Bush & Silman Amazonia Revisited

8/11/2019 Bush & Silman Amazonia Revisited

1/9

457

The Ecological Society of America www.frontiersinecology.org

The first scientists to enter Amazonia encountered awonderland of undescribed organisms living in whatappeared to be unoccupied and untouched forest.Adjectives they used to describe the forest included vir-gin, pristine, and timeless, a vision which becameincorporated into scientific thinking. Explanations ofhigh Amazonian diversity invoked the stability and themuseum-like quality of unchanging environments thataccumulated species and minimized extinctions (Stebbins

1974). Coupled with this view was the romantic idyll ofhuntergatherers living in harmony with nature.

Such views have been altered by excavations of archaeo-logical sites from the mouth of the Amazon to the HighAndes, which reveal a long record of human occupation,ceramics manufacture, and agriculture (eg Roosevelt 1991;Roosevelt et al. 1991). For at least some groups, a trajectory

of increasing populations and greater reliance on agricul-ture is evident for the past several thousand years. Asanthropological and paleoecological knowledge of thesesystems has deepened, the view of Amazonia as untram-meled and changeless has disappeared (Clark 1996).

The realization that human populations throughout theAmericas declined sharply following European contactaltered expectations of the level of human disturbanceand modification of systems prior to 1492. Indeed, a series

of articles (eg Clark 1996; Erickson 2000; Heckenbergeret al. 2003; Erickson 2006) and a recent book (Mann2005) suggest that the pendulum of scientific opinion hasswung from the extreme view of the Amazon as virgin,has passed a midpoint of disturbance localized aroundmain waterways, and is now headed toward the otherextreme of widespread and pervasive human distur-bance. The title of Heckenberger et al.s (2003) article,Amazonia 1492: pristine forest or cultural parkland?, wasdeliberately provocative, but if taken literally depicts asimple dichotomy. We suggest the possibility of a middlepath. Here, we provide an alternative interpretation ofthe existing data, and caution that uncritical acceptance

of Amazonia as a manufactured landscape may be mis-guided and could lead to unsound policy.

In Amazonia, there has been a commendable use of sci-ence by governments, particularly the BrazilianGovernment, to set conservation policy. Refuge theorywas widely accepted in the mid-1980s and was used at thetime, together with maps of diversity, to prioritize areasfor conservation (Dinerstein et al. 1995). Although thistheory is now largely discredited (Colinvaux et al. 2001),the basic biogeographic patterns of high local endemicityand diversity used to identify refugia also formed effec-

PALEOECOLOGY PALEOECOLOGY PALEOECOLOGY

Amazonian exploitation revisited: ecologicalasymmetry and the policy pendulumMark B Bush1* and Miles R Silman2

The influence of pre-Columbian human populations on Amazonian ecosystems is being actively debated. The long-

standing view that Amazonia was only minimally impacted by human actions has been challenged, and a new par-

adigm of Amazonia as a manufactured landscape is emerging. If such disturbance was the norm until just 500

years ago, Amazonian ecosystems could be far more ecologically resilient to disturbance than previously supposed.

Alternatively, if the manufactured landscape label is an overstatement, then policy that assumes such resilience

may cause substantial and long-lasting ecological damage. We present paleoecological data suggesting a middle

path, in which some areas were heavily modified, but most of Amazonia was minimally impacted. Bluffs adjacent

to main river channels and highly seasonal areas appear to have been the most extensively settled locations. Away

from areas where humans lived, their influence on ecosystems was very local. Consequently, we see no evidence sug-

gesting that large areas at a distance from rivers or in the less seasonal parts of Amazonia were substantially altered

by human activity. Extrapolating from sites of known human occupation to infer Amazon-wide landscape distur-bance may therefore potentially lead to unrealistic projections of human impact and misguided policy.

Front Ecol Environ2007; 5(9): 457465, doi:10.1890/070018

In a nutshell:

Pre-Columbian human influence on Amazonia was spatiallyheterogeneous, with some sites intensively altered

The large majority of Amazonia was probably barely influenced

by human activity Fossil pollen and charcoal data point to localized influencearound widely scattered occupation sites rather than a uniforminfluence

Arguments that the Amazon forest is a manufactured land-scape, and hence resilient to human activity, are only locallytrue and should not be used to set regional management policy

1Department of Biological Sciences, Florida Institute of Technology,

Melbourne, FL 32901 *([email protected]); 2Department of Biology,

Wake Forest University, Winston Salem, NC 27104


2/9


3/9

MB Bush and MR Silman Exploitation of the Amazon revisited

459


and complex societal hierarchies (Bale1989; Erickson 2000).

A further legacy of human endeavor isthe distinctive soil layering, up to 2 mdeep, of black or brown soils with ele-vated nutrient availability, known as terrapreta del Indio (Figure 2). These nutrient-rich black soils often contain brokenceramics and appear to have been alteredby the addition of ash, green manure, or,in some cases, fish meal (Lehmann et al.2003). Such soil modification by humansmay have held the key to sustainableexploitation of the land.

Some Amazonian locations wereoccupied for several thousand years.Shellfish and fish were undoubtedlyimportant dietary components, but sotoo were maize and manioc. On mostAmazonian soils, five harvests of maize

in a span of 23 years exhausts the soilof essential nutrients. A fallow period ofabout 2030 years is needed before theland can again support crops of maize(Kellman and Tackberry 1997). A prevailing view is thatthese villages remained occupied year after year at popu-lation densities too high to allow such long fallowingbetween periods of cultivation. The elevated nutrientavailability (particularly phosphorus) of terra preta soilsmay have allowed near-permanent cultivation of thesesoils, potentially supporting a much larger human popula-tion density than the native soils (Glaser et al. 2001).Indeed, the terra preta model may offer insight into how

Amazonia could withstand intensified agriculturalexploitation in the future (Glaser et al. 2001).

Based on a wide array of evidence, Bale (1989) esti-mated that 11% of Amazonian vegetation in pre-Columbian Amazonia was extensively used and alteredby human activity. That use ranged from slash-and-burnagriculture to enrichment with fruit-bearing trees (egBactris gasipaes [peach palm], Bertholettia excelsa [Brazilnut],Annona spp [includes guanabana or sweet soursop],and Mauritia flexuosa [aguaje, moriche, or nontoca; apalm whose fruit is used to make a drink]).

Inferring past human disturbance

Fire is the oldest human tool for both small- and large-scale manipulation of the landscape, and one of the majorsources of evidence for past human disturbance of tropicalforest systems comes from fire histories. Charcoal is pro-duced by forest fires and is gradually buried within a soilprofile. Charcoal layers from soil pits can be radiocarbondated, yielding an age estimate for fire events (generally 50100 years). If old wood is burned, the resulting 14Cdating of charcoal overestimates the time since the fire(Gavin 2001). Sources of such old wood could be the

heartwood of an old tree killed by the fire, or dating long-dead but undecomposed wood lying on the soil surface. Asmost tropical forest fires char bark but not heartwood(sensu Cochrane 2003), it is unlikely that heartwood fromthis source is represented in the charcoal. Also, becauserapid rates of decay in the tropics lessen the fuel load ofundecomposed heartwood lying on the forest floor com-pared to temperate or boreal settings, the probability ofcharcoal being composed of heartwood is similarly low.

The charcoal ages in tropical forest soils are thereforelikely to be reasonably representative of the actual firedate (taking into account the errors of 14C dating). When304 ages for Amazonian soil charcoal are plotted againsttime (Figure 3 a,b), the data strongly suggest an expansionof agricultural activity at ca 250 AD, then a series of peaksof fire activity between ca 700 and 1550 AD, followed bya sudden collapse at ca 1600 AD. One of the more surpris-ing results of this analysis is that the highest peak of appar-ent fire activity is not at ca 1550 AD, but at ca 700800AD. Nevertheless, the trajectory of implied disturbancefits well with archaeological estimates of village expansion(eg Roosevelt 1980; Heckenberger et al. 1999).

Hammond et al. (2006) conducted the first systematicsurvey for soil charcoal in a 60 000-ha area of Guyana.Their analysis of > 280 soil profiles revealed charcoal in allof them. The authors considered it unlikely that these werenatural fires although this section of Guyana is subject toEl Nio-related droughts and concluded that there hadbeen extensive human influence in upland easternAmazonian forests within the past millennia.

Pollen, charcoal, and phytolith data from Amazonianlakes also provide evidence for widespread human occupa-tion prior to European contact (Figure 4), and there is vir-

CourtesyofB

Glaser

Figure 2.Natural and modified Amazonian soils. (a) An oxisol and (b) a terra pretasoil. Oxisols are the most common dry land soils in Amazonia and when modified withthe addition of ash and organic waste are stained black to form terra preta.

(a) (b)

CourtesyofB

Glaser


4/9

Exploitation of the Amazon revisited MB Bush and MR Silma

460

www.frontiersinecology.org The Ecological Society of Americ

tually no sign of human presence at almost every site forthe period 15001900 AD. However, these data do notportray the intensity or scale of land use at a local level.

Within the archaeological community, a full spectrumof opinion can be found, from those suggesting minimalpre-contact disturbance (Meggers 1954, 2003) to thoseciting extensive local impacts (Denevan 1996), to thosewho believe in regional, even basin-wide, habitat modi-fication (Erickson 2000). Some anthropologists are nowreferring to the forests of Amazonia as a cultural park-

land, human created, or built landscapes (Erickson2000; Heckenberger et al. 2003; Mann 2005). Indeed,Charles Clement, an anthropologist based at INPA(National Institute for Amazonian Research) in Brazil isquoted in Mann (2005) as stating, I basically think[Amazonian lowland forest is] all human created. ClarkErickson is quoted in Mann (2005) as saying thatbuilt environment applies to almost [all], if not all,

Neotropical landscapes. As these ideas are repeatedthey tend to gain credence, and may subsequently influ-ence public perceptions and policy. But how robust are

the data underlying the assertions of widespread alteration of Amazonia?

The hypothesis of widespread Amazonian landscapmanagement is based on analyses of archaeological sites anthe assumption that there was a large pre-contacAmazonian population (> 10 million people). A caveamust be applied to these data, and indeed all of the datthat we have to date about human disturbance in thAmazon, which is that they are derived from just a few locations, and do not represent either a systematic or a randomized sampling design. There is no ecological componenpredicting which forest was most likely to be occupied. Wadisturbance spread evenly across all of Amazonia or concentrated near human habitation? Is it safe to extrapolatresults from sites where we know human habitation occureto the rest of Amazonia?

Ecologists are familiar with problems of scale. Indeedthe problem of relating phenomena across scales is thcentral problem in biology and in all of science (Levin1992). Some general observations from ecology raise a

least three concerns regarding how we may extrapolate datfrom the kind of dot map shown in Figures 3 and 4. Firstlandscapes are heterogeneous, and the resulting variabilitin environmental factors produces characteristic patterns odistribution and habitat use in nearly all species. Secondalthough species distributions can be widespread, theioccurrence may be very local. For example, a dot map of thpainted turtle (Chrysemys picta) would show that it is distributed across much of the continental US, but its actuaoccurrence is highly localized within a landscape. Third, species can be widespread and also have a very substantiaimpact on the system where it is found, but this is not thsame as saying that a species has a widespread, intens

impact, or makes wholesale changes to the habitat in whichit lives. Thus, extrapolating observations from dot maps cabe dangerous, especially when the dots represent discretactivities of limited spatial extent (eg terra preta formation

A first step in making the conjectures that are so criticato ecosystem management in Amazonia is recognizing thpotential sources of bias that affect what we infer from dataThe data in Figure 4 have two major potential sources obias. First, lakes used to study paleoecology may not accurately represent the landscape as a whole, because they arattractive places for human settlement humans like to livby water now and, apparently, always have (eg Bush et a2007b). Also, while maps of terra preta sites are available

we have no data on the distribution of soil pits that did noshow modified soils. It is a statistical certainty that wheextrapolations about land use are made from known archaeological centers and exclude other samples, the analysewill exaggerate human impacts. So how do we go aboupainting a more accurate picture of past human impacts?

Investigating occupation: the scale of impact

Fossil pollen and charcoal records (Figure 5 a,b) derivefrom the analysis of lake sediments can provide detaile

Figure 3. Soil carbon data from Amazonia. (a) Sketch mapshowing major Amazonian rivers, locations of radiocarbon-datedsoil charcoal, and terra preta. (b) Ages of 228 carbonized layers in

Amazonian soils. Data are derived from published accounts of soilcharcoal in Amazonia (Sanford et al. 1985; Saldarriaga and West1986; Desjardins et al. 1996; Piperno and Becker 1996; Pessendaet al. 1998; Tardy 1998; Santos et al. 2000; Walker 2000;Francis and Knowles 2001; Neves et al. 2004; Pessenda et al.

2004; Hammond et al. 2006). All ages are expressed in calibratedcalendar-years before present (Stuiver and Reimer 1993).

Known terra preta

Soil charcoal dated

AtlanticOcean

PacificOcean

(a)

0 500 1000

Kilometers

(b)

Num

berofrecords

18

16

14

12

10

8

6

42

0

5015025

035

045055

065

075

0850

950

1050

1150

1250

1350

1450

1550

1650

1750

1850

1950

Year AD


5/9


461


insights into Amazonian history. Pollenand spores released from plants andblown or washed into lakes becomeincorporated into the sediment. Yearafter year, the sediment accumulates,burying and preserving these plantmicrofossils in anoxic mud. Lake sedi-ments are retrieved using coring rigs sup-ported by inflatable boats. The verticalcolumns of mud and the buried layers ofmicrofossils they contain record thecomposition of vegetation that grewaround the lake. Neotropical pollen andspores can often be identified to genus,and sometimes even to species. Theradius of land represented in the micro-fossil record the vegetation directlyrepresented in the record depends onlake size. In the moderate-sized lakes wediscuss here, the great majority of the

pollen would have been derived within afew kilometers of the lake.

During a forest fire, charcoal particlesare created in both the smoke and thecharred plant remains. This charcoalfalls or is washed into nearby lakes. Thefinest fraction of charcoal reflects regional fire histories,whereas relatively large particles (ie those > 160 m inlength) indicate past fire in the adjacent watershed(Clark 1988). Today, fire is rare in most natural Amazonsystems, and many sediments from undisturbed settingscontain no charcoal. Finding evidence of regular burnsmay therefore indicate past hunting activity, where fire is

used to drive game or improve its habitat, as a mechanismto clear forest for semi-permanent dwellings, or toincrease populations of those light-demanding plants pre-ferred by huntergatherers. Finding both charcoal andpollen from crops such as corn (Zea mays) or manioc(Manihot esculenta) is a clear indicator of past agriculture.Zea mays is not native to Amazonia, and its pollen is dis-tinctive, due to its surface pattern and large size (com-monly 80110 m). Maize pollen is particularly poorlydispersed, making it a good marker of local crop use.

Given these two proxies for human activity microfos-sils and charcoal we can start to look at regional com-parisons of where humans have altered the landscape

(Figure 4). Of the 22 pollen and charcoal records fromlakes shown as yellow squares, 13 contained evidence ofpre-Columbian occupation. A minority of these sites laywithin the area predicted by Mann (2005) to be heavilyoccupied, but this pattern may reflect the relative abun-dance of lakes suitable for paleoecological study as muchas it does human distribution. Most of these lakes werechosen for the purpose of looking at the vegetation his-tory of the particular area, not as a random or even repre-sentative sample of Amazonia. At the scale of Figure 4, itis not always possible to show individual lakes, and so if

one or more lakes in a cluster show human activity, thenthe district is circled. At the broadest scale whether aregion shows human impacts or not it appears thatAmazonia was broadly, and possibly universally, influ-enced by people.

Looking at the lake data on a finer scale, however,shows the dangers of extrapolation. Three lake districts in

which multiple lakes have been analyzed provide similarspatial extents of human activity in widely separatedAmazonian landscapes (Athens and Ward 1999; Weng etal. 2002; Bush et al. 2007 b,c). In two settings, Prainhaand Maldonado (Figure 6 a,b), one lake has an unambigu-ous record of occupation and agricultural use from about4000 years ago until ca 1600 AD. The third location, aswamp named Maxus 5, near Yasuni in Ecuador (Figure6c), lies in wet forest that has very little seasonality. Thissite has a long record of charcoal in its sediment, whiletwo neighboring wetlands have none. Human occupationin this landscape is more tentative than in the other set-tings, as no crop pollen or artifacts are associated with

Maxus 5. Nevertheless, the discovery of charcoal at justone of three settings is a likely indicator of human activ-ity. The forest at Yasuni receives ~ 2800 mm of precipita-tion each year, with no distinct dry season, making nat-ural fires exceptionally rare, if not totally absent. Thatonly one of the three sites contains charcoal, and that thecharcoal was found in multiple samples, adds strength tothe suggestion by Athens and Ward (1999) that the char-coal was produced by local human activity.

In the Maldonado lake district, the paleoecologicalrecord of human activity at Lake Gentry is supported by

Figure 4. Evidence of widespread pre-Columbian occupation in Amazonia. Mapshows the known distribution of modified (terra preta) soils, lakes with

paleoecological records with and without evidence of human occupation, the major

archaeological settings of (1) Maraj Island, (2) Santarm, (3) the upper Xingu,(4) Manacapuru, and (5) the Beni. The region that Mann (2005) maps as having ahigh probability of dense settlement is shown. Terra preta data from Lehmann et al.(2003) with additional data from A Zimmerman. Pollen data from numerousauthors summarized in Bush et al. (2004; see also Bush et al. 2007a).

12

3

4

5


6/9


462


the presence of stone tools and a midden adjacent to thelake. The paleoecological markers of human occupationat this lake, corn and manioc pollen and charcoal, indi-cate that agricultural activity in the region began around4400 years ago, with as much as 3000 years of burning

prior to that time. Similarly, at Lake Geral in the Prainhagroup, corn pollen was found regularly between 4000 and400 years ago (Bush et al. 2000, 2007c), and regularlyoccurring charcoal appeared between 8000 and 400 yearsago. However, in both the Maldonado and Prainhagroups, other lakes showed either a small increase incharcoal abundance or no charcoal whatsoever, and nonecontained pollen from corn or manioc. These records sug-gest that human occupation and land conversion werelocal in nature. Sites within 35 km of an occupationcenter appeared to be quite heavily used but, beyond this,human influence declined markedly, so that at distancesof 50 km there was no evidence of human activity. This

pattern is very similar to that described by modernanthropologists, where hunting and land management isconcentrated in a 13 km radius around the center ofhabitation (Glanz 1991; Apaza et al. 2002).

The absence of humans in these records is also strikingin another way. In both the Prainha and Maldonado lakedistricts, human population collapses followed Europeancontact, as both crops and charcoal disappear from therecords in the past 400 years. Notably, the apparent col-lapse is evident even at sites that were never visited byEuropeans, providing unambiguous support for theanthropological and historical hypotheses of widespreadepidemics.

An overall picture is clear: the paleoecological data con-firm that Amazonia was exploited by indigenous peopleswho practiced agriculture and developed urban centers(Roosevelt 1980, 1991; Roosevelt et al. 1991; Hecken-berger et al. 2007), and the populations collapsed shortlyafter European contact. However, the data do not supportthe extrapolation of observations from occupied sites toinfer uniformly widespread impacts and landscape transfor-mation. There is no doubt that humans are importanttransformers of Amazonian landscapes and have been sothroughout much of the Holocene. Determining how

widespread those effects were in the vasarea of the Amazon basin, particularly thosfar from water, requires further studySpecifically, it requires a research programthat integrates existing data with new samples taken in ways that allow inferences tbe made about broader Amazonia.

Two other observations emerge fromanalysis of the charcoal data fromAmazonian soils. The peak of fire frequencwas observed not when human populationwere presumably at their largest (ie immediately prior to European contact), but at c700800 AD (Figure 3b). This period wathought to be one of peak El Nio activit

(Thompson 2000), when droughts beset Amazonia anforests were more highly flammable. It is highly probablthat the great majority of these fires were humaninduced, but that during periods of intense drought, largareas were burned (making them more likely to b

detected today) when small-scale fires escaped to becomwildfire. In this way, the spatial scale of human disturbance in Amazonia as a whole does not necessarily tracthe intensifying history of local land use in the most populous areas. The capacity of humans to disturb the systemappears to have been strongly influenced by climatic conditions, not simply a growing human presence. In addition, the time for post-disturbance forest recovery inthese settings is not the 500 years implied by the diseasmodel, but > 1000 years.

One model of human settlement in Amazonia doeincorporate known and inferred densities at the landscape level, albeit through observations of the preference

of modern indigenous populations. Denevan (1996) proposed a bluff model of human occupation, in whichindigenous peoples preferentially settled sandy bluffalongside rivers and where there were views over wetlands. Denevan estimates that as many as 10 million people lived in these densely populated riverbank settingWith just 400 000 scattered through all the rest oAmazonia, the impact of people outside the high-densitlocations may have been very limited. The paleoecological data are entirely consistent with the long-term aggregation of people around key landscape features implied bDenevans settlement model, and the inference that threst of Amazonia was lightly settled.

Seasonality and location

Although our knowledge of past human occupation inAmazonia is still in its infancy, the case for extensive occupation and use of the landscape by pre-Columbian humanactivity in Mesoamerica is clear. We need to develohypotheses that predict where humans would have existein Amazonia at densities sufficient to alter landscapes. Foexample, one initial prediction might be based on seasonality. Civilizations such as the Olmec, Maya, and Azte

(a) (b)

Figure 5. Markers of human occupation from Amazonian lake sediment. (a) Apollen grain of corn (Zea mays) and (b) microscopic charcoal fragments.


7/9


flourished in the highly seasonal settings of the Yucatanand central Mexico. In landscapes with stronger gradientsbetween ever-wet and seasonal forests, such as in Panamaand Costa Rica, the dense populations may have defor-ested the seasonal areas (Piperno et al. 1991), whereasimpacts on the dense, wet forests of the Caribbean slope

were never as great (Dickau et al. in press).A similar pattern may also have prevailed in Amazonia,

with highly favorable areas supporting large populations.Figure 7 shows the major archaeological sites that havebeen described. The main Amazon channel is clearlyimportant as a trade route, and also as an excellent sourceof fish, shellfish, and game. Within this setting, the con-fluences of rivers became centers of settlement and popu-lation hubs, as did the mouth of the Amazon. Away fromthis ribbon of occupation, signs of human habitation canbe found scattered across Amazonia. Of the areas furtheraway from the major rivers, the most complex culturaldevelopment appears to have occurred in the seasonal

(defined as having the greatest intra-annual monthlyvariability in rainfall) locations of southern Amazonia(eg Heckenberger et al. 2007). This incorporation of sea-sonality as a factor paralleling geographic location mayform a powerful predictive model for where major occu-pation sites are likely to be found.

Our contention that pre-Columbian transformation ofAmazonian landscapes was essentially local and spatiallypredictable does not contradict the development of com-plex societies (Heckenberger et al. 2003; Heckenberger etal. 2007) or that substantial areas were influenced by

463


human activity (Bale 1989), but it is

at odds with assertions regarding thegenerality of disturbance in Ama-zonia, particularly broad-scale, inten-sive disturbances. More importantly,our admittedly preliminary analysisprovides a set of predictions and ameans for testing them, so that we

can update our model of past human disturbance in theAmazon, and of the effects of the different kinds andintensities of disturbances on Amazonian forests.

Asymmetry of policy

Some years ago, Bush and Colinvaux (1994) showed thattwo lake records from the Darien region of Panamademonstrated surprisingly long histories of human distur-bance and abandonment. Since then, logging companieshave attempted to use that paper to justify timberingsome of these remote forests, arguing that, if the regionsupported cornfields 500 years ago, the forest ecosystemsmust be young and resilient, and that, therefore, loggingnow would not lead to loss of biodiversity. The problemwith this argument is that the real scale of disturbance isignored. The extended-family, slash-and-burn farmingstyle of pre-Columbian times would have had a very lim-ited impact on these systems. Certainly, there may have

been much larger impacts around major population cen-ters, but the case has yet to be made that the halo ofhuman influence was anything other than local for thegreat majority of settlement sites.

There is no question that more effort needs to beinvested in obtaining a clearer picture of the extent ofpre-Columbian settlement and alteration of Amazonianhabitats. In the meantime, it would be prudent for ecolo-gists to dig a soil pit on their research site and analyze thesoil for charcoal (in tropical soils, manganese and tita-nium also form black accretions that are strikingly similar

Figure 6. A spatial context for pre-Columbian disturbance within three Amazonianlake districts. (a) Lakes near Prainha, Brazil (14123.20 S, 533344.41 W); (b)lakes near Puerto Maldonado, Peru (12951.51 S, 69559.74 W); (c) lakes nearYasuni, Ecuador (05353.02 S, 761024.58 W).

(a)

Santa Maria:lightly used,not farmed

Geral:ocupied 7000

years, farmed4000 years,abandonedc 1600 AD

Saracuri:used, but notfarmed

N

0 5

km Amazon River

50

0

50

0

(b)

Werth: notused

Gentry: 4000 yearsfarming, abandonedc 1600 AD

Parker:used, butnot farmed

Vargas:little use,not farmed

R. Madre de Dios

0 10

km

N

N

0 20

km

(c)

Maxus 1,not used

Maxus 5,charcoal andoccupation

50

0 Maxus 4,not used

Rio Tipitini

R. Madre de Dios

Maxus 1,not used

Maxus 5,charcoal andoccupation

Maxus 4,not used


8/9


in appearance to charcoal) to determine the probabilitythat they are working on an impacted site. However, it

would be sensible for policy makers to assume minimalpast impacts at sites away from major rivers and knownarchaeological centers.

If policy makers err on the side of underestimating pasthuman impacts and the forests are truly more resilientthan we believe, the forest may be overprotected. Atworst, this policy conserves a resource for later exploita-tion and buffers against disruptions of ecological interac-tions, such as trophic cascades, which may be vitallyimportant to the structure of tropical forests (Silman et al.2003; Terborgh et al. 2006). If, on the other hand, theerror is in overestimating past exploitation and assumedforest resilience, then forests are likely to be used unsus-

tainably, and in ways that may have broad repercussionsin the face of climate change. A particular worry is thatincreasing fragmentation will lead to more frequent firesin the future (Nepstad et al. 2001), and will also impedeplant migration in the face of climate change (Ibez etal. 2006). Worse still, Amazonian forests are much morethan just trees, and the plantanimal interactions crucialto the maintenance of diversity, and regeneration afterdisturbance, may also be disrupted by such disturbance(Peres 2005). Trophic cascades and other strong ecologi-cal interactions may greatly increase the amount of forest

that needs to be conserved to retain functioning ecosystems.

It is not hyperbole to state that, together, these factorcould precipitate an Amazonian extinction crisis. Thualthough the pendulum of scientific opinion can swinthrough an arc from pristine wilderness to cultural parkland, the implications of the different resulting policiecould be ecologically asymmetric. Determining the truscale of pre-Columbian influence on Amazonian ecosystems will undoubtedly lead to a more sophisticated viewof these landscapes, in which their ecological, climaticand cultural heterogeneity are recognized.

Acknowledgements

Thanks to our collaborators and field companions, M dToledo, P Colinvaux, C Weng, and C Listopad, and to DPiperno, C Peres, and J Terborgh for many formative conversations. This research was funded by the NationaScience Foundation grants DEB 9732951 and 0237682 t

MBB and BSR 9007019 to P Colinvaux, DEB-023768and a Wake Forest University research grant to MRS, ana CNPq grant to M de Toledo. The manuscript wastrengthened by input from S Jackson.

ReferencesApaza L, Wilkie D, Byron E, et al. 2002. Meat prices influence th

consumption of wildlife by the Tsimane Amerindians oBolivia. Oryx36: 38288.

Athens JS and Ward JV. 1999. The late Quaternary of the westerAmazon: climate, vegetation and humans. Antiquity 73287302.

Bale W. 1989. The culture of Amazonian forests. In: Posey DAand Bale W (Eds). Resource management in Amazoni

indigenous and folk strategies. New York, NY: New YorBotanical Garden.

Black FM. 1992. Why did they die? Science258: 173940.Bush MB. 2004. The lowland neotropics. In: Mackay A (Ed

Holocene climate change. London, UK: Academic Press.Bush MB and Colinvaux PA. 1994. A paleoecological perspectiv

of tropical forest disturbance: records from Darien, PanamEcology75: 176168.

Bush MB, Gosling WD, and Colinvaux PA. 2007a. Climatchange in the lowlands of the Amazon basin. In: Bush MB anFlenley JR (Eds). Tropical rainforest responses to climachange. Chichester, UK: Praxis Springer.

Bush MB, Listopad MCS, and Silman MR. 2007b. A regional studof Holocene climate change and human occupation iPeruvian Amazonia.J Biogeogr. doi:10.1111/j.1365-2699.200701704.x.

Bush MB, Silman MR, de Toledo MB, et al. 2007c. Holocene firand occupation in Amazonia: records from two lake districtPhilos T Roy Soc B362: 20918.

Bush MB, Miller MC, De Oliveira PE, and Colinvaux PA. 2000Two histories of environmental change and human disturbancin eastern lowland Amazonia. Holocene10: 54354.

Clark DB. 1996. Abolishing virginity.J Trop Ecol33: 38592.Clark JS. 1988. Particle motion and the theory of charcoal analysi

source area, transport, deposition, and sampling. Quat Res306780.

Cochrane MA. 2003. Fire science for rainforests. Nature 42191319

Colinvaux PA, Irion G, Rsnen ME, et al. 2001. A paradigm to b

464


Figure 7. Seasonality and major settlement sites. If, as inCentral America, colonization of seasonal sites was moreextensive than colonization of ever-wet sites, predictions ofhuman population densities might be improved by inclusion ofseasonality as a factor. Seasonality is shown here on a relativescale of intra-annual monthly variability, with a score of 0 beingaseasonal data from Tropical Rainfall Monitoring Missionsatellite 19982004. Legend: (1) Guyana charcoal study area,(2) Maraj Island, (3) Santarm, (4) Manacapuru, (5) theupper Xingu, and (6) the Beni (Silman 2007).


9/9


discarded: geological and paleoecological data falsify the Hafferand Prance refuge hypothesis of Amazonian speciation.Amazoniana16: 60946.

Denevan WM. 1976. The aboriginal population of Amazonia. In:Denevan WM (Ed). The native population of the Americas in1492. Madison, WI: University of Wisconsin Press.

Denevan WM. 1996. A bluff model of riverine settlement in pre-historic Amazonia.Annu Assoc Amer Geogr86: 65481.

Denevan WM. 2003. The native population of Amazonia in 1492reconsidered. Revista Indias62: 17588.Desjardins T, Filho AC, Mariotti A, et al. 1996. Changes of the for-

estsavanna boundary in Brazilian Amazonia during theHolocene revealed by stable isotope ratios of soil organic car-bon. Oecologia108: 74956.

Dickau R, Ranere J, and Cooke RG. Starch grain evidence for thepreceramic dispersal of maize and root crops into tropical dryand humid forests of Panama. Proc Natl Acad Sci USA. In press.

Dinerstein E, Olson DM, Grraham DJ, et al. 1995. A conservationassessment of the terrestrial ecoregions of Latin America andthe Caribbean. Washington, DC: World Bank.

Erickson CL. 2000. An artificial landscape-scale fishery in theBolivian Amazon.Nature408: 19093.

Erickson CL. 2006. The domesticated landscapes of the BolivianAmazon. In: Bale W and Erickson CL (Eds). Time and com-

plexity in historical ecology: studies in the neotropical low-lands. New York, NY: Columbia University Press.Foster DR and Motzkin G. 2003. Interpreting and conserving the

openland habitats of coastal New England: insights from land-scape history. Forest Ecol Manage185: 12750.

Francis JK and Knowles OH. 2001. Age of A2 horizon charcoaland forest structure near Porto Trombetas, Par, Brazil.Biotropica33: 38592.

Gavin DG. 2001. Estimation of inbuilt age in radiocarbon ages ofsoil charcoal for fire history studies.Radiocarbon43: 2744.

Glanz WE. 1991. Mammalian densities at protected versus huntedsites in central Panama. In: Robinson JG and Redford KH(Eds). Neotropical wildlife use and conservation. Chicago, IL:University of Chicago Press.

Glaser B, Haumaier L, Guggenberger G, and Zech W. 2001. Theterra preta phenomenon: a model for sustainable agriculture

in the humid tropics.Naturwissensch88: 3741.Hammond DS, ter Steege H, and van der Borg K. 2006. Upland

soil charcoal in the wet tropical forests of central Guyana.Biotropica39: 15360.

Heckenberger MJ, Kuikuro A, Kuikuro UT, et al. 2003. Amazonia1492: pristine forest or cultural parkland? Science 301:171014.

Heckenberger MJ, Peterson JB, and Neves EG. 1999. Village sizeand permanence in Amazonia: two archaeological examplesfrom Brazil. Lat Amer Antiq10: 35376.

Heckenberger MJ, Russell JC, Toney JR, and Schmidt MJ. 2007.The legacy of cultural landscapes in the Brazilian Amazon:implications for biodiversity. Philos T Roy Soc B362: 197208.

Hemming J. 1970. The conquest of the Incas. New York, NY:Harcourt.

Ibez I, Clark JS, Dietze MC, et al. 2006. Predicting biodiversitychange: outside the climate envelope, beyond the speciesareacurve. Ecology87: 189606.

Kellman M and Tackberry R. 1997. Tropical environments: thefunctioning and management of tropical ecosystems. London,UK: Routledge.

Lehmann J, Kern DC, Glaser B, and Woods WI. 2003. Amazoniandark earths: origin, properties, management. Dordrecht,Netherlands: Kluwer Academic Publisher.

Mann CC. 2005. 1491: new revelations of the Americas beforeColumbus. New York, NY: Knopf.

Meggers BJ. 1954. Environmental limitation on the development

of culture.Amer Anthropol56: 80124.Meggers BJ. 2003. Natural versus anthropogenic sources of

Amazonian biodiversity: the continuing quest for El Dorado.In: Bradshaw GA and Marquet PA (Eds). How landscapeschange. Berlin, Germany: Springer.

Nepstad D, Carvalho G, Barros AC, et al. 2001. Road paving, fireregime feedbacks, and the future of Amazon forests. Forest EcolManage154: 395407.

Peres CA. 2005. Why we need megareserves in Amazonia.ConservBiol19: 72833.Pessenda LCR, Gomes BM, Aravena R, et al. 1998. The carbon iso-

tope record in soils along a forestcerrado ecosystem transect:implications for vegetation changes in the Rondonia state,southwestern Brazilian Amazon region.Holocene8: 599603.

Pessenda LCR, Gouveia SEM, Aravena R, et al. 2004. Holocenefire and vegetation changes in southeastern Brazil as deducedfrom fossil charcoal and soil carbon isotopes. Quatern Int114:3543.

Piperno DR and Becker P. 1996. Vegetational history of a site inthe Central Amazon basin derived from phytolith and charcoalrecords from natural soils. Quaternary Res45: 20209.

Piperno DR, Bush MB, and Colinvaux PA. 1991. Paleoecologicalperspectives on human adaptations in central Panama. I. ThePleistocene. Geoarchaeol6: 20126.

Roosevelt AC. 1980. Parmana: prehistoric maize and manioc sub-sistence along the Amazon and Orinoco. New York, NY:Academic Press.

Roosevelt AC. 1991. Moundbuilders of the Amazon: geophysicalarchaeology on Maraj Island, Brazil. San Diego, CA:Academic Press.

Roosevelt AC, Housley RA, Imazio da Silveira M, et al. 1991.Eighth millennium pottery from a prehistoric shell midden inthe Brazilian Amazon. Science254: 162124.

Saldarriaga JG and West DC. 1986. Holocene fires in the northernAmazon basin. Quaternary Res26: 35866.

Sanford RL, Saldarriaga J, Clark KE, et al. 1985. Amazon rainforestfires. Science227: 5355.

Santos GM, Gomes PRS, Anjos RM, et al. 2000. 14C AMS datingof fires in the central Amazon rain forest.Nucl Instrum Methods172: 76166.

Silman MR. 2007. Plant species diversity in Amazonia. In: BushMB and Flenly JR (Eds). Tropical rainforests and climatechange. Chichester, UK: Praxis-Springer.

Silman MR, Terborgh JW, and Kiltie RA. 2003. Population regula-tion of a dominant rainforest tree by a major seed predator.Ecology84: 43138.

Smith A. 1990. Explorers of the Amazon. Chicago, IL: Universityof Chicago Press.

Stebbins GL. 1974. Flowering plants: evolution above the specieslevel. Cambridge, MA: Belknap Press.

Stuiver M and Reimer PJ. 1993. Extended 14C database and revisedCALIB radiocarbon calibration program. Radiocarbon 35:21530.

Tardy C. 1998. Paloincendies naturels, feux anthropiques, et envi-ronments forestiers de Guyana Franaise du tardiglaciare l'Holocne rcent (PhD dissertation). Montpellier, France:Universit Montpellier 2.

Terborgh J, Feeley K, Silman MR, et al. 2006. Vegetation dynamicsof predator-free land-bridge islands.J Ecol94: 25363.

Thompson LG. 2000. Ice core evidence for climate change in thetropics: implications for our future. Quaternary Sci Rev 19:1935.

Walker JH. 2000. Raised field abandonment in the upper Amazon.Cult Agric22: 2731.

Weng C, Bush MB, and Athens JS. 2002. Holocene climatechange and hydrarch succession in lowland AmazonianEcuador. Rev Palaeobot Palynol120: 7390.

465


Documents

Bush & Silman Amazonia Revisited