Chapterhead: Detail of Paracas embroidered textile (drawing courtesy of Ann Peters).
7. The Huarango:
The Genus Prosopis on the South Coast
How may we unravel the profound consequences to human ecology of the long and gradual
story of deforestation that we have traced through the archaeology and history of the south
coast of Peru? Naturally, we should start by better understanding the tree that, along with
humans, dominates its riparian ecology: the huarango. These are a species of the genus
Prosopis, one of the most common plants found along the watercourses of New World
deserts and members of a family of nitrogen-ﬁxing, bean-producing plants—the legumes—
whose importance to humankind is second only to that of the cereal grasses and with which
our relationship is, if anything, even older.
Fortunately for our purposes, because of its importance in arid environments, Prosopis
has been widely studied by agroforesters, botanists and ecologists. We can draw upon this
work to supplement our own observations in the last fragments of old-growth woodland that
remain on the Peruvian south coast. Today, as we will see, perceptions of the genus are
deeply divided between appreciation of its value on the one hand, and intense dislike of it as
a thorny, invasive weed on the other. In this chapter we will sift through the reasons for this
and a history of misidentification, so as to identify the particular characteristics of the
huarango and, thereby, its true value as a human resource in the past. Indeed, we will go on
to construe that thousands of years of co-evolution with humans have left their mark on the
tree’s form on the south coast of Peru.
Problems of Taxonomy
The quarter of the earth’s land surface that is sub-tropical and dry is dominated by trees
of two genera: Acacia and Prosopis (Geesing et al. 2000). These nitrogen-fixing, semi-
deciduous, drought-resistant trees occupy more than three million square kilometres
(Pasiecznik et al. 2001: 2). Both Acacia and Prosopis are of the same sub-family,
Mimosoideae, having small, compound leaves and producing small flowers clustered into
inflorescences, within the family Leguminosae (or Fabaceae), whose members are popu-
larly known as legumes, since they produce seeds in legumes or pods. Plants of this large
family, whose seeds are relatively rich in protein and include lentils, peas and beans, have
an importance to human diet second only to the cereal grasses (see for instance Vaughan
and Geissler 2000: xviii). And the time-depth of our relationship with them, and in partic-
ular the tree legumes, stretches far back into our hunter-gatherer past (see Jones 2009 and,
for the significance of Prosopis in very early human diets in Mesoamerica, Smith 1967:
241, Flannery 1968, Callen 1969: 239).
While more than a thousand species of Acacia are recognised worldwide, Prosopis is
but a small genus—represented by only some forty-four species. It is also predominantly an
American genus, 90 per cent of its species being native to the New World (Simpson and
Solbrig 1977, Catalano et al. 2008). Yet the taxonomy of Prosopis is still confused and sub-
ject to ongoing revisions.
This complicated taxonomy (or ‘systematics’) arises precisely because of the adaptive
flexibility necessary to inhabit arid environments. Most Prosopis species, and even individ-
uals, show great plasticity in form (or ‘phenotype’), in response to environmental fluctua-
tions (Ibrahim 1992). Furthermore, species frequently hybridise because they are
genetically polymorphic, meaning that they have few chromosomal differences (Solbrig et
al. 1977). These flexibilities convey the advantages of relatively quick physiological and
morphological adaptation for plants living in the patchy habitats and climatic unevenness of
deserts (see for instance Harris and Cambell 1981). But for botanists concerned with classi-
fication, they make for complication. Indeed, Pasiecznik et al. (2001: 2) describe the history
of the nomenclature of Prosopis as ‘one of conflicting definitions and changing systematics
leading to great confusion in the definitions of species’.
As we have seen, these first confusions began in Peru when the Spanish conquerors
called the tree el algarrobo de las Indias, because it resembled another legume, the carob
tree of the Mediterranean basin (Ceratonia siliqua). ‘Algarrobo’ is today the vernacular
name for trees of the genus Prosopis throughout South America. The name ‘huarango’ is
restricted to the Peruvian south coast, although, again confusingly, on the north coast this is
the common name for another Mimosoideae tree, Acacia macracantha (see for instance
Brack Egg 1999: 13). In the United States, meanwhile, various Prosopis species are known
Scientific classification of Prosopis began with Linnaeus, who assigned the name in
1767 to the single species of which he was aware: P. spicigera (now synonymous with P.
cineraria), one of the few Old World species (Fagg and Stewart 1994). The derivation of the
name Prosopis is ‘towards abundance’, from the ancient Greek ‘pros’ for towards; and
‘Opis’, wife of Saturn and the goddess of abundance and agriculture (Pasiecznik et al. 2001:
19). It is purely serendipitous for our purposes here that ‘towards agriculture’ would be an
equally satisfying etymology.
130 THE LOST WOODLANDS OF ANCIENT NASCA
Burkart (1976) subsequently divided the genus Prosopis into five distinctive groups of
species or ‘sections’. Within these, species can be very similar and frequently hybridise. By
far the largest is section Algarobia, which includes more than half the New World species of
the genus, and wherein most taxonomic confusion lies. This led Burkart to believe that
Prosopis is an old genus, separated into principal lineages, within which episodes of isola-
tion have produced partial speciation (Burkart and Simpson 1977). Thirty-five of the forty-
four species recognised are native to South America, which suggests that this was the
plant’s place of origin (see for instance Burkart and Simpson 1977, Pasiecznik et al. 2001).
Indeed, this is confirmed by the latest molecular phylogeny studies (Catalano et al. 2008),
which date the origin of the genus to the beginning of the Oligocene, some 34 million years
ago, and link its subsequent spread and divergence to an expansion of arid conditions over
this great time-depth.
The history of Prosopis classification in Peru mirrors its wider problems, as Díaz Celis
(1995: 47; my translation) summarises: ‘the systematic treatment of the algarrobo in Peru
has been variable and confused’. This history can be followed through the work of
MacBride (1943), Weberbauer (1945), Burkhart (1976), Burkhart and Simpson (1977),
Ferreyra (1987), Díaz Celis (1995), Pasiecznik et al. (2001), Mom et al. (2002) and Harris
et al. (2003) (and see Beresford-Jones 2005 for a comprehensive review).
Rather unsurprisingly then, difficulties with Prosopis taxonomy abound also in the
archaeological literature. Yacovleff and Herrera (1934: 291) originally identified artefacts
of ‘algarrobo’ wood as Prosopis juliflora. Towle (1961: 56), likely following MacBride
(1943), considered this identification mistaken, claiming that ‘juliflora does not occur in
Peru’ and that the Peruvian equivalent is P. chilensis, ‘a plant very similar to P. juliflora and
often confused with it’. And most archaeological authors continue to follow Towle in iden-
tifying materials as P. chilensis (see for instance Menzel et al. 1964, Shimada and Shimada
1985, Silverman 1993, Cook 1999, Silverman and Proulx 2002).
Scrupulously, these identifications are almost certainly incorrect. P. chilensis is, in fact,
one of the more distinctive Prosopis species in its live condition, by virtue of its larger leaf
morphology (Pasiecznik et al. 2001: 33). None of the many trees that we observe on the
south coast of Peru today show the P. chilensis leaf type (Whaley 2004, Beresford-Jones
2005). Following Burkart and Simpson (1977), Díaz Celis (1995), Pasiecznik et al. (2001)
and Harris et al.’s (2003) systematics, and in agreement with most other current taxonomic
authorities, they would be identified as P. pallida (H. and B. ex. Willd.) H.B.K. (cf.
Recently, however, Mom et al. (2002) published the most authoritative description yet
of the Prosopis pallida populations of Peru. Their morphological analysis calls for the cur-
rent classification of P. pallida to be differentiated into two species and in so doing they
reintroduce the binomial originally used by Bentham (1875), and subsequently discarded by
Burkhart (1976), of P. limensis. Mom et al.’s (2002: Fig. 3) description of the form of P.
limensis is the most accurate that we have seen for the trees of the valleys of Ica and Nazca
(Whaley 2004, Beresford-Jones 2005, Whaley et al. 2010b). Indeed, Mom et al. (2002: 44)
7: THE HUARANGO 131
assign the name P. limensis to all the Prosopis collections they examine from the south coast
which include specimens from Ocucaje, Ica and Jumana and Poroma, Nazca. The type spec-
imen assigned the name P. limensis by Bentham is held in the herbarium of the Royal
Botanical Gardens. It is identified only as from ‘Lima’—hence the name assigned by
Bentham—yet this may well only be the place from which it was shipped. And its mor-
phology fits Mom et al.’s description for the rehabilitated P. limensis binomial (Oliver
Whaley, personal communication).
So, while I will continue here to use the encompassing binomial P. pallida—or the
wider so-called ‘P. juliflora–P. pallida complex’ (and see Pasiecznik et al. 2001 for back-
ground on this useful identification class)—for the purpose of corresponding with the exist-
ing botanical literature, the more precise classification of the huarango of the south coast
should almost certainly be P. limensis (Whaley et al. 2010a: 14). The distinctive features
defined by Mom et al. (2002: 43–4) as defining P. limensis—the huarango—include its
well-developed brachyblasts with distinctive resin ducts; leaves with very many, very small
leaflets (less than 7 mm long); and long, straight, yet also wide yellow pods. They are noted
in the desiccated Prosopis macrofossils recorded during our archaeological investigations
(see Figure 5.4).
Indeed, archaeological wooden artefacts from anywhere along the coast of Peru are
rather unlikely to be of P. chilensis (Molina) Stuntz, which is restricted to isolated popula-
tions in higher altitude, dry intermontane locations, mainly in the south of the country
(Burkart and Simpson 1977, Díaz Celis 1995, Brack Egg 1999, Pasiecznik et al. 2001), and
much more likely to be of the so-called ‘P. juliflora–P. pallida complex’. The botanist
Fortunato Herrera was certainly well aware of this when he originally identified coastal
algarrobos as P. juliflora (Yacovleff and Herrera 1934), since he had already identified the
Prosopis of the high intermontane valleys as P. chilensis (Díaz Celis 1995).
Fortunately, however, such specific identifications are unnecessary for many purposes
of archaeological interpretation. For all of them fall into Burkart’s section Algarobia, whose
members show strong similarities, even at the very detailed botanical levels of wood
anatomy,protein and enzyme characteristics, leaf physiology, and DNA (Pasiecznik et al.
2001: 28). Consequently, their ecological characteristics are broadly similar and, as Felgar
(1977: 150) puts it, ‘ethnobotanically it is both convenient and realistic to treat the North
American members of the section Algarobia as a unit’. Indeed, within the P. juliflora–P.
pallida complex itself, these similarities are so marked that species are, as Harris et al.
(2003: 154) note, ‘difficult to distinguish even for those with considerable experience in
semi-arid forestry’. Here, therefore, we may use appropriately selected data from studies of
the species of section Algarobia—and particularly from the P. juliflora–P. pallida complex
—to help understand the ecological importance and ethnobotany of the huarango on the
south coast of Peru. Nonetheless, more specific identifications will still be necessary for
some of our purposes here.
Until very recently, for instance, many North American authors continued to follow
Bentham (1875) in assigning the one binomial Prosopis juliflora to all the plants long since
132 THE LOST WOODLANDS OF ANCIENT NASCA
ascribed by Burkhart to thirty different species within the New World section Algarobia
(Fagg and Stewart 1994, Pasiecznik et al. 2001). Trees of this section, and in particular of
the P. juliflora–P. pallida complex, have been introduced to many exotic locations because
of their ecological and resource value in arid environments, such that today they are effec-
tively pan-tropically distributed (Figure 7.1; Fagg and Stewart 1994, Pasiecznik et al.
2001). Yet, all too often, inappropriate taxonomic classifications followed these introduc-
tions and thence were perpetuated in those exotic habitats. As Pasiecznik et al. (2001: 21)
caution, ‘initial misidentification without correction can be repeated continually’.
Sifting through these identifications is important because modern impressions of the
genus are deeply divided. For, while Prosopis has long been greatly valued within its native
ranges, exotic trees assumed to be ‘P. juliflora’ are now regarded as a serious invasive pest
in others, such as parts of India, the Sudan and Australia. Indeed, local distaste finds expres-
sion in their names: ‘viper’, ‘bastard thorn’ and dangerous thorn’ in the Sahel (Butterfield
1996); ‘Mexican thorn’ in Ascension Island; ‘white man’s thorn’ in Senegal; ‘Syrian thorn’
in Iraq; and vilayati babul (‘foreign thorn’) and ganda babul (‘mad thorn’) in India
(Pasiecznik et al. 2001).
Yet these names would seem quite incongruous to those who live with the genus on the
south coast of Peru, where the predominant form of the huarango has no thorns. Forms of
the tree with and without thorns are recognised locally as macho (male) and hembra
(female), respectively (Donaire 1998: 199). But thorny trees are rare and they mostly are
only saplings. On the north coast, meanwhile, Ferreyra (1987) and Díaz Celis (1995) also
recognise two forms of P. pallida: forma armata with thorns and forma pallida without.
And in countries to which plants of known Peruvian P. pallida origin have been imported,
such as Hawaii, Brazil and Haiti, Prosopis trees are also known as thornless (see for
instance Skolmen 1990, Lee et al. 1992). Indeed, this feature of the south coast huarango,
together with others that we will come across in this chapter, hints at possible deep-time
human–Prosopis relationships on the south coast of Peru.
The Form of the Huarango
The obvious problem faced by plants and, indeed, all life in arid regions is water availabil-
ity. The relatively few plants that survive in desert environments have evolved adaptations,
either to find other sources of water or to make more conservative use of available water,
to meet their metabolic requirements (Horton and Hart 1998). These include, for instance,
the thick cuticles and specialised photosynthesis (crassulacean acid metabolism, or
‘CAM’) of the cacti and succulents. Trees of the genus Prosopis, meanwhile, are so-called
‘phreatophytes’—plants with exceedingly long taproots that thereby access deep ground-
water unattainable by most plant species (Simpson and Solbrig 1977, Mooney et al. 1980).
They also have some so-called ‘xerophytic’ (or ‘dry-loving’) features and physiological
responses. Indeed, as we will see here, most of the features that define the genus Prosopis
7: THE HUARANGO 133
and the form of the south coast huarango reflect their remarkable adaptations to their
Describing the underground root architecture of a tree is obviously no easy matter.
Carefully excavating an entire tree is almost impossible, and roots can be difficult to
identify at any taxonomic level (Jackson et al. 2000). Nonetheless, records of the extents of
tree roots do exist and, perhaps unsurprisingly, show the greatest measures in arid, tropical
environments (Canadell et al. 1996). Some roots reach apparently extraordinary depths.
Well-diggers in the Kalahari Desert, for instance, encountered roots at 68 metres depth,
often cited as the deepest roots thus far recorded (Schulze et al. 1998). Stone and Kalisz
(1991) survey the evidence of root depths for 211 woody plant species in 96 genera. Few
exceed 20 metres in depth. The greatest they record is for Juniperus monosperma, encoun-
tered in a mineshaft at 61 metres. Yet, for a Prosopis juliflora of only modest size, at 6
metres tall, Stone and Kalisz (1991: 80) report a root depth of 53 metres and lateral radial
extents of more than 19 metres.
Indeed, there are many reports of very deep Prosopis roots on the south coast of Peru.
Galera (2000: 195; my translation) notes that the lateral roots of P. pallida ‘develop at up to
two to three times the diameter of the tree’s crown’, and that ‘roots have been found up to
60 metres in length’. Félix Quinteros (personal communication) of Universidad de Ica, who
has spent a lifetime studying the huarango of the south coast, reports living, radial roots at a
horizontal distance of 81 metres from an individual P. pallida encountered during the dig-
ging of a well in the La Victoria district of Ica, and Oliver Whaley (personal communica-
tion) of the Royal Botanic Gardens at Kew has observed similar root dimensions. These put
on record that the roots of the south coast huarango are amongst the longest and deepest of
any tree species worldwide.
After germination, usually in short-lived conditions of temporary moisture, Prosopis
roots grow very rapidly and can reach a depth of 40 centimetres in eight weeks (Garg 1998,
1999, Pasiecznik et al. 2001). Once mature, its root architecture becomes ‘dimorphic’—two
distinct systems with different hydraulic functions (see for instance Sudzuki 1985b, Gile et
al. 1997, Galera 2000, Pasiecznik et al. 2001). One or two main taproots form a deep system
to access permanent groundwater. These can become very thick and, as we have seen, very
long (Figure 7.2D). Meanwhile, a second, shallower, radial lateral root system extends about
the tree, forming a dense, complex root mat (Figure 7.2B). These roots also show ‘xero-
phytic’ adaptations, for, as Simpson and Solbrig (1977: 15) put it, they ‘can extract water
from soils too arid for most plant species’. Reductions in soil water under Prosopis have been
recorded down to minus 15 bar (Mooney et al. 1977), enabling the tree to extract water held
even under very high matric forces. As we will see in the next chapter, this extraordinarily
deep and dimorphic root architecture has crucial significance, not only for the water relations
of the huarango itself, but also for the entire ecosystem of which it is part.
7: THE HUARANGO 135
136 THE LOST WOODLANDS OF ANCIENT NASCA
A Transverse section of huarango bough, Usaca.
B Huarango timbers (‘horcones’) sold in Ica.
C Excavated huarango trunk and main rootbole,
D Huarango rootbole excavated by carboneros, Ica.
E Lateral huarango roots exposed in riverbank,
(Photographs C and D by O. Whaley)
Figure 7.2. The roots and wood of the south coast huarango
Huarango leaves: poña
Yet, despite its specialised root system, Prosopis is still exposed to the stresses of dry desert
air and it has various adaptations to reduce excessive water loss through its leaves by tran-
spiration. Its leaves are highly variegated with an epidermal surface covered by a coating of
wax. Leaf morphology is one of the main features used to distinguish species within the
genus and P. pallida and P. limensis have very well-developed ‘brachyblasts’—the smallest
lignified leaf branches (cf. Mom et al. 2002)—and many more, much smaller leaves than
other species. These leaves are ‘bipinnate’—split into pairs of pinnae—each of which has
up to twenty-nine pairs of tiny leaflets less than 7 mm long (see Figure 7.3; Burkart 1976,
Díaz Celis 1995, Pasiecznik et al. 2001).
This mass of brachyblasts and tiny leaflets in the huarango will usually become
enmeshed into ‘dreadlocks’ of spider-webs and dead tree litter material (Figure 7.4A and B).
Whaley (personal communication) observes that this may confer twofold advantages in an
arid climate. Firstly, the air within the dense brachyblast/leaflet cluster itself forms a micro-
climate that reduces transpiration. And secondly, these clusters create a greatly expanded
surface area upon which high night-time humidity may condense and thereby be captured
by the plant through mechanisms that we will examine further in Chapter 8. The highly var-
iegated leaf canopy of the tree nonetheless allows much more sunlight to filter through than
would larger, simpler leaves. And as we will also see, this has great significance for the
growth of other, so-called ‘understory’ plants, elegantly expressed by the great Peruvian
writer José María Arguedas—‘los guarangos dejan pasar el sol, pero quitándole el fuego’
(the huarangos let the sun pass, but take out its fire—my translation).
Like the wood of many slow-growing desert trees, that of the huarango is extremely fine-
grained, heavy, strong and hard (see Figure 7.2C). Indeed, Rogers (2000: 74) calls Prosopis
‘one of the world’s best woods’. It is also a very high-quality fuel. These characteristics,
which derive from its structure and chemical constituency, underlie much of its value as a
human resource and, thereby, its occurrence in the archaeological record.
Prosopis heartwood is red in colour, darkening with drying and quite distinct from the
much lighter, cream sapwood (see Figure 7.2A). This is because this heartwood has more
than twice the content of resins, tannins and oils of most hardwoods. These contribute both
to its quality as a fuel and to its virtual invulnerability to fungi and insect attack (see for
instance Goel and Behl 1996, 2001, Rogers 2000, Pasiecznik et al. 2001). Its growth rings
are lightly distinct, irregular dark bands (Acevedo Mallque and Kikata 1994).
The mechanical properties of the tree’s wood derive from its cellular structure. This is
described by wood anatomists using thin sections cut as precisely as possible along three
perpendicular planes, which, when considered together, describe the microscopic three-
dimensional structure of wood (see for instance Schweingruber 1978). Acevedo Mallque and
7: THE HUARANGO 137
Kikata (1994) and Moutarde (2006) provide anatomical descriptions of P. pallida wood from
the coast of Peru. By way of brief summary the features of huarango wood are as follows.
The wood pattern is diffuse porous (meaning that the diameters and frequency of vessels are
more or less the same throughout the season’s growth and thus across a growth ring). Vessels
(or pores, the xylem tubes of the tree’s vascular system) are few, solitary and small, with
simple perforation plates. Parenchyma cells (the tree’s storage cells) are abundant and non-
storied. Medullary rays (which give the wood structural strength in the radial plain from pith
to outer cortex) are homogeneous, multiseriate and non-storied. Fibres (which give the wood
its vertical structural strength) are abundant, short and non-storied with very thick walls.
Gums and chambered crystals are usually present in vessels, parenchyma and rays. Together,
these microscopic features—small and solitary vessels; abundant, narrow but very thick
walled fibres; and numerous, multiseriate rays—mean that huarango wood is heavy and
extremely hard with great tensile strength (D’Antoni and Solbrig 1977). Moreover, its timber
has strength in all three dimensions and has excellent—that is, low and uniform—shrinkage
138 THE LOST WOODLANDS OF ANCIENT NASCA
Figure 7.3. Nomenclature of Prosopis parts (adapted from Simpson et al. 1977, Solbrig et al. 1977 and Pasiecznik et
characteristics (Rogers 2000). The density of wood, expressed as its specific gravity relative
to water, encapsulates many of these characteristics in a single measure. They will vary, of
course, with growing conditions. Within the genus, the densest woods are those of the arid
Peru–Chile Desert: P. tamarugo,P. chilensis and P. pallida (Pasiecznik et al. 2001). The
specific gravity of P. pallida from the Peruvian coast is 0.88 (Acevedo Mallque and Kikata
1994), as compared with around 0.75 for English oak. Indeed, huarango wood is over one
and half times as hard as teak (Tectona grandis) and is much harder and stronger than oak,
mahogany or walnut (Ibrahim 1992, Pasiecznik et al. 2001: 79).
These remarkable mechanical properties of Prosopis wood cannot be better illustrated
than by the uses to which it has been put. In the early 1900s, timber blocks of Prosopis were
used to pave the streets of San Antonio, Texas (Rogers 2000) and Buenos Aires, Argentina
(D’Antoni and Solbrig 1977). These still underlie the asphalted road surface of many streets
of these cities. Even more extraordinarily, the first shots of the Mexican war of independ-
ence were fired from a cannon that the insurrectionists, lacking artillery, made from a large
hollowed-out trunk of mesquite. The wooden cannon worked well and is now an exhibit at
the National Museum in Mexico City (Rogers 2000).
Its high density and tannin content also make Prosopis wood such an excellent fuel that
it has been called ‘wooden anthracite’ (Pasiecznik et al. 2001: 4). It does not spit, spark or
emit much smoke but has a very high heat of combustion with a calorific value of around
5,000 kcal/kg (Arya et al. 1992) and thus burns with a hot, even heat. Today, Prosopis wood
is an important source of fuel for millions of people in arid areas worldwide (see for
instance Díaz Celis 1995, Lea 1996, Pasiecznik et al. 2001). Its green wood burns well
when freshly cut so Prosopis firewood does not require storage and drying (Pasiecznik et al.
2001). Indeed, such is its flammability that the charcoal burners (or carboneros) of the
Peruvian south coast report that large huarangos cannot easily be felled through controlled
use of fire, since it is impossible to prevent the entire tree ‘exploding’ into flame.
Burning Prosopis wood under anaerobic conditions produces a high-quality charcoal.
This is lighter and less bulky to transport than firewood: around five kilograms of wood
must be burned to produce one of charcoal (Pasiecznik et al. 2001). Today on the coast of
Peru, charcoal burning provides livelihoods for those with few economic options through
the supply of distant urban markets, rather than providing a domestic fuel source in the area
of production itself. It is still manufactured here in traditional earth ovens. Stacks of wood
cut into equally sized pieces are moistened and covered with soil to form the oven before
being fired (Figure 7.5). Burning conditions within are carefully maintained over a period
of about a week through the opening and closing of air holes. Once an even burn has been
achieved throughout, the oven is broken open and allowed to cool, before bagging and sale
to middlemen for transport and sale in Lima. And, whereas for domestic firewood small
branches are preferred (Pasiecznik et al. 2001), for charcoal the greatest efficiencies are
obtained by selecting the largest available trees and removing them in their entirety, includ-
ing their main root wood, the consequences of which are written in the age profile of the
huarango population of the south coast today.
140 THE LOST WOODLANDS OF ANCIENT NASCA
7: THE HUARANGO 141
A Felling huarango with chainsaws.
B Wood logged and ready for burning.
C Charcoal oven.
D Charcoal prepared and bagged.
Figure 7.5. Producing charcoal from huarango, Usaca woodland
The fruit of Prosopis on the south coast: miskyhuaranga
In a desert environment of limited and erratic water supply, Prosopis needs to disperse its
seed away from the parent tree to locations where they may successfully germinate. To this
end the huarango produces its seeds within copious quantities of sweet, highly nutritious,
bean-like fruit—known as miskyhuaranga on the south coast. Their structure—in three dis-
tinct layers, much like the common bean—and their composition are consequences of an
adaptation for dispersal by animals. For these pods—or legumes—are indehiscent, that is,
covered with a thin, leathery ‘exocarp’, which does not split open on maturity, but instead
falls from the tree as a single structure. Within this is a fleshy ‘mesocarp’, high in carbohy-
drates and proteins, within which are a number of stony ‘endocarp’ segments, each contain-
ing a soft, oval, brown seed. The sugars and starch in the mesocarp make the pods highly
palatable and they are eaten avidly by both wild and domesticated mammals and various
types of reptile, bats and birds (Mooney et al. 1977, Kingsolver et al. 1977, Ansley et al.
1997, Pasiecznik et al. 2001). The stony endocarp, meanwhile, protects the seed from ani-
mal chewing. Passage through the digestive tract then scarifies the seed, which enhances
germination. In due course it is deposited in animal dung, providing an immediate source of
nutrients for the young seedling. Germination occurs only upon the reception of suitable
cues such as during seasonal river flows. Indeed, water itself is also an important agent of
dispersal: both pods and seed endocarps float and survive extended periods, even in seawa-
ter (Pasiecznik et al. 2001). Desert river flows are erratic, of course, and, if necessary,
Prosopis seed can remain dormant and viable for up to forty years (Ibrahim 1992). Once
germinated, the seedling dedicates much of its initial energy to finding groundwater,
thereby developing its extraordinary root system. This mechanism of dispersal likely
evolved with large American mammals such as camelids, mastodons and giant ground
sloths (Mooney et al. 1977). Most of these mega-fauna became extinct towards the end of
the Pleistocene, perhaps owing to the first arrival in the Americas of human hunters (see for
instance Martin 1967), and today domesticated livestock such as cattle, horses and goats are
the primary agents of dispersal for Prosopis throughout most of its American distribution
ranges (Fisher 1977, Pasiecznik et al. 2001). These animals are, of course, relatively recent
introductions, following contact with the Old World and, as we will see, they have wrought
considerable changes upon the distribution and habit of Prosopis.
Yet, while adaptation for animal dispersal is common to all species of Prosopis, there is
nonetheless great variation between them in the size and composition of their legume pods.
Because of their nutritional value, Prosopis pods have been the subject of considerable
research and studies consistently note that Peruvian coastal accessions of the P. pallida–P.
juliflora complex produce larger pods, higher in sugars, carbohydrates and proteins, in
much larger numbers than do other members of the genus. Pasiecznik et al. (2001: 85–6),
for instance, review the results of thirty-two such studies from locations worldwide of the
pods of the P. pallida–P. juliflora complex, and conclude that these ‘suggest that Peruvian
Prosopis pods are of an inherently better quality than pods from P. juliflora of the northern
142 THE LOST WOODLANDS OF ANCIENT NASCA
races’. In Peru itself, research is ongoing at the Algarrobo Research Project at the
Universidad de Piura, on the far north coast. Here the composition of P. pallida fruit-pulp is
shown to be 46 per cent sucrose and 8–13 per cent protein with significant amounts of vita-
mins E and C and potassium, while the seeds themselves are some 36 per cent protein, one
of the highest levels for any legume (Grados and Cruz 1996). Meanwhile, Lee et al. (1992:
11) report that in Brazil P. pallida of Peruvian accession have ‘the longest pods we have
ever observed (about 50 cm)’. Pods of the south coast huarango are between 10 and 45 cm
long (Figure 7.4D).
Furthermore, Peruvian P. pallida trees are more productive than North American
Prosopis species, which typically yield between 19 and 34 kilograms per mature tree per
year (Ibrahim 1992). By contrast, exceptional P. pallida trees on the north coast of Peru
yield up to 300 kilograms (Grados and Cruz 1996) and fruit productions of up to 420 kilo-
grams are reported from P. pallida trees of Peruvian accession in Brazil (Pasiecznik et al.
2001: 93). Although P. pallida on the north coast of Peru produce fruit once a year in June
and July (Díaz Celis 1995), on the south coast we see P. pallida usually producing two fruit
crops in a year: once during a main season between December and March when rivers flow;
and again, in a lesser production season—‘la cosecha de San Juan’—between June and
July, when winter air humidity is at its highest. Figure 7.4E shows the pod harvest from a
single ten-year-old tree in the Samaca Basin at the end of the fruiting season in April 2002.
Accessing deep groundwaters P. pallida here produces fruit almost regardless of the
highly erratic hydrology of the Río Ica (cf. Simpson et al. 1977). This ability to produce
copious, nutritious fruit for animal fodder or human consumption, somewhat independent
of short-term climatic variations, underpins the deep-time importance of the genus to
humankind in arid areas of the New World. Indeed, as Felker (1981) notes, no tree legume
has been more widely used as a food source.
The size and form of the huarango
How an individual Prosopis tree grows and attains its eventual form is, naturally, a matter
both of its genetic makeup and of the environment in which it finds itself. Prosopis growth
rates therefore vary widely between species, and within populations (see for instance
Pasiecznik et al. 2001). Agroforestry trials using P. pallida trees of Peruvian accessions
show faster growth rates, attaining larger sizes on maturity, than other species of the genus.
Trials in Senegal, Haiti and India compare the growth of various Prosopis species from
diverse origins in Argentina, Chile, Haiti, Mexico, Peru and the United States and record
that those of Peruvian origin grew significantly faster, taller and straighter than others,
including those of native origin to the trial area (Lee et al. 1992, Felker and Patch 1996,
Harris et al. 1996, 2003, Harsh et al. 1996, Pasiecznik et al. 2001, Deans et al. 2003). Harris
et al. (1996: 4–27), for instance, conclude that ‘Peruvian genetic stock is near the top in
evaluations in three distinctly different environments: Haiti, Cape Verde, and the interior
deserts of India’.
7: THE HUARANGO 143
Tree growth can also be greatly affected by human intervention. The practices of
pruning, coppicing and pollarding have been used in the management of woodland
resources for millennia. Such adjustments to tree architecture are employed to diverse
ends (see Reynel and Felipe-Morales 1987 for details in Peruvian contexts), but, in gen-
eral, they greatly augment the rate of production of woody biomass by placing the tree
under stress. And Prosopis is no exception. Elfadl and Luukkanen (2003: 448) report, for
instance, that heavy pruning of P. juliflora in the Sudan increases their wood production
by up to six times.
All these studies of the growth rates of Prosopis, however, pertain to rather young
trees. Rapid initial growth is of obvious advantage to plants seeking to establish them-
selves successfully in unpredictable desert environments. But data on the longer-term
growth of Prosopis species are largely lacking and, typical of many desert hardwood trees,
the huarango, as we will see shortly, can live for a very long time. Today, the form (or
‘habit’) of Prosopis, throughout its New World range, is usually of dense stands of desert
shrub or shrubby trees (Figure 7.6A). And in some grassland areas it has become estab-
lished as a noxious, invasive woody weed. However, conflicting perceptions of the genus
in appreciation of its value, on the one hand, and intense dislike as a thorny, invasive weed,
on the other, are mostly the result of relatively recent human disturbances. For, as
Prosopis species of section Algarobia have had their distribution expanded by deliberate
human introductions worldwide, so too have they extended subsequently in some new
locations, through invasions. This success, of course, derives from those same features of
morphological and genetic plasticity that so well adapt Prosopis to desert environments, or
what Pasiecznik et al. (2001: 13) call its ‘broad ecological amplitude’. Indeed, so success-
ful is Prosopis that, once established in arid environments outside its native range, it can
become impossible to eradicate. So Peattie (cited in Pasiecznik et al. 2001: 1) describes
mesquite on the Texas rangelands as ‘something more than a tree, it is almost an elemental
force, comparable to fire—too valuable to extinguish completely and too dangerous to
On the cattle rangelands of the south-west United States and north-west Argentina, for
instance, species invasions by thorny Prosopis scrub have spread far and wide. Ranchers
have responded with attempts at mechanical and herbicide control which damage the tree but
lead to heavy, multi-stemmed regrowth from the trunk base. Decades of these attempts have
succeeded only in creating ever-denser thickets of shrubs (Ansley et al. 1997) and thousands
of hectares have thereby been rendered useless for cattle grazing. Although geographically
distant the Argentinean and United States rangelands share a common land-use history.
Beginning around 1850, cattle numbers in these areas increased by many orders of magni-
tude. Along with this came overgrazing of grass cover and severe soil surface disturbance, all
of which created ideal conditions for rapid invasion by Prosopis—often dispersed by the
livestock themselves (Fisher 1977). These problems come as little surprise to agroforesters.
As Felker and Patch (1996) have shown, the only secure defence against invasions by dense
stands of Prosopis shrubs are large Prosopis trees. Large trees accomplish self-thinning,
144 THE LOST WOODLANDS OF ANCIENT NASCA
intra-specific competition, preventing encroachment by younger saplings. So, in less
disturbed Prosopis woodland, the tree population decreases as larger and larger trees come
to dominate the stand. And since mature Prosopis trees obtain water from deep horizons,
they offer far less inter-specific competition than do dense stands of small trees. Indeed, as
we will see in Chapter 8, they exert a host of soil (or so-called ‘edaphic’) effects, which com-
bine to stimulate understory herbaceous growth, in direct proportion to the age of the
The coast of Peru, meanwhile, presents an entirely different environment to that of the
American rangelands. Sufficient rainfall to support extensive dry forest savannah prevails
only in the far north near the border with Ecuador. Elsewhere, Prosopis is limited to isolated
riparian habitats scattered over one of the world’s driest deserts. Here, as we have seen,
human impact has also been dramatic, but with very different causes and effects. Indeed,
there are, as Pasiecznik et al. (2001: 11) observe, ‘no records of P. pallida as a weed in its
native range’ (2001: 11). Instead, here Prosopis deforestation has led often to processes of
desertification. For in this hyperarid desert, human disturbance may extinguish completely
even the ‘elemental force’—as Peattie called it—of Prosopis. Or it may prevail, but with a
dramatically altered age profile, dominated now by small trees and shrubs. This, in turn, has
led to the oft-used term of ‘scrub-forest’ (or its local variant monte), to describe the riparian
vegetation of the south coast (Figure 7.6A). Yet, just as with the Prosopis invasions of
grasslands elsewhere in the Americas, this monte scrub of the south coast of Peru is an arte-
fact: the product of human action. In fact, as we will see presently, given time, the huarango
of the coast of Peru are enormous trees.
Relict Huarango Forms on the South Coast of Peru
Although the vast majority of its huarango population has been reduced to monte scrub,
there are still, at the time of writing at least, some tiny fragments remaining on the Peruvian
south coast of far, far more ancient Prosopis habits. These include scattered numbers of
exceptional older, isolated trees protected by accidents of fortune; some extensively man-
aged mature woodland fragments along stretches of the lower Río Nasca; and a single relic
of what might be termed ‘old-growth’ Prosopis woodland, in the Quebrada Usaca. The
Prosopis species of the Peru–Chile Desert, P. pallida,P. juliflora, P. chilensis and P.
tamarugo, all may become substantial trees of between 15 and 20 metres high (Brack Egg
1999, Díaz Celis 1995, Grados and Cruz 1996). But, because of its extraordinary form, the
measure of tree height quite fails to convey the true size of a mature huarango on the south
As we have seen from the studies showing the superior growth and form of Peruvian
coastal Prosopis in locations worldwide, much of its variation in habit is genetically con-
trolled. Yet, as Harvard (cited in Pasiecznik et al. 2001: 59) observed in 1884, ‘there is
hardly any soil, if it is not habitually damp, in which mesquite cannot grow; no hill too
7: THE HUARANGO 145
146 THE LOST WOODLANDS OF ANCIENT NASCA
A Huarango dominated monte scrub, Samaca, Río Ica.
B Prostrate form huarango, Cahuachi, Río Nasca.
C Mature huarango in full foliage, Río Ica.
D Three year old sapling growing on yapana, Samaca, Río Ica.
E Huarango growing on rock, Río Poroma.
F Erect form, twenty year old huarango, Hacienda Elias, Ica.
G Erect form huarango, Cahuachi, Río Nasca.
Figure 7.6. The forms of the huarango on the south coast of Peru
rocky or broken, no flat too sandy or saline, no dune too shifting . . . to entirely exclude it’.
And this ability of the genus to persist in the most variable of sites naturally gives rise to a
huge variety of tree forms (Figure 7.6). Forms also vary, of course, with maturity. While
younger trees have erect branches, as the huarango gets older its boughs become ever larger
and more decumbent. In time they come to evoke Calancha’s 1639 description of the
huarangos of old Ica as ‘tall, growing sideways as contorted mountains’.
One such ancient individual tree grows near Palpa, alone on a relict river terrace of the
Río Santa Cruz, some 40 metres above the current floodplain. Known locally as ‘El
Huarango Milenario’, its age was determined in the 1980s to be 1,064 years old by investi-
gators from the universities of Huamanga and La Agraria, La Molina through counting
growth rings on a damaged main bough (Carlos Reynel, personal communication). This tree
has a short trunk of tremendous girth (14.4 metres in circumference and almost 5 metres in
diameter) that divides into seven huge decumbent boughs, which extend to a maximum of
43 metres from the main trunk. As Figure 7.7 shows it is well-nigh impossible to capture a
tree of such dimensions in a single photograph.
Indeed, the extraordinary seven-stemmed form of the Huarango Milenario has all the
hallmarks of a tree subject to human interference in its youth. For it has a striking resem-
blance to those oldest markers of English boundaries: an ancient coppice stool (Rackham
1980). The location of this ancient huarango on the edge of the main Río Santa Cruz flood-
plain also marks the entrance to a narrow quebrada in which lies a major Late Intermediate
Period settlement site: Huayuri (see for instance Mächtle et al. 2009). The suggestion that
the form, location and even preservation of the Huarango Milenario may identify it as an
artefact and landscape marker is of course highly speculative. Yet nonetheless, this ancient
tree certainly stands as a powerful illustration of the different timescales over which
human–huarango relationships need to be evaluated. For it was a sapling during the Middle
Horizon which preceded the Huayuri occupation and was already old when the Inca
Pachakuti Yupanqui conquered the south coast, some four hundred years later. It is still, just
about, alive today.
The age of the Huarango de Huayuri also offers us a ‘rule-of-thumb’ for Prosopis
growth rates on the south coast of Peru. The tree shown Figure 7.6F, for example, has a
trunk diameter of half a metre and a known age of twenty years, equating to a 2.4 centime-
tre diameter per year growth rate (and see Felker and Patch 1996). Of course, growth rates
will slow with age. The ancient trunk of the Huarango de Huayuri indicates an average
growth rate over its tremendous lifespan of only 0.43 cm per year. Figure 7.8 then illustrates
this expected decrease in growth rate with tree age. Obviously this estimate is very crude:
an extrapolation between two points of data, which ignores any number of possible local
environmental variations. Nonetheless, it may be useful, since controlled agroforestry trials,
even over the thirty years that some Indian studies now approach, can of course only be
conducted over human timescales.
Aside from scattered giants such as the Huarango de Huayuri, there are also still today
some tracts of secondary huarango woodland along the lower reaches of the Río Nasca.
7: THE HUARANGO 147
148 THE LOST WOODLANDS OF ANCIENT NASCA
A Detail of main trunk.
B Tree in context; note person for scale.
C Trunk and procumbent boughs.
Figure 7.7. The Huarango of Huayuri (or the ‘Huarango Milenario’), Río Santa Cruz, Nazca
This river runs west, along the southern edge of the Nazca Pampa, famous for its concen-
tration of geoglyphs. The Río Nasca provides a perfect example of the misleading impres-
sions of hydrological poverty caused by surface river flows on the south coast, which we
saw in Chapter 2. For on the surface it appears an unimportant tributary of the Río Grande
de Nazca: comprising a mere seven per cent of its eventual total surface flow (see Table
2.4). And yet, land under irrigation today in the Río Nasca drainage is some 15,000 acres,
much more than that of the Ingenio, Upper Grande and Palpa rivers combined. Its sub-
surface water flow is clearly very significant. The introduction of deep tube wells in the
1930s has allowed increased exploitation of this groundwater and agriculture expansion on
the upper reaches of the tributaries of the Río Nasca, but archaeological investigation shows
that the significance of its groundwater flow was also understood and exploited in the
pre-Hispanic past (Rossel Castro 1977, Schreiber and Lancho Rojas 1995). Phreatophytic
huarangos are, as we have seen, adapted to accessing these hidden water supplies. So it is
unsurprising that the last vestiges of riparian woodland on the south coast are to be found
precisely along the lower stretches of the Río Nasca and its tributaries. The forms of these
woodlands are, however, changing fast.
Alongside the Nasca site of Cahuachi, ‘the place where water appears’ (Silverman 1993:
12), is a grove of huarangos of about a century in age, protected, appropriately enough, by
the guardian of the archaeological site (Figure 7.6B and G). Its trees, however, have lost all
foliage and may be diseased (Oliver Whaley, personal communication). Travelling further
7: THE HUARANGO 149
0 200 400 600 800 1000 1200
AG E (YEARS)
MAIN TRUNK DIAMETER (C M )
Estimated average growth rate
Growth rate 0–20 years
Growth rate 0–1,000 years
Figure 7.8. ‘Rule of thumb’ estimate of huarango growth rate on the south coast of Peru
downstream along the Río Nasca, however, there are still quite extensive stretches of sec-
ondary riparian woodland. In Majuelos and Jumana, for instance, the landscape remains
dominated by woodland. As recently as the early 1990s most of these woodlands were
divided between families of goat herders who, depending upon Prosopis fruit for their live-
stock, resisted the deprecations of carboneros and protected seedlings from their animals
(Figure 7.9C). Since then, however, large areas have been cleared for cotton growing and in
the long dry season they are now full of charcoal burners. The centurial tree photographed
in 1995 (Figure 7.9B), for instance, no longer exists. So Jumana and Majuelos are today
currently undergoing the transition, already complete on the rest of the south coast, from
mixed-age secondary woodland to scrub woodland dominated by young trees and shrubs
A small tributary, the Río Poroma, joins the Río Nasca just upstream from Majuelos.
Up this tributary in Usaca is a fragment of what might be termed ‘old-growth’ woodland —
to our knowledge the last remaining relict of ancient woodland on the south coast of Peru.
The Quebrada Usaca is a steep, narrow valley, dominated to the south by an enormous wall
of dunes blown up by the winds from the mouth of the Río Grande. Along its higher edges
are extensive Nasca Period cemeteries (Isla 1992). The floor of the quebrada is, however,
covered with dense riparian woodland, which spreads up its flanks to considerable heights
above the Río Poroma watercourse (Figure 7.10A). Usaca has a tiny forest fragment, per-
haps 40 hectares. It is also clearly secondary woodland, considerably altered in form and
composition by human activity in its environs, primarily that of goat herding. It is,
nonetheless, perhaps the only and closest observable analogue left on the south coast for
the riparian woodlands which once characterised the region’s vegetation. Its woodland
includes Acacia and Salix trees of considerable size, but it is dominated by huarango, a few
of which have reached truly enormous proportions (Figure 7.10).
The largest are to be found growing along the outer edge of the woodland fragment and
have main trunks of up to 7 metres in circumference—or more than 2 metres in diameter—
and reach heights of around 30 metres. Their main boughs are decumbent, often draped up
the steep slopes of the quebrada’s edge. Sometimes these vast pendulous boughs replant
themselves at some distance from the main bough and have sprouted anew, such that an
individual tree may sprawl across an area of well over a thousand square meters. Our ‘rule-
of thumb’ estimate would conservatively put their ages at about four hundred years old.
Yet today, the somnolent Usaca woodlands are disturbed often by the harsh rip of the
chainsaw—but one more, perhaps the last, chapter in the long story that we reviewed in
Chapter 6. And the results are grim, dramatic and absolute (see Figure 7.5). When charcoal
burners worked with hand axes they avoided giant trees like those in Usaca, because of the
huge efforts required to fell them. The tremendously hard wood of the huarango takes its
toll even on chainsaws—they require hours of resharpening after every use. But a large tree
can provide several tonnes of high-quality charcoal. After centuries, the ancient giants of
the Quebrada Usaca are falling, to be burnt, bagged and transported to the oblivious barbe-
ques of faraway Lima.
150 THE LOST WOODLANDS OF ANCIENT NASCA
The Huarango: an Artefact?
Those general characteristics that we have seen appreciated by agroforesters for the
Prosopis species of Peruvian coastal accessions, and in particular those of P. pallida forma
pallida—or rather, following Mom et al. (2002), P. limensis—lead rather naturally to spec-
ulation that they could be the product of long co-evolution with humans. To recapitulate,
these characteristics include faster and more erect growth to much larger sizes, the produc-
tion of more copious quantities of larger, sweeter pods, and indeed the preponderance of
thornless forms, when compared with the rest of the P. juliflora–P. pallida complex, or
indeed with any of the rest of the genus Prosopis.
As we will see in further detail in Chapter 9, Prosopis is recognised as the single most
valuable wild plant food resource in the ethnobotany of all American arid environments.
Certainly the huarango is not a ‘domesticate’ or even in any strict sense a cultivar, for it
requires no human intervention to persist. Nonetheless, as has been long appreciated, the
control and cultivation of food sources (ultimately, agriculture) was not an ‘event’, but
rather a protracted co-evolutionary process between humans and parts of their environment
7: THE HUARANGO 151
A Majuelos woodland, Río Nasca, 1995.
B Century old huarango, Marjuelos, 1995.
C Livestock herding, Marjuelos, 1995 (note protected sapling in foreground).
D Open huarango woodland, Jumana, Río Nasca, 2002.
Figure 7.9. Huarango woodland in Majuelos 1995
152 THE LOST WOODLANDS OF ANCIENT NASCA
A View south of the Quebrada Usaca, Río
B D Ancient trees of the Usaca woodland.
Figure 7.10. Ancient huarango woodland, Quebrada Usaca, Río Poroma, Nazca
(Rindos 1980). The categories of ‘wild’ and ‘cultivated’ thus fall along a long continuum of
plant use by humans. Some plants growing wild—such as the Brazil nut—provide foods on
an industrial scale, while other cultivated food plants—such as the carrot—are rather simi-
lar to their wild progenitors (see for instance Vaughan and Geissler 2000).
This is particularly so with tree species. In the case of the olive (Olea europaea), for
instance, the precise phenotypic or genotypic differences between wild and cultivated vari-
eties remain unclear (see for instance Lumaret and Ouazzani 2001), except for the obvious
lack of thorns in the latter—typically the initial stage in any process of plant selection by
humans. This is partly because of complications arising through grafting between woody
plant species that do not naturally hybridise. But it is also because the individual lifespan of
an olive or Prosopis may begin to approach the duration of entire processes of domestica-
tion for other plants— such as the kiwifruit. So the living genotype of an ancient huarango
will preserve any influence of human selection from as far back in time as the Middle
As Pasiecznik et al. (2001: 5) obverve, ‘human influences including intentional intro-
duction, weedy invasions and deforestation over the last 500 years have significantly
altered . . . the native range of the P. juliflora–P. pallida complex’. And they go on to spec-
ulate that the isolated populations along the Peruvian coast today ‘may have been connected
before deforestation during the last 500 years’. Deforestation within riparian ecosystems on
the south coast of Peru is of course our theme here. Yet, while Pasiecznik et al.’s observa-
tion of greater connection between populations in the recent past may well be true over
much of the range of the P. juliflora–P. pallida complex in semi-arid regions of the coun-
tries north of Peru, it certainly cannot be true outside the riparian contexts of the south coast
of Peru. For these have been completely isolated by stretches of hyperarid desert over far
greater time-depths. Indeed, the isolation here of P. pallida populations within their riparian
contexts is further defined, on the one hand, by the Pacific Ocean, and on the other, by the
Andes mountain range. For, as Pasiecznik et al. (2001: 2) note, P. juliflora–P. pallida is a
truly tropical complex, ‘killed to the ground . . . at temperatures of only −6 °C’. So the range
of P. pallida on the Peruvian coast does not extend above 1,500 m of altitude (Brack Egg
1999: 414). Furthermore, possible seed dispersals by the ocean are limited along the south
coast because of its paucity of wide alluvial river deltas. Huarango populations of the ripar-
ian ecosystems of the south coast have therefore experienced particular isolation from the
larger population of the P. juliflora–P. pallida complex far to the north. Of course, such
restrictions do not present absolute barriers to Prosopis dispersal over the great time-depth
of its botanical evolution, but they certainly do represent relative barriers to dispersal,
hybridisation and backcrossing. More recently the coast of Peru has experienced five hun-
dred years of profound changes, including the introduction of many new animal agents of
seed dispersal from the Old World, the construction of transport infrastructure linking its
valleys in modern times and, indeed, the fading of the importance of Prosopis as a food
resource for humans. And yet the productive, thornless form of P. limensis—the huarango
—of the isolated riparian systems of the Peruvian south coast remains distinct, thereby rep-
7: THE HUARANGO 153
resenting, perhaps, relict populations, whose seed germ plasm—especially of ancient trees
—preserves uniquely here the effects of human selection over those past millennia during
which it was far more important to human subsistence than it is today. Indeed, as Pasiecznik
et al. (2001: 51) note, ‘disjunct populations near to ancient centres of population suggest
that man has also been a dispersal agent in historic and prehistoric times’. Certainly, it is
remarkable that the coast of Peru—the location in which Prosopis varieties exhibit all those
characteristics for which humans would, and still do, choose to select—is, coincidentally
that location whose archaeological record reflects the longest human cultural trajectories of
the entire Americas, and wherein sedentism and complex society seem to have developed in
Peru (see for instance Quilter and Stocker 1983, Dillehay et al. 2004, Heggarty and
154 THE LOST WOODLANDS OF ANCIENT NASCA