Panama forms the land bridge between North and South America. Today, a large part of Panama is covered by tropical forest which are some of the world’s most biologically diverse areas.
Many of the Central- and South American tropical rainforests were considered as sparsely populated and pristine areas prior to the arrival of the Europeans in 1502 (Denevan, 1992; Meggers, 2003). While it was known pre-Columbian people occupied Panama at least 11,000 years ago (Piperno, Bush and Colinvaux, 1990; Cooke, 2005), their impact was once argued to be minimal because of resource limitations in tropical forests (Bush et al., 2015). A growing number of historians and geographers now believe in a contrary vision; one which sees the Central- and South American rainforests as a ‘human-modified landscape’ instead of ‘pristine areas’(Erickson, 2000; Heckenberger et al., 2003; Stokstad, 2003; Bush et al., 2015). Growing evidence of dense human populations and large agricultural areas in the forests lead to the vision of a ‘human-modified’ landscape.
Natural fires in humid tropical forests rarely occur. Lightning can cause natural fires in these forests, but most of the forests are too wet and therefore burn themselves out in a few tens of meters (Bush et al., 2015). Therefore, fires in humid tropical forests have primarily been a result of human activities (Bush et al., 2015). The use of fire by humans dates back to more than a million years ago (Shimelmitz et al., 2014), but after the rise of agriculture people used fire more intensively to clear land, remove weeds and control diseases (Brandt, 1966; Pausas and Keeley, 2009). Fire in these forests have ecological consequences. Anthropogenic fire disturbances can alter the vegetation structure and species composition of a forest. A loss of primary forest species has been observed in burned forests (Barlow and Peres, 2008) and overall species diversity decreases in a forest with repeated burns (Cochrane and Schulze, 1999).
The Holocene (11,700 – present) has been a time of large population and agricultural expansion. In the early Holocene, hunter–gatherers began to domesticate specific plants and animals, but human activities, especially agricultural development, increased in the late Holocene (Carcaillet et al., 2002; Oliver, 2008). In Panama, ceramics were used by c. 4500 yr BP (Iizuka, 2017) and people cultivated a variety of domesticated seeds and crops by 7000 yr BP (Piperno et al., 1985, 2000; Piperno, Bush and Colinvaux, 1991; Piperno, 1998; Dickau, 2010).
Moreover, palms are well known to be important sources of food and building materials for humans and some economically-important palms are domesticated (Clement, 2006; McMichael et al., 2015; Sosnowska, Walanus and Balslev, 2015). During pre-Columbian times Bactris gasipaes (peach palm) was domesticated and widely used for human consumption, flour and oil production in the lowland tropical rainforests (Clement and Urpí, 1987). At the time of the arrival of the Europeans, the fruit of the peach palm was an important food crop and it was used to make a fermented drink (Clement et al., 2017). Some other useful palms that are native to the Panamanian landscape are Oenocarpus mapora, Astrocaryum standleyanum, Euterpe precatoria, Cocos nucifera and Elaeis oleifera. These palms are harvested for multiple purposes; for obtaining oils, the use as a food or building material (Henderson, Galeano-Garces and Bernal, 1997; Eynden, Cueva and Cabrera, 2003).
Panama was part of the Spanish Empire for over 300 years (1513 – 1821 A.D.). In the early colonial period a population collapse of Native Americans occurred between 1500 and 1650 A.D. (Newson, 1993; Clement, 1999). After the Spanish colonial era the Panama Canal was constructed (ca. 1881-1914 A.D.) to link the Atlantic and Pacific Oceans. However, conclusions that tropical forest composition reflects pre-Columbian cultivation patterns fail to consider four centuries of European influence, including the population collapse and the construction of the Panama Canal.
All these past activities may have left ecological legacies on the modern vegetation. There are, however, few paleo-environmental data that directly link past disturbances with modern forest composition. Previous studies showed that the distribution of some useful plants across modern Amazonia is associated with past human occupation and disturbances (Ross, 2011; Levis, 2018). An example is the Brazil nut (Bertholletia excelsa), which is found frequently in close proximities to Amazonian Dark Earths (Thomas et al., 2015; Levis, 2018). Furthermore, ancient Maya “forest garden” species – tree species used for daily household needs- still persist in the modern vegetation after centuries of abandonment (Ross, 2011). This raises the question if more useful plants are found in close proximities to places with high human impact. If forests, that are considered pristine, are still recovering from past human disturbances then our understanding of tropical forest biodiversity patterns and processes may be skewed, and this would be useful information for conservation decisions.
To investigate the vegetation and fire history, phytoliths and charcoal analyses will be performed. Phytoliths and charcoal are useful proxy investigators for the reconstruction of human impact on the vegetation. Phytoliths are silica-based remains of plants that preserve in soils providing information about the vegetation in geological timescales and human activity by providing evidence of 1) cultivars (e.g., maize, manioc and arrowroot), 2) some economically-important palms (Morcote-Ríos, Bernal and Raz, 2016) and 3) forest clearings and canopy openings. Phytoliths have several advantages for investing human impact on the vegetation. Firstly, unlike pollen, phytoliths do not degrade in terrestrial soils. Therefore, they are an essential tool to reconstruct past vegetations in the tropics (Piperno and Becker, 1996; Watling et al., 2016). Secondly, phytoliths are released into the soil where the plant dies and therefore show a local signal of the vegetation, which is beneficial for investigating human impact on the vegetation (Piperno, 2006; Watling et al., 2016).
The Gigante Peninsula, which is a part of the Barro Colorado Nature Monument in Panama, is one of the most researched areas in the world (Smithsonian Tropical Research Institute, no date). However, whether prehistoric people occupied these forests, and whether they left ecological legacies in Gigante is unknown. To determine whether the forests of Gigante may still be recovering from past disturbances, the local fire and vegetation history of the Gigante Peninsula in Panama will be reconstructed using phytolith and charcoal analyses. More specifically in this study the following questions will be answered:
1) What is the fire history of Gigante?
It is expected that the highest likelihood of past fire will be after the arrival of the Europeans approximately 500 years ago, because of agricultural intensification.
2) What is the vegetation history of Gigante?
It is expected that arboreal phytoliths will decrease and palm phytoliths will increase through time as a result of land use intensification, because several reports showed increased densities of palms at sites with human disturbances (Woods and McCann, 1999; McMichael et al., 2015). Moreover, the abundance of palm phytoliths will be significantly and positively correlated with charcoal abundances linked to human activity, whereas the abundance of arboreal phytoliths will be significantly and negatively correlated with charcoal abundances.
3) What is the spatial distribution of human impact within the plot?
Denevan (1996) stated that the location of villages was usually on high elevations (e.g. hills and elevated areas overlooking a river). Therefore, it is expected that past disturbances (signals of fire events and vegetation change) most likely occurred in highest ridges of the plot.
4) Is the modern vegetation a result of past disturbances?
Areas within a plot containing the highest levels of past disturbance will also contain higher number of palms, because previous research showed increased densities of palms at sites with human disturbances (Woods and McCann, 1999; McMichael et al., 2015). Moreover, if palm domestication happened in the plot it is expected that palms are more abundant in closer proximities to plots with a higher human impact.
Methods
2.1. Site description
The Gigante Peninsula plot (-79.85342, 9.102242, Fig. 1) is a 26.6 ha plot owned by the Smithsonian Tropical Research Institute and it is among the best researched and monitored tropical forest on Earth. It is located approximately 1 km inland from Gatun Lake, the main channel of the Panama Canal. Gigante is ~4.5 km south of the Forest Dynamics plot on Barro Colorado Island (BCI). The Gigante Peninsula and BCI form part of the Barro Colorado Nature Monument (BCNM).
The Gigante Peninsula research site is a densely forested seasonal tropical forest system
with high precipitation levels (average 2600 mm per year) (Windsor, 1990; Shapiro and Pickering, 2000; Corre et al., 2010). Rainfall occurs during the wet season between May through mid-December (Corre et al., 2010). Humidity varies between 62 and 95 percent in wet and dry seasons. Temperature remains between 22 and 35° (Windsor, 1990). Soils are considered relatively fertile for lowland tropical forests (Wright et al., 2011). The elevation of the Gigante plot ranges between 50 and 90 meters above sea level (m.a.s.l).
The forests of Gigante have been studied intensively since Gigante has been administered by the Smithsonian institute. Since 1997, a 38.4-ha forest dynamics plot was established on the Gigante Peninsula. Every free-standing tree of more than 20-cm diameter at breast height (DBH) is recorded in the plot. The plot has been recounted in 1998, 1999, 2000, 2001, 2003, 2008 and 2013. Thirty-six 40-by-40 m subplots within the 38.4 forest dynamics plot are mapped and identified for all trees > 10 cm DBH, used for a fertilization experiment (Wright et al., 2011; Yao et al., 2018).
Figure 1; Location of Gigante.
A: Map of the Barro Colorado Nature monument and the location of the Gigante Peninsula research site, B: Map showing the 42 core locations in the 26.6. ha plot. Black circles show the cores from which charcoal is obtained. Red circles show the cores from which charcoal and phytoliths were obtained. A SRTM-derived digital elevation model (DEM) is displayed over the extent of the study site, with elevations ranging between 50 and 90 meters above sea level.
2.2. Sample collection
Within the 26.6 ha Gigante plot, 42 soil cores were collected at 5 or 10 cm intervals up to 80 cm or as far as the water ground level allowed (Fig. 1B). In moist tropical forests, a depth of 80-100 cm usually contains soil and fossils from the last 3000-5000 years (McMichael et al., 2015). These types of soils are general stratigraphically intact. However, mixing and reversals of layers can take places in these soils (McMichael et al., 2015). The soils were subsampled for charcoal and phytolith analyses. Cores were collected at various elevations (50-90 m.a.s.l.). One of the cores used for phytolith analyses (core 40) fell into a fertilization experiment subplot.
2.2.1 Fire history – Charcoal analyses
The methodology of soil charcoal analysis followed standard techniques (McMichael, Correa-Metrio and Bush, 2012). Soil was poured into a cylinder with 30 ml of water to calculate the volume of soil. Sampled used for charcoal analyses contained a volume between 4 and 9 ml of soil. Samples were deflocculated with 3% H2O2 and shaken for 24h, followed by sieving at 500 μm. Remaining material on the sieves was collected in Petri dishes and charcoal was identified and photographed. Surface area was calculated using ImageJ software (Rasband, 2012). Surface area calculations were converted to estimates of charcoal volume per sample using the equations derived by Weng (2005).
2.2.2. Vegetation history – Phytolith extraction, identification and counting
From the 42 soil cores, a subset of 8 cores was analysed for phytolith composition. The selection of these cores was based on a gradient of charcoal abundances, ranging from 0% to 67% of depth intervals containing charcoal in a core. Phytoliths were extracted following the standard procedures based on Piperno (2006). Organic material was removed from the soil by the oxidation of organic matter using a 33% solution of H2O2 and KMNO4, followed by carbonate removal using 10% HCL. After removal of and grains and clay particles by sieving using 212-micron sieves, phytoliths were separated from the remaining soil using bromoform (CHBr3), with a specific gravity of 2.3, and mounted on microscope slides with Naphrax.
For phytolith analysis, a minimum of 250 phytoliths were counted at a magnification of 1000x using immersion oil. Identification of the phytoliths were done by using the reference material from Piperno (2006), Morcote-Ríos (2015, 2016) and Huisman (unpublised). Dark brown or black phytoliths were counted separately as “burned”. Phytoliths from the most common tree species in the plot were identified by comparing with a modern reference collection collected at the Leiden Herbarium.
2.3. Data Analyses
2.3.1. Fire history – Charcoal analyses
Charcoal fragments were submitted to the DirectAMS laboratory (47.775990, -122.182812) for 14C dating. To obtain the age of the most recent fire event, fragments from the uppermost soil layers were dated. Moreover, fragments from lower depths of cores are dated, particularly when there appeared to be multiple fire events within a core. Calibration of reported dates was undertaken in Rstudio Version 1.1.453 (RStudio Team, 2016) using the “Bchron” package (Parnell, 2016) and the IntCal13 atmospheric curve (Reimer et al., 2013). To investigate the spatial patterns of fire, Rstudio was used to map the elevation of the plot with the amount of charcoal that was found in all samples. Elevation data was obtained from the Shuttle Radar Topography Mission (SRTM), which resulted in a 30-m resolution Digital Elevation Model (DEM).
2.3.2. Vegetation history – Phytolith analyses
Phytolith percentages per taxon were calculated and graphed using C2 software Version 1.7.7. (Juggins, 2016). Two metrics were used to investigate the magnitude and direction of vegetation change. The first metric that was used is the trend metric (trend = surface abundance – basal abundance) of a given phytolith type to estimate the direction of vegetation change through time within a core. The second metric, the delta metric (delta = maximum abundance – minimum abundance), was used to calculate the overall magnitude of change within a core. To investigate the patterns of phytoliths between the different cores and the depth intervals a Detrended Correspondence Analyses (DCA) was performed.
To test whether changes in the abundance and distribution of palms through time is associated with past fire disturbance, a Spearman correlation test was used to compare phytolith abundances with charcoal abundance. The average of phytolith and charcoal abundances were calculated for every depth interval and used in the correlation tests. Extreme values of charcoal abundances were excluded from the correlation tests.
2.3.3. Spatial distribution of human impact
The spatial distribution of human impact was assessed by developing a core-level disturbance metric for each core that was analysed for charcoal and phytoliths within the plot. A “Human Impact Score” was derived for each core based on the following variables within soil cores: (1) the percentage of samples that contained charcoal; (2) the percentage of samples that contained >40% palm phytoliths; (3) the percentage of samples that contained more than 5% conical phytoliths (possible evidence of palm domestication); (4) the magnitude of change arboreal taxa; (5) the magnitude of change of palm taxa; and (6) the magnitude of change of grass taxa.
The percentage of samples that contained charcoal was used because fire abundance can be directly linked to human activities in the plot. The percentage of cores that contained more than 40% palms was used because palms could be a result of palm cultivation and looking at the obtained phytolith data, more than 40% of palm phytoliths seemed like a turning point in the vegetation. The percentage of depths that contained more than 5% conicals was used because this may reflect peach palm domestication. The magnitude of changes in phytolith taxa was used because rapid changes in vegetation may reflect human activity.
Because fire is a direct result of human activities, the percentage of depths that contained charcoal were given a weighting of 3. The percentage of depths that contained more 5% conicals were given a weighting of 2, because this may be a direct result of human activities but could also be due to climatic change. All the other variables were given a weighting of 1 because these are possible indirect results of human activities but can also have other causes, such as climatic or other forms of ecological changes. The weighted sum of all proxies was used to create the “Human Impact Score (HIS)”.
2.3.4. Relation to modern vegetation
To determine whether areas containing higher levels of past disturbance have higher palm abundances in the modern forests (a legacy of palm enrichment), a Spearman correlation test was performed on the core-level disturbance metric (Human Impact Score) and the percentage of palms in the closest forest inventory subplot from the fertilization experiments of the sample site. The subplots in the forest inventory plots were used because palms often have a lower DBH than 20 cm and the subplots contain all trees with a DBH>10 cm.
A Spearman correlation test was performed on the percentage of palm phytoliths in the surface samples and the percentage of palms in the modern vegetation to investigate the relationship between phytolith abundances and the modern vegetation.
Results
3.1. Fire History
3.1.1. Charcoal Analyses Table 1. Radiocarbon data and calibrated ages for Gigante.
Date Reported Core and depth Radiocarbon age (BP ± yr) Calibrated age (yr BP)
05-07-18 44, surface Modern –
05-07-18 15, 0-5 cm 2195 ± 26 2218 – 2307
05-07-18 16, 0-10 cm 674 ± 23 642 – 671
05-07-18 2, 0-10 cm 732 ± 26 657 – 697
05-07-18 6, 0-10 cm 751 ± 20 667 – 702
08-02-17 25, 0-10 cm 8.060 ± 30 8971 – 9030
08-02-17 16, 10-20 cm 765 ± 15 674 – 705
08-02-17 6, 10-20 cm 940 ± 20 796 – 885
08-02-17 8, 10-20 cm 965 ± 15 903 – 926
08-02-17 16, 20-30 cm 795 ± 25 678 – 737
08-02-17 12, 30-40 cm 2.280 ± 15 2312 – 2342
08-02-17 25, 40-50 cm 770 ± 25 673 – 726
13-02-17 32, 50-60 cm 1.320 ± 15 1261 – 1289
13-02-17 38, 50-60 cm 2.760 ± 20 2790 – 2882
Charcoal was present in 41 of the 42 cores sampled, and abundances ranged from 0.006 mm3/ml to 30.990 mm3/ml where it was found. The majority of the samples contained less than 5 mm3/ml charcoal (Fig. 2). Sixteen of the 42 surface samples contained charcoal.
Most charcoal was found in the upper 30 cm depth intervals (Fig. 3). 40% of the surface samples contained charcoal, 81% of the 0-10 cm samples, 50% of the 10-20 cm samples, 48% of the 20-30 cm samples, 14% of the 30-40 cm samples, 17% of the 40-50 cm samples, 21% of the 50-60 cm samples, 14% of the 60-70 cm samples and 18% of the 70-80 cm (Fig. 3). 14 fragments of charcoal were used for 14C dating. Radiocarbon ages resulted in calibrated ages ranging from modern to 9030 cal. yr BP (Table 1). Two of the calibrated ages were considered as outliers (core 15, 0-5 cm and core 25, 0-10 cm).
.
.
3.1.2. Spatial patterns of fire
Most cores containing charcoal at depths greater than 30 cm were clustered on the intermediate elevations (Fig. 4A). Charcoal was found in the entire plot in the upper 30 cm depth intervals (Fig. 4B).
Figure 4; Spatial distribution of fires within the Gigante plot.
A: charcoal found in the 30 to 80 cm depth intervals of the 42 cores, B: charcoal found in the upper 30 cm of the 42 cores. The area of the filled circles is proportional to the abundances of charcoal (mm3/ml) found in the cores. Red stars indicate cores where no charcoal is found in these depths. A SRTM-derived digital elevation model (DEM) is displayed over the extent of the study site, with elevations ranging between 50 and 90 meters above sea level.
3.2. Vegetation history
3.2.1 Patterns of vegetation change
The majority of phytoliths found in the Gigante soils were composed of arboreal and palm elements, with very few grasses (Fig. 5). Arboreal abundances decreased over time (average delta = 58.2, average trend= -41.4) and mainly existed of small rugose spheres (Fig 5, Table 2). Palm phytolith abundances increased in all cores (average delta= 53.3, average trend=41.8), especially in the upper 20/30 cm (Fig 5, Table 2). Conical (palm) phytolith abundances increased over time and reached up to 12.7% (Fig. 6). Grass phytoliths were rare in the cores (average = 5.8%), and reached 19% maximum (average delta=9.6, average trend= -3.7) (Fig 5, Table 2). Nodulated spheres follow the abundances of palms and increased over time (average delta = 13.4, average trend=3.3) (Fig. 5, Table 2). Nodulated spheres were most abundant on the highest elevation of the plot (core 40), were phytoliths reached up to 33% (Fig. 5, Appendix B.6.). Nodulated spheres and conical abundances decreased in the surface samples in some cores. The majority of burned phytoliths consisted of arboreal phytoliths (70%).
Figure 6; Conical percentages of Gigante, Panama.
Percentage data of conical phytoliths of the eight cores used for phytolith analyses at Gigante. Stars (*) indicate charcoal pieces used for C14 dating with the corresponding calibrated ages.
Table 2: Magnitude and direction of change of selected phytolith taxa from Gigante, Panama.
Values of the delta (magnitude of vegetation change, = maximum abundance-minimum abundance) and trend (direction of vegetation change, =surface abundance-basal abundance) metric of all cores separately and the averages of all cores.
Core Delta arboreal Delta
palms Delta nodulated Delta grasses Trend arboreal Trend palms Trend nodulated Trend grasses
1 41.6 48.3 4.4 10.2 -31.7 37.7 -0.3 -5.7
2 47.5 53.0 6.3 4.5 -42.0 46.5 -2.0 -2.4
6 62.2 57.9 14.0 10.2 -55.7 55.6 6.0 -5.9
15 65.4 55.7 17.3 18.1 -28.5 32.8 -3.8 -0.5
23 51.7 59.6 6.7 16.9 -34.1 47.5 -0.3 -13.0
40 55.9 41.5 28.8 9.5 -33.8 26.7 9.7 -2.6
41 90.3 69.5 19.3 1.5 -89.9 69.5 19.3 1.1
44 50.8 40.8 10.7 5.4 -15.7 18.3 -2.3 -0.4
average 58.2 53.3 13.4 9.6 -41.4 41.8 3.3 -3.7
3.2.2. Detrended Correspondence Analyses
Species scores for axis 1 (DCA1, eigenvalue= 0.2303) placed the rugose spheres (large and small) at the negative extreme and the palms (conicals, spherical globular echinate, globular echinate elongate variant 1 and 2 and other palms), grasses and nodulated spheres (large and small) at the positive extreme (Fig. 7). The negative extreme of axis 2 (DCA2, eigenvalue= 0.07841) was characterized by Heliconia, grasses, elongated echinates (variant 1 and 2) (palms), globular echinates (palms) and other palms (Fig 7). The positive extreme of DCA Axis 2 was dominated by rugose spheres (large and small), conical palm phytoliths and nodulated spheres (large and small).
The DCA analyses showed a separation of most phytoliths samples according on depth and cores. Samples from all cores were evenly distributed on the positive and negative extremes of axis 1 (Fig. 7A). Samples from upper 20 cm were clustered together on the positive extreme of axis 1, whereas samples from the 20 to 80 cm depth intervals were dispersed mainly on the negative extreme of axis 1 (Fig. 7B). Samples from core 1, 2, 6, and 23 had negative scores on axis two. Samples from core 40, 15 and 44 had positive scores on axis 2 (Fig. 7A). Samples from core 41 are evenly distributed on the positive and negative extremes of axis 2. Samples from all depths were evenly distributed on axis 2, except for the surface samples. Surface samples were clustered on the negative extreme of axis 2 (Fig. 7B).
Figure 7; Detrended Correspondence Analyses of phytolith data.
Species that characterized the axes are shown by their relative placement. A: samples are divided based on cores, B: samples are divided based on depth.
3.2.3. Correlations of charcoal abundances and phytolith abundances
Based on the distribution of charcoal abundances, three extreme values of charcoal abundances (> 5 mm3/ml: 12,583 mm3/ml from a surface sample and 30.990 mm3/ml and 6.741 mm3/ml from 0-10 cm samples) are considered outliers and are excluded from the correlation tests (Fig. 2). Arboreal phytolith abundances were negatively (r=-0.85) and significantly (p=0.0037) correlated with charcoal abundances, whereas palm phytolith abundances were positively (r=0.87) and significantly (p=0.0025) correlated with charcoal abundances (Fig. 8). Nodulated spheres and conical abundances both had a positive (r=0.85 and r=0.75 respectively) and significant (p=0.0037 and 0.02 respectively) correlation with charcoal abundances (Fig. 8).
Figure 8: Correlations of specific phytoliths and charcoal abundances.
Shown are the correlations between of phytolith taxa with charcoal abundances (mm3/ml). A: correlation of arboreal phytoliths and with charcoal abundances (mm3/ml), B: correlation of palm phytoliths with charcoal abundances (mm3/ml), C: correlation of nodulated spheres phytoliths with charcoal abundances (mm3/ml), D: correlation of conical phytoliths with charcoal abundances (mm3/ml).
3.3. Spatial distribution of human impact
The weighted Human Impact Scores (HIS) between the cores ranged from 228 (core 41) to 394 (core 15), with higher scores meaning more evidence of human impact and vegetation change (Table 3). With the exception of core 41, cores on higher elevations had a higher HIS (Fig. 9).
Table 3; Human Impact Scores from Gigante.
Human Impact Scores (HIS) were calculated based on %Char (the percentages of samples that contained charcoal) and its weighting (W(%Char)), palm>40 (the percentages of samples that had more than 40% palm phytoliths) and its weighting (W(palm>40)), Con>5 (the percentage of samples that contained more than 5% conical phytoliths) and its weighting (W(con>5)), magnitude of change of arboreal phytoliths (dA), nodulated spheres (dN), palms (dP) and grasses (dG) phytoliths and their weightings (W(dA), W(dN), W(dP) and W(dG)).
Core %Char W(%Char) Palm>40 W(palm>40) Con>5 W(con>5) dA W(dA) dN W(dN) dP W(dP) dG W(dG) HIS
41 0 3 25 1 11 2 90 1 19 1 69 1 2 1 228
23 25 3 25 1 0 2 52 1 7 1 60 1 17 1 235
2 25 3 71 1 11 2 48 1 6 1 53 1 5 1 279
44 67 3 13 1 0 2 51 1 11 1 41 1 5 1 322
1 22 3 50 1 67 2 42 1 4 1 48 1 10 1 355
6 44 3 63 1 11 2 62 1 14 1 58 1 10 1 361
40 60 3 33 1 17 2 56 1 29 1 41 1 10 1 382
15 56 3 20 1 25 2 65 1 17 1 56 1 18 1 394
Figure 9; Spatial patterns of the Human Impact Scores.
The 42 core locations are shown, with black circles representing the cores from which charcoal is obtained and red circles representing the cores from which charcoal and phytoliths were obtained. Numbers represent core numbers and Human Impact Scores between brackets. An SRTM-derived digital elevation model (DEM) is displayed over the extent of the study site, with elevations ranging between 50 and 90 meters above sea level.
3.4. Relation to modern vegetation
A significant correlation (p=0.0068) was found between the Human Impact Score and the percentage of palms in the modern vegetation (Fig. 10). Core 41 was excluded from the correlation test because it was considered an outlier. When a core had a higher HIS (more evidence of human impact and vegetation change), lower palm abundances were found in the modern vegetation. Similarly, when a core had a low HIS (less evidence of human impact and vegetation change), higher palm abundances were found in the modern vegetation.
No significant correlation (p=0.65) was found between the percentage of phytolith palms in the surface samples and the percentage of palms in the modern vegetation of the closest subplot (Appendix E).
Discussion
4.1. Fire history
The fire history of Gigante has been investigated to reconstruct past human activities at this site. Charcoal is found in 98% of the cores, even in pre-Columbian times. The majority of the dated charcoal have maintained stratigraphic intact. An age-depth model cannot be made of these cores because multiple cores are used to give an indication about the ages. Besides, mixing and reversals of layers and re-deposition of charcoal may have taken place, due to bioturbation or other disturbances (i.e. agriculture or slope wash) (Whitlock and Larsen, 2001; McMichael et al., 2015) . An example of soil overturning or re-deposition can be seen in the calibrated age 2218-2307 in the 0-5 cm depth interval of core 15, and the calibrated age of 8971-9030 in the 0-10 cm interval of core 25. Therefore, these calibrated ages are considered as outliers.
Because fire is not a natural phenomenon in tropical rainforests (Bush et al., 2015), the presence of charcoal suggest that humans were present in the area of Gigante both in pre-Columbian times and in the modern era. The youngest fire disturbance was found after 1950 A.D., which suggest that some parts of the forest of Gigante can be considered as young forests. It is interesting to note that the phytolith analyses of the surface samples did not show a big vegetation change, except for a decrease in nodulated spheres and in some parts of the forest a decrease of conical phytoliths.
The highest likelihood of past fire abundance was expected around 500 years ago. The most intensive period of fire was in the upper depth 30 cm, with calibrated ages of approximately 900 years BP. The increase of fire suggests that human populations increased in the region over the last 900 years and that pre-Columbian people settled in area that is now the forest dynamics plot on the Gigante Peninsula.
The spatial distribution of fire disturbances indicated that before approximately 900 years ago fire disturbances mainly occurred in areas with intermediate elevations, which supports the hypothesis that past fire did not occur on the lowest elevations. Prior studies showed that the topographic location of villages was usually on high elevations (e.g. hills and elevated areas overlooking a river) (Denevan, 1996). A possible explanation for the highest intensity of burning in the intermediate elevations can therefore be sought in the avoidance of burning higher elevations, where houses were located, and the incapability of burning the lower elevations, where flooding’s could have occurred more regularly. After approximately 900 years ago fire disturbances are found throughout the entire plot, which suggest that humans also migrated to the lower elevations.
4.2. The agricultural and cultivation history
No evidence of maize was found in the phytolith samples. Therefore, there is no direct evidence of agriculture in forest plot. The forest plot of Gigante is at an elevation between 50 and 90 meters above sea level. Maize agriculture is often found at frequently/yearly flooded areas or periodic swamp forests (Denevan, 1996; Oliver, 2008), so a possible explanation for the absence of maize is that the elevation was too high for maize cultivation.
The results from this study supported the hypothesis that arboreal phytoliths would decrease and the palm phytoliths increase over time. Palms are well known to be important sources for food and building materials for humans (Clement, 2006; McMichael et al., 2015; Sosnowska, Walanus and Balslev, 2015) and several reports showed increased densities of palms at sites with human disturbances (e.g. archaeological sites and Amazonian Dark Earths) (Woods and McCann, 1999; McMichael et al., 2015).
A negative correlation is found between the percentage of trees with the amount of charcoal and a positive correlation is found between the percentage of palms (including the conical shaped phytoliths) and the amount of charcoal, which supported the hypotheses. Moreover, 70% of all burned phytoliths consisted of arboreal phytoliths. A possible explanation for these results is that trees are burned down and replaced with palms for palm cultivation. Furthermore, palm cultivation less likely occurred on the highest elevations because the lowest delta and trend of palms is found on the highest elevation of the plot (core 44). This is line with the results that the houses were probably located on the highest elevations. Therefore, palm cultivation most likely occurred on the lowest and intermediate elevations.
Oenocarpus mapora, which produces elongated echinates (variant 1), is one of the most common palm species in the modern vegetation according to the forest inventory tree data. Euterpe precatoria, another species that produces globular echinate elongates (variant 1 and 2) and globular echinates, is also common in the forest plot of Gigante but is less abundant than Oenocarpus mapora (Huisman and McMichael, unpublished ; Bozarth et al., 2009; Morcote-Ríos, Bernal and Raz, 2016). Euterpe precatoria and Oenocarpus mapora are harvested for the use as a food and thatch in pre-Columbian times (Henderson, Galeano-Garces and Bernal, 1997; Eynden, Cueva and Cabrera, 2003; Sosnowska, Walanus and Balslev, 2015). Therefore, the increase of palms in Gigante could indicate palm cultivation.
Palm species that produce conical phytoliths (including Bactris and Astrocaryum spp.) were present in much rarer quantities than those producing globular echinate phytoliths or elongated echinates but increased in all cores (Huisman and McMichael, unpublished; Bozarth et al., 2009; Morcote-Ríos, Bernal and Raz, 2016; Watling et al., 2018). Bactris and Astrocaryum spp. are important economic palms used in prehistoric times (Piperno, McMichael and Bush, 2015). Bactris gasipaes (peach palm) was cultivated widely in human tropical areas of Central and South America for especially their fruits during the pre-Columbian era (Henderson, Galeano-Garces and Bernal, 1997; Clement et al., 2017). Therefore, the increase of conical phytoliths could indicate peach palm domestication in the plot. Astrocaryum standleyanu is also common in the area of Gigante. Astrocaryum standeleyanum provides fruit, oil and especially fiber and could also be cultivated in Gigante (Fadiman, 2008).
Nodulated spheres were not abundant in the cores but follow the abundance of palms and have a positive correlation with charcoal abundances. Nodulated spheres occur in four families, all part of the order Zingiberales: the Marantaceae, Cannaceae, Costaceae and Zingiberaceae (Chase, 2004; Piperno, 2006). These families are usually large herbaceous plants with rhizomatous root systems and have species of economic importance with many species of local use (Prince and Kress, 2002). Zingiberaceae (or ginger family) is a family with many important ornamental, spice, and medicinal plants such as ginger, cardamom and turmeric. Nodulated spheres were most abundant on the highest elevation (core 40). As mentioned before, houses were probably located on the highest elevations of the plot. Tropical homegardens are land-use systems within close proximities of houses with selected trees, shrubs and herbs grown for food and aesthetic and ecological benefits (Kunhamu, 2014). Because many of the Zingiberales are economically important species, the high abundance of nodulated spheres in the highest elevations could indicate homegardens.
A striking pattern is the decline of nodulated spheres and conicals in the surface samples.
Surface samples are the only samples that contain charcoal that has been dated after the arrival of the Europeans. After the arrival of the Europeans a population collapse occurred between 1500 and 1650 (Newson, 1993; Clement, 1999). Many of the crop genetic diversity required human maintenance. A consequence of this population collapse is a loss in crop genetic diversity (Clement, 1999). This could explain the decline of nodulated spheres and conicals after the arrival of the Europeans.