearson, R. G., & Dawson, T. P. (2003). Predicting the impacts of climate change on the distribution of species:
are bioclimate envelope models useful? Global Ecology and Biogeography, 12(5), 361–371. http://doi.org/10.1046/j.1466-822X.2003.00042.x
Piperno, D. R. (2006). Phytoliths. A comprehensive guide for archeologists and paleoecologists. New York: AltaMira Press (Rowman & Littlefield). http://doi.org/doi:10.1017/S0016756807003159
Pitman, N. C. A. (2000). A large-scale inventory of two Amazonian tree communities. Duke University . Rull, V., & Montoya, E. (2014). Mauritia flexuosa palm swamp communities: natural or human-made? A
palynological study of the Gran Sabana region (northern South America) within a neotropical context.
Quaternary Science Reviews, 99, 17–33. http://doi.org/10.1016/j.quascirev.2014.06.007
ter Steege, H., Pitman, N. C. A., Sabatier, D., Baraloto, C., Salomão, R. P., Guevara, J. E., … Ree, R. H. (2013). Hyperdominance in the Amazonian tree flora. Science (New York, N.Y.), 342(6156), 1243092. http://doi.org/10.1126/science.1243092
Thompson, L. G., Mosley-Thompson, E., Davis, M. E., Zagorodnov, V. S., Howat, I. M., Mikhalenko, V. N., … Ramirez, E. (2013). Annually resolved ice core records of tropical climate variability over the past ~1800 years. Science (New York, N.Y.), 340(6135), 945–50. http://doi.org/10.1126/science.1234210
van Breukelen, M. R., Vonhof, H. B., Hellstrom, J. C., Wester, W. C. G., & Kroon, D. (2008). Fossil dripwater in
stalagmites reveals Holocene temperature and rainfall variation in Amazonia. Earth and Planetary Science Letters (Vol. 275). http://doi.org/10.1016/j.epsl.2008.07.060
Vuille, M., Burns, S. J., Taylor, B. L., Cruz, F. W., Bird, B. W., Abbott, M. B., … Novello, V. F. (2012). A review of the South American monsoon history as recorded in stable isotopic proxies over the past two millennia.
Clim. Past, 8, 1309–1321. http://doi.org/10.5194/cp-8-1309-2012
Whitlock, C., & Larsen, C. (2002). Charcoal as a Fire Proxy. In Tracking Environmental Change Using Lake
Sediments (pp. 75–97). Dordrecht: Kluwer Academic Publishers. http://doi.org/10.1007/0-306-47668-1_5
Whitmore, T. C. (1984). Gap Size and Species Richness in Tropical Rain Forests. Biotropica, 16(3), 239. http://doi.org/10.2307/2388058
Whittaker, R.H.,Levin, S.A., Boot, R. B. (1973). Niche, Habitat, and Ecotope. The American Naturalist , 107(955), 321–338. Retrieved from http://www.jstor.org/stable/2459534
Willis, K. J., Whittaker, R. J., Gould, S. J., Rahbek, C., Graves, G. R., Crawley, M. J., … Lawton, J. H. (2002). Ecology. Species diversity--scale matters. Science (New York, N.Y.), 295(5558), 1245–8.
http://doi.org/10.1126/science.1067335
Zambrana, N. Y. P., Byg, A., Svenning, J.-C., Moraes, M., Grandez, C., & Balslev, H. (2007). Diversity of palm uses in the western Amazon. Biodiversity and Conservation, 16(10), 2771–2787.
