Vivid valleys, pallid peaks? Hypsometric variations and rural-urban land change in the Central Peruvian Andes.
Free full text in Europe PMC
Abstract
Free full text
Vivid valleys, pallid peaks? Hypsometric variations and rural–urban land change in the Central Peruvian Andes
Abstract
What happens to the land cover within the hinterland's altitudinal belts while Central Andean cities are undergoing globalization and urban restructuring? What conclusions can be drawn about changes in human land use? By incorporating a regional altitudinal zonation model, direct field observations and GIS analyses of remotely sensed long term data, the present study examines these questions using the example of Huancayo Metropolitano – an emerging Peruvian mountain city of 420,000 inhabitants, situated at 3260 m asl in the Mantaro Valley.
The study's results indicate that rapid urban growth during the late 1980s and early 1990s was followed by the agricultural intensification and peri-urban condominization at the valley floor (quechua) – since the beginning of Peru's neoliberal era. Moreover, regarding the adjoining steep slopes (suni) and subsequent grassland ecosystems (puna), the research output presents land cover change trajectories that clearly show an expansion of human land use, such as reforestation for wood production and range burning for livestock grazing, even at high altitudes – despite rural–urban migration trends and contrary to several results of extra-Andean studies.
Consequently, rural–urban planners and policy makers are challenged to focus on the manifold impacts of globalization on human land use – at all altitudinal belts of the Andean city's hinterland: toward sustainable mountain development that bridges the social and physical gaps – from the bottom up.
Highlights
► Huancayo Metropolitano's hinterland is undergoing major land cover change. ► Agricultural intensification follows rapid urban growth since the mid-1990s. ► Increasing activities of human land use reach to high-altitude puna grasslands. ► High-altitude zones need to be integrated into urban and peri-urban planning.
Introduction
Background and aims
Due to the influence of relief and altitude, applied research on land use and cover in mountain regions has been declared a major goal of the UNESCO-supported GLOCHAMORE research strategy (Björnsen Gurung, 2006). The latter aims to better understand the causes of global change and its impacts on mountains: in order to achieve sustainable development, the peculiar environmental (e.g. high mountain vs. valley climates) and socio-cultural diversity (e.g. indigenous vs. neoliberal Weltanschauung) should be considered in land change analyses – from the valley up to the peak.
With respect to the Central Andes, recent studies on hypsometric variations can draw on a large body of valuable literature that evolved over the last decades (Dollfus, 1982; Gade, 1992; Lauer, 1993; Mayer, 1979; Sarmiento, 2000; Stadel, 1992; Troll, 1968; Zimmerer, 1999). Especially within the Peruvian context, Pulgar Vidal's framework of “the eight natural regions” (Pulgar Vidal, 1946) – it has recently been examined by Zimmerer and Bell (2013) – represents a useful altitudinal zonation of land use and cover that is adopted for the present study. A common characteristic of the literature mentioned is the focus on remote rural areas – giving little importance to rural–urban interfaces. However, against the background of rapid urban growth (Bolay & Rabinovich, 2004; Hardoy & Satterthwaite, 2007; Satterthwaite, McGranahan, & Tacoli, 2010) and the increasingly fragmented structures and processes of urban land use in Latin America (Borsdorf, Bähr, & Janoschka, 2002; Borsdorf & Hidalgo, 2010), it becomes more and more important to understand land cover change – e.g. “farmscape” transformations (Sarmiento, 2008; Young, 2008) or settlement processes in risk zones (Haller, 2010; Tobin & Whiteford, 2002) – in the hinterlands of Andean and other mountain cities (Bertrand & Vanpeene-Bruhier, 2007; Thapa & Murayama, 2010), in order to strengthen the positive and to handle the negative impacts on these socio-culturally and environmentally diverse rural–urban continua (Zeleke, 2007).
Although the Peruvian sierra – especially the quechua altitudinal belt at 2500–3500 m asl (Pulgar Vidal, 1946) – is now undergoing major globalization-driven restructuring, the mountain research community is paying little attention to land change in this part of the Andes. Hence, the present case study generally aimed to enhance the scientific knowledge about this actual but little researched topic for supporting sustainable regional planning and development. Moreover, the specific aims were as follows: (1) to detect land cover change within the study area during the periods of 1988–1998 and 1998–2008; (2) to quantify and explain change trajectories, giving emphasis to the different altitudinal belts; (3) to interpret the rural–urban interactions against the background of direct field observations conducted in the second half of 2011. Thus, the study temporally focused on the last 25 years of global change – a period often highly relevant for applied research, planning and policy making (Yarnal, 1996). Epistemologically, the research process, which mainly followed the inductive method, was clearly situated at the nexus of empiricism, logical positivism and humanistic approaches (Gatrell, Bierly, & Jensen, 2012).
Study area
The Mantaro Valley in the Peruvian Central Andes is formed by the homonymous river that issues at approximately 4100 m asl near Lake Junín or Chinchaycocha (probably from Quechua chincha for “north” and qocha for “lake”) and flows southeast between the Western and Eastern Cordillera, passing Huancayo Metropolitano. The emerging agglomeration – once an “indigenous reduction” of several ayllukuna – is situated 20 km southwest of the Huaytapallana mountain range (from Quechua huayta for “flower” and pallar for “to harvest”) at 12°4’S and 75°12.5′W (urban center, 3260 m asl) on the Shullcas River alluvial fan (Fig. 1).
Huancayo Metropolitano and its hinterland with the quechua, suni and puna altitudinal belts. The map has been elaborated on the basis of 2008 Landsat 5 TM and Aster GDEM data.
It represents the valley's largest continuous settlement area and – with about 420,000 inhabitants distributed over seven districts in 2012 (Haller & Borsdorf, in press) – Central Peru's most populated mountain city. Regarding the soils' land use capability, the valley floor around Huancayo – mainly consisting of alluvial, fluvial and glacial Quaternary deposits (Instituto de Geología y Minería, 1975) – was considered suitable for the production of annual and biennial crops by the former National Office of Natural Resources Evaluation ONERN (Oficina Nacional de Evaluación de Recursos Naturales, 1981). Even though the annual precipitation around Huancayo Metropolitano reaches a multiannual average (1960–2000) of approximately 750 mm (Instituto Geofísico del Perú, 2005), the possibilities for agriculture are limited by the availability of irrigation water during the dry season.
Although the agglomeration's area is entirely located within the quechua altitudinal belt, there are several rural–urban linkages (e.g. production and commerce of agricultural goods, day excursions for recreation and tourism or commuting to urban-based employment) that make the study area reach at least to the border of the suni (3500–3800 m asl) and puna (3800–5200 m asl) altitudinal belts; the hamlet and ex-hacienda Acopalca (from Quechua aqo for “sandy” and pallqa for “valley”), for example, is situated at 3950 m asl, 9–10 km northeast of Huancayo Metropolitano. Hence, for the present study, the “rural–urban” is rather seen as the city and its hinterland as a whole – from purely rural to clearly urban areas – than an exclusively hybrid type of space (peri-urban or rurban).
Due to the environment's physical characteristics (marked relief, several altitudinal belts) and socio-economic dynamics (rapid urban growth, agricultural intensification), the valley section is eminently suitable for the present research's purposes. In order to define the study area, the agglomeration's continuously built-up area was manually demarcated on a 1998 Landsat TM composite (bands 6, 7, 5 in RGB; band 6 prior resampled to 30 × 30 m cell size) via on-screen digitization. Subsequently, the polygon was extended by a buffer (10 km) that served for calculating the minimum bounding rectangle; the latter limited the study area.
Material and methods
Data sources
For the objectives of the present multi-temporal analysis and due to the study area's extent, Landsat 5 TM satellite images (path 6, row 68) – acquired in 1988 (August 7), 1998 (August 3) and 2008 (July 13) – represented a well-suitable source of data; at an output scale of 1:100,000, the side of a square pixel would be 0.3 mm. As the images were acquired during the dry season, almost no cloud cover occurred and thus the effect of urban heat islands was well pronounced – an important fact for differentiating urban settlements from bare rock in mountain regions. In order to analyze altitudinal variations, Terra Aster GDEM elevation data (28 m resolution) were incorporated. Complementary primary (training samples for classification, GPS ground points for accuracy assessment) and secondary information – a topographic map (Instituto Geográfico Nacional; scale 1:100,000) and recent high resolution satellite images including metadata (12 scenes from GEOEYE-1 taken during 2009–2011; six from IKONOS-2 taken during 2007–2010) – were acquired in situ or searched and displayed using the GeoFUSE Toolbar for ArcMap™ 10 and GeoFUSE Google Earth™ Integration Tools.
Land cover classification
The first and probably most important step of processing data consisted of extensive field observations during July, August and September 2011. Led by the traditional concept of Lautensach (1953), the focus was both on the central–peripheral and hypsometric variations of geographic forms (Formenwandel). Subsequently, by interpreting and comparing the remotely sensed data (feature space) and the “real space” (Seger, 1995), a set of six land cover classes that matched with the input data's quality was defined (Table 1). Woodland and shrubland were merged to a single class, as many woody species appear both as trees and shrubs. Further, harvested cereal cropland was not distinguishable from ichu grasslands, and ch’ampa meadows had spectral characteristics similar to those of cultivated pastures – e.g. alfalfa – and certain plants grown for vegetables. Additionally, regarding the differentiation between cereals and pastures, rotational cycles had to be taken into account (Knapp, 2010). Next, by using the ArcGIS® 10 software package, settlement areas were separated from non-settlement land cover. Therefore, the thermal (resampled to 30 × 30 m resolution) and both medium infrared bands were used to create a 675 (RGB) false color composite, which mainly enabled the distinction of rock at high altitudes (suni and puna) and built-up land – two surfaces with similar spectral characteristics; for separating settlements and rock of river bars, the Mantaro River and two tributaries (the Cunas and Shullcas rivers) had to be digitized as polyline and buffered by 350 m and 75 m, respectively. Subsequently, a supervised digital classification (maximum likelihood algorithm; equal a priori probability weighting) was carried out. Post-classification processing comprised the exclusion of the river masks, the elimination of single, isolated pixels (majority filtering, considering eight neighbor pixels) and the smoothing of class boundaries (boundary clean). Further, it was assumed that a pixel classified as settlement could not afterward change into another land cover class.
Table 1
Defined land cover classes and their description.
| Class ID | Class name | Explanation | Examples |
|---|---|---|---|
| 1 | Wood or shrubland (WSL) | Non-cultivated woody plants; cultivated wood crops | Queñual (Polylepis spp.), Eucalyptus (Eucalyptus spp.), Pine (Pinus spp.), Tara (Caesalpinia spp.) |
| 2 | Ichu grassland or cereal crops (IGC) | Natural herbaceous vegetation dominated by tussock-forming graminoids (known as ichu); graminoids cultivated for grains | Ichu (Stipa ichu), Wheat (Triticum spp.), Corn (Zea mays) |
| 3 | Ch’ampa, vegetable or pasture crops (CVP) | Natural herbaceous vegetation dominated by carpet-forming graminoids (known as ch’ampa); graminoids and legumes cultivated for forage; several plant species cultivated for leaves, buds, roots or tubers | Paku paku (Aciachne pulvinata), Alfalfa (Medicago sativa), Potato (Solanum tuberosum) |
| 4 | Bare soil areas (BSA) | Damp, moist or wet bare soil; especially harvested, fallow or abandoned cropland | |
| 5 | Settlements (SET) | Buildings, open spaces smaller than 6 pixels (0.54 ha) inside the continuous built-up area | |
| 0 | Other | Rocks and river bars; dry bare soil (sandy and well drained); rivers, lakes and wetlands |
Thus, the 2008 and 1998 settlement classifications had to be merged with all their respective previous data, in order to correct misclassification. After that, the three results were visually cross-checked against the GEOEYE-1 and IKONOS-2 true color images and, where possible, manually enhanced by masking and exclusion. A few small areas of bare rock remained misclassified; thus, the absolute amount of settlement areas probably was slightly overestimated. Yet, as these pixels mainly appeared in all three classifications, their influence on the relative results of land cover change detection was considered insignificant. In order to define the areas of the remaining five land cover classes, a 742 (RGB) false color composite was created for each year. Then, the classification (maximum likelihood algorithm) and relevant steps of the post-classification process previously described (majority filtering, boundary clean) could be repeated. Finally, by creating a mosaic of all land cover types classified – including the river mask – the three thematic maps were ready for accuracy assessment.
Accuracy assessment and change detection
As ground checking a hundred randomly sampled points over large mountainous terrain was not feasible – due to spatial, temporal, financial or legal restrictions – , reference data for accuracy assessment were sampled and checked by two different manners: 30 points (other than the training sites) per class were purposively sampled during the 2011 walking survey; therefore, sites with an estimated extent of at least 30 × 30 m that seemed clearly assignable were selected. Where possible, they were equally distributed over both largely homogenous and more heterogeneous areas. After that, these ground data were compared with the 2008 thematic map. Further, a total of 180 points (distributed over the study area; distance between points >30 m) were additionally created within ArcMap™ 10 using simple random sampling and then visually compared with the mentioned high resolution true color images from GeoEye®. Next, a simple confusion matrix – for calculating the percentage of correctly allocated cases – required a careful assessment (Foody, 2002). Cohen's Kappa coefficient of agreement, still common for the – intended – assessment of accuracy (Aguirre-Gutiérrez, Seijmonsbergen, & Duivenvoorden, 2012; Beekhuizen & Clarke, 2010), was not calculated; as recent studies underline, this and many other Kappa indices are useless and misleading for the practical applications in remote sensing (Pontius & Millones, 2011; Van Vliet, Bregt, & Hagen-Zanker, 2011).
Although neither ground points nor high resolution satellite images were available – and hence no quantitative assessment could be carried out – , it was assumed that the accuracy of the 1988 and 1998 results would not differ decisively from the 2008 map due to the following reasons: (1) the same sensor, Landsat 5 TM, acquired the imagery; (2) the acquisition dates, phenological and atmospheric conditions were nearly identical; and (3) the same classification workflow was applied. Further, a final interpretation of the three classification results revealed both the almost inexistence of rather unlikely land cover change trajectories – such as WSL–BSA–Other – as well as the existence of plausible and consistent development paths within each class and altitudinal belt.
In order to analyze land cover changes, the classes' areas per altitudinal belt and analysis date were calculated and their respective percentages compared. For computing spatially explicit change trajectories (1988–1998 and 1998–2008), post-classification pixel-wise comparison (image addition) was applied; therefore 3-digit (class ID × 102), 2-digit (class ID × 101) and 1-digit pixel values (class ID × 100) were first assigned to the land cover classes of 2008, 1998 and 1988, respectively. After adding up the three reclassified images, the land cover change trajectories were easily interpretable via the new pixel values.
Results and discussion
Accuracy assessment
On the basis of 360 sample points – five out of six classes counted 46 points or more and thus complied with Congalton's (1991) rule of thumb – , the percentage of correctly allocated pixels reached a value of 86.39%; thereby, the majority of classes had individual values of more than 80% – the only exception was BSA with 37 sample points and an individual accuracy of 54.1% (Table 2).
Table 2
Results of accuracy assessment per class and sample type.
| Reference data | Sample type | Classified data | ||||||
|---|---|---|---|---|---|---|---|---|
| WSL | IGC | CVP | BSA | SET | Other | Accuracy [%] | ||
| WSL | Purposive | 27 | 2 | 1 | 0 | 0 | 0 | 90.0 |
| Random | 10 | 1 | 5 | 0 | 0 | 0 | 62.5 | |
| Sum | 37 | 3 | 6 | 0 | 0 | 0 | 80.4 | |
| IGC | Purposive | 0 | 30 | 0 | 0 | 0 | 0 | 100.0 |
| Random | 3 | 56 | 3 | 0 | 0 | 2 | 87.5 | |
| Sum | 3 | 86 | 3 | 0 | 0 | 2 | 91.5 | |
| CVP | Purposive | 0 | 0 | 29 | 0 | 0 | 1 | 96.7 |
| Random | 0 | 1 | 47 | 0 | 1 | 1 | 94.0 | |
| Sum | 0 | 1 | 76 | 0 | 1 | 2 | 95.0 | |
| BSA | Purposive | 0 | 4 | 6 | 17 | 3 | 0 | 56.7 |
| Random | 0 | 0 | 4 | 3 | 0 | 0 | 42.9 | |
| Sum | 0 | 4 | 10 | 20 | 3 | 0 | 54.1 | |
| SET | Purposive | 0 | 0 | 1 | 1 | 28 | 0 | 93.3 |
| Random | 0 | 0 | 3 | 0 | 12 | 1 | 75.0 | |
| Sum | 0 | 0 | 4 | 1 | 40 | 1 | 87.0 | |
| Other | Purposive | 0 | 0 | 0 | 2 | 0 | 28 | 93.3 |
| Random | 1 | 1 | 0 | 0 | 1 | 24 | 88.9 | |
| Sum | 1 | 1 | 0 | 2 | 1 | 52 | 91.2 | |
| Total | 86.39 | |||||||
However, in the latter case, it can be assumed that the lower value is rather a result of imperfect reference data (Foody, 2010) than a product of misclassification. Owing to crop rotation, agricultural land cover is highly dynamic (cereal crops, cultivated pastures, fallow land); thus, both ground reference data from 2011 as well as the various high resolution reference images contain a high probability of being different from the situation on July 13, 2008 (Landsat 5 TM acquisition date). Against this background, the overall accuracy targets set by Thomlinson, Bolstad, and Cohen (1999) – more than 85% with no individual class below 75% – could be considered accomplished. Thereby, the purposively sampled points led to slightly better results than those randomly sampled – although the difference seemed to decrease with a rising number of random points per class.
Land cover change
A first analysis of the land cover classes' portions and changes within the total study area shows clear tendencies (Table 3). The area covered by wood or shrubland (WSL) doubled between 1988 and 2008; thus, it represented the class with the highest relative growth rates. Yet, as WSL made up only a small share of the total study area in 1988, the absolute increase was modest. Ichu grassland or cereal crops (IGC), the largest class in 1988, showed a slight but continuous decrease of approximately 2500 ha or 10% until 2008. The proportion of ch’ampa, vegetable and pasture crops (CVP) – covering a quarter of the study area in 1988 and 1998 – incremented considerably to 36% in 2008. With about 5000 ha in 1988 and 1998, BSA (damp, wet or moist bare soil areas) presented no major change. Nevertheless, between 1998 and 2008, this land cover class' extent declined to 3500 ha. The settlement area (SET) – in relative terms – grew about a third between 1988 and 1998 and remained almost constant during 1998–2008. Other land cover made up 27% of the study area in 1988 and was nearly halved to 15% in 2008. Thereby, the change was mainly related to dry bare soil, as rock and water areas were supposed to be stable. In sum, the numbers presented convey clear development tendencies for each land cover class. However, due to the hypsometric variations of geographic forms, these results may lead to misinterpretations if the land cover's distribution at different altitudinal belts (quechua, suni, puna) is not taken into account (Fig. 2).
Distributions of land cover classes at different Andean altitudinal belts in 1988, 1998 and 2008.
Table 3
Portions and changes of land cover classes within the total study area (percentages rounded to integer values).
| 1988 | 1988–1998 | 1998 | 1998–2008 | 2008 | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Area [%] | Area [ha] | Count [pixel] | Change [%] | Area [%] | Area [ha] | Count [pixel] | Change [%] | Area [%] | Area [ha] | Count [pixel] | |
| WSL | 3 | 1916 | 21294 | 42 | 4 | 2723 | 30258 | 45 | 5 | 3936 | 43738 |
| IGC | 35 | 26225 | 291385 | −1 | 34 | 25841 | 287120 | −8 | 32 | 23756 | 263951 |
| CVP | 24 | 18450 | 205000 | 4 | 25 | 19129 | 212542 | 43 | 36 | 27341 | 303794 |
| BSA | 6 | 4743 | 52697 | 8 | 7 | 5124 | 56936 | −33 | 5 | 3439 | 38210 |
| SET | 5 | 3661 | 40673 | 36 | 7 | 4994 | 55488 | 5 | 7 | 5266 | 58512 |
| Other | 27 | 20054 | 222826 | −14 | 23 | 17238 | 191531 | −34 | 15 | 11310 | 125670 |
| Total | 100 | 75049 | 833875 | 100 | 75049 | 833875 | 100 | 75049 | 833875 | ||
A comparison of the results from 1988 (a), 1998 (b) and 2008 (c) in Fig. 2 reveals several interesting trends. Regarding WSL it can be observed that both the suni (+114% during 1988–2008) as well as the puna altitudinal zone (+568% during 1988–2008) achieved a stronger relative growth than the lower quechua (+18%). While there were several artificially reforested areas (Eucalyptus spp. and Pinus spp.) observed below 3800 m asl during the direct field observations, the growth within the puna belt was seen to be a result of natural growth, especially on steeper slopes.
The loss of IGC between 1998 and 2008 occurred especially within the quechua (−45%) – often seen as the uppermost zone for cereal cultivation – and the suni (−16%) transitional zone. Ichu bunch grasses, although potentially existing (White, 1985), were not seen below 3800 m asl. The IGC's development was thus primarily interpreted as a result of a declining wheat or maize production. With respect to the puna, the tussock grassland's area showed a modest increase during 1998–2008 (+5%) and hence cushioned the general decrease of IGC.
Between 1988 and 2008, areas covered by carpet meadows, vegetable or pasture crops expanded both on the quechua valley floor (+49%) and the superior suni (+67%) and puna (+38%). Since major areas of carpet meadows were relatively rare below 3500 m asl, as the 2011 field observations indicated, CVP's notable increase at this altitudinal zone (about 6400 ha) could clearly be assigned to the cultivation of crops previously mentioned (Table 1). Although the influence of misclassified pixels can never be completely excluded – and more facts are required – , the latter land cover change seems to be linked to the almost simultaneous loss of BSA at the quechua level (−27%; approximately 1600 ha) and may hence sustain the observation of agricultural intensification made by Haller and Borsdorf (in press) – a process partially driven by urban and peri-urban settlement and population growth. On the other hand, the growing area of carpet meadows at the puna was obviously a result of range burning and livestock pressure, especially sheep (Ovis aries) and camelid (Lama glama and Vicugna pacos) grazing (Baied & Wheeler, 1993; Gade, 1999; Godoy, 1984; Postigo, Young, & Crews, 2008; White, 1985).
Trajectories in the hinterland
For validating the hypotheses of rural–urban land cover change, the calculation of change trajectories for each altitudinal level offers insightful results. Moreover, for evaluating the direct impacts of urban growth on the agricultural landscape, the land changed into settlement is examined separately, interpreted against field observations in 2011 and linked to the overall tendencies detected.
The 10 major change trajectories at the quechua – their area represents 25% of the total altitudinal level – show that the conversion from cereal cropland into areas of vegetables and cultivated pastures was the most important (Table 4). Moreover, the results reveal that this was a continuous change that occurred both during 1988–1998 and 1998–2008. A similar tendency can be observed regarding dry bare soil (in this context the only reasonable land cover within the class “Other”) and vegetable or pasture crops. Regional agricultural statistics prove the results obtained via GIS analysis; they further suggest, that the decline of wheat and the rise of alfalfa cultivation started particularly in 1996 – while the production of potato, for example, was relatively stable until 2010 (Fig. 3). These tendencies of agricultural intensification are probably due to an increasing global or national consumption of dairy products and thus an increasing need for cultivated pastures (Bebbington, 2001; Fernández Baca & Bojórquez, 1994) or because of an international demand for certain crops such as artichokes (Cynara cardunculus), as CEPES reported (Centro Peruano de Estudios Sociales, 2007).
Relative changes regarding the size of production area (alfalfa, potato and wheat) in the Junín region 1988–2010. Source: Compendio Estadístico de la Oficina de Estudios Económicos y Estadísticos (Ministerio de Agricultura, 2010).
Table 4
Major trajectories of land cover change at the quechua and their shares of the total altitudinal zone.
| Rank | Change trajectories | Coverage | |||
|---|---|---|---|---|---|
| 1988 | 1998 | 2008 | Area [%] | Area [ha] | |
| 1 | IGC | CVP | CVP | 4.2 | 1665 |
| 2 | IGC | IGC | CVP | 4.0 | 1555 |
| 3 | Other | Other | CVP | 3.3 | 1291 |
| 4 | CVP | IGC | CVP | 2.8 | 1097 |
| 5 | Other | CVP | CVP | 2.7 | 1046 |
| 6 | Other | Other | IGC | 2.1 | 832 |
| 7 | CVP | Other | CVP | 2.0 | 797 |
| 8 | Other | IGC | CVP | 1.4 | 538 |
| 9 | CVP | BSA | CVP | 1.3 | 492 |
| 10 | BSA | BSA | CVP | 1.2 | 487 |
At the adjoining suni, the 10 major trajectories comprise an area of 35% of the total zone (Table 5). The decrease of IGC was evidently linked to the growth of WSL and CVP – between 1988 and 1998 as well as within the subsequent period. As directly observed in situ, areas artificially reforested with species cultivated for wood contributed in large part to this development. The third clear tendency – interpreted as a reaction to the two previously mentioned trends – is the change from dry bare soil to IGC.
Table 5
Major trajectories of land cover change at the suni and their shares of the total altitudinal zone.
| Rank | Change trajectories | Coverage | |||
|---|---|---|---|---|---|
| 1988 | 1998 | 2008 | Area [%] | Area [ha] | |
| 1 | Other | Other | IGC | 7.8 | 662 |
| 2 | IGC | IGC | CVP | 6.7 | 569 |
| 3 | IGC | IGC | WSL | 4.4 | 371 |
| 4 | Other | IGC | IGC | 4.0 | 338 |
| 5 | IGC | CVP | CVP | 3.6 | 303 |
| 6 | IGC | WSL | WSL | 2.7 | 234 |
| 7 | IGC | Other | IGC | 1.8 | 156 |
| 8 | Other | Other | CVP | 1.7 | 145 |
| 9 | CVP | IGC | CVP | 1.7 | 141 |
| 10 | Other | WSL | WSL | 1.1 | 95 |
Table 6 gives an overview about the 10 most frequent trajectories' areas (covering 26% of the total altitudinal belt's area) at the puna level. The modest increase of ichu grassland is clearly a result of vegetation recovery since 1988, as the tussock-forming graminoids expanded especially on dry bare soil (land cover class “Other”). The second most important and continuous trend relates to the conversion of bunch grassland into carpet meadows. As mentioned before, this is due to range burning and – regarding the maintenance – livestock grazing; activities that may be interpreted as a state-supported (through credits from the economic ministry's Banco Agropecuario) reaction of smallholders to the globally growing demand for alpaca fiber and derivatives, which are often produced by Huancayo's strong textile industry.
Table 6
Major trajectories of land cover change at the puna and their shares of the total altitudinal zone.
| Rank | Change trajectories | Coverage | |||
|---|---|---|---|---|---|
| 1988 | 1998 | 2008 | Area [%] | Area [ha] | |
| 1 | Other | Other | IGC | 5.7 | 1447 |
| 2 | Other | IGC | IGC | 4.8 | 1203 |
| 3 | IGC | IGC | CVP | 4.5 | 1137 |
| 4 | IGC | CVP | CVP | 3.6 | 900 |
| 5 | IGC | Other | IGC | 2.1 | 518 |
| 6 | IGC | IGC | WSL | 1.3 | 329 |
| 7 | IGC | CVP | IGC | 1.3 | 325 |
| 8 | CVP | CVP | IGC | 1.1 | 288 |
| 9 | CVP | IGC | CVP | 0.8 | 205 |
| 10 | CVP | IGC | IGC | 0.7 | 177 |
Settlement growth trajectories
Due to its role as a transition zone between the city and the hinterland, the peri-urban interface is the scene of multiple human-environmental changes (Allen, 2003). It hence can be considered a preview of what rural–urban developments may be about to happen in the rest of the quechua hinterland. While rapid peri-urban growth in Huancayo Metropolitano occurred predominantly in the late 1980s and early 1990s, influenced by the terrorism-driven rural–urban migration, settlement expansion during 1998–2008 decelerated noticeably in quantitative terms and changed even qualitatively – toward the “condominium-dotted vernacular” (Haller & Borsdorf, in press). Interpreting the trajectories of land changed into settlement during 1988–2008, in combination with results from direct field observations in 2011, permits to draw certain conclusions concerning socio-spatial processes and the “nature/culture gestalt” (Gade, 1999). As shown in Fig. 4, major conversion from open land to settlement areas took place between 1988 and 1998, principally on bare soil – dry as well as damp, moist or wet (425 ha and 390 ha, respectively) – and land previously covered by ch’ampa, vegetables or pasture crops (401 ha). Unsurprisingly, woodland and shrubland were not affected. Settlement expansion between 1998 and 2008 was clearly less intense – a result that fits in with the metropolitan area's average annual population growth rates during 1981–1993 (3.81%) and 1993–2007 (1.81%) calculated by Haller and Borsdorf (in press); with respect to the trajectories, especially former persistent areas of bare soil (1988–1998) were urbanized; a fact reminiscent of both land speculation and Hartke's (1953) early concept of the “social fallow” (Sozialbrache) – where he analyzes fallow agricultural land against the background of socio-economic changes such as industrialization or tertiarization.
Amount of land changed into settlement during 1988–2008 by their respective trajectories. Areas less than 1 ha are not shown.
Direct field observations identified real estate developers to be among the principal forces of urbanization-driven land cover change around the city's quechua zone: by constructing planned condominiums (“condominization”) for an emerging middle class that demands environmental amenities (Fig. 5), as well as by boosting land prices and thus motivating peri-urban land owners to sell lots (Fig. 6).
Land cover transition from carpet meadows to planned settlements in San Carlos, Huancayo, 2011.
From Triticum to condominium? Cropland as a resource for peri-urban residential development in Pilcomayo, 2011.
Conclusions and outlook
The findings of the present case study in the Peruvian Central Andes clearly show land changes (1988–2008) that have to be interpreted as the consequences of land use intensification around Huancayo Metropolitano – at all altitudinal belts of the rural–urban landscape. While quantitative urban growth had its peak during the period from 1988 to 1998, non-urban changes – especially agricultural intensification – predominated during 1998–2008.
At the quechua level, irrigation-dependent vegetable or pasture producers and the emerging urban society – with its resource-intensive lifestyle in privatized spaces – compete for ecosystem services (Fig. 7), such as the melt water from the surrounding glaciers (Guevara Gil, 2010; Su, Xiao, Jiang, & Zhang, 2012). Thus, the observed socio-ecological consequences of the peri-urban change (Haller & Borsdorf, in press) shown in Fig. 8 are generally supported by the present study's quantitative results. However, the large increase of intensively used agricultural land suggests that condominization is just one part of a globalization-driven land or environmental change, which entails risks and opportunities for the entire hinterland's sustainable development (Coy, 2010; Marchant, 2010; Zimmerer, 2006).
Announcement of The Shamrock: “the best ecological and residential site of Huancayo”, San Carlos, Huancayo 2011.
Scheme of the condominization-driven peri-urban change at the quechua valley floor. Source: modified and adapted from Haller and Borsdorf (in press).
The results also prove that rural–urban migration toward and rapid urban growth of Huancayo Metropolitano did not lead – as one could suppose in view of extra-Andean experiences (Borsdorf & Bender, 2007; Camacho, Dobremez, & Capillon, 2008; Lasanta, Arnáez, Errea, Ortigosa, & Ruiz-Flaño, 2009; Pelorosso, Leone, & Boccia, 2009) – to the decline of high altitude agricultural land use and the increase of the negatively connoted shrub encroachment (Eldridge et al., 2011) at the hinterland's puna: if the expansion of ch’ampa carpet meadows is considered a spatial manifestation of the “nature/culture gestalt” (Gade, 1999), then range burning and livestock grazing activities continue during global change and – regarding the rural–urban study area – even increase. The suni transition zone follows these tendencies of intensification, since their slopes are more and more reforested with highly demanded wood crops (Eucalyptus spp. and Pinus spp.) and used for cultivating potatoes (Solanum tuberosum), oca (Oxalis tuberosa) or olluco (Ullucus tuberosus).
An integration of political and cultural ecological analyses into land change studies – as proposed by Turner II and Robbins (2008) – could explain the motivations for changes in land use and, as a consequence, foster synthetic–integrative (or even holistic) applied geography: how did the transition from a community-based Agrarian Society of Social Interest (SAIS Cahuide) – established during Velasco's leftist military government (1968–1975) and destroyed by the Maoist terrorist movement Shining Path in 1988 (Mayer, 2009) – to the neoliberal era (initiated by ex-president Alberto Fujimori in 1990) influence the observed land change at the suni and puna around Huancayo? In what way are cultural–economic globalization, neoliberal policies and privatization driving land use expansion and intensification by smallholders at high altitudes? How do the relations between rural smallholder families or comunidades campesinas (peasant communities), the urban-based food and textile industry and governmental initiatives (e.g. “Sierra Exportadora”, a program that aims at linking smallholders and global markets) contribute to the analyzed rural–urban landscape changes?
In sum, as claimed by the UNESCO-supported GLOCHAMORE research strategy, applied research on land change in mountain areas is needed more than ever. Moreover, the present study underlines the necessity of integrating all altitudinal zones of the hinterland into the processes of rural–urban planning and policy making, for the impacts of globalization or urbanization on the rural–urban society and environment are not limited to the hot spots of change at the valley floor. Thereby, the methodology used for the present study – incorporating conventional, medium scale land cover maps with Pulgar Vidal's (1946) model of altitudinal belts and actual empirical results of field work – enables land use interpretations that take the “real space's” (Seger, 1995) horizontal and vertical socio-cultural and environmental diversity into account. Nevertheless, following a more historicist perspective, extended analyses of the cultural landscape's genesis (Antrop, 2005; Bender, Boehmer, Doreen, & Schumacher, 2005; Gade, 2011) would lead to an even better understanding of current land transformations and thus improve rural–urban decision making.
The Zonificación Ecológica Económica (Consejo Nacional del Ambiente, 1999) – a multi-level spatial planning instrument currently implemented by the Peruvian Ministry of Environment – could offer practical solutions to some of the conspicuous rural–urban challenges. However, it still remains to be seen whether the self-imposed aims of sustainable, inclusive and conflict-poor land use can be fulfilled, as the influence of public planning is declining in times of privatization and weakening social cohesion (Fig. 8). If the project would succeed, this regional path to the peak of sustainability could become a global road to success for mountain regions – and even beyond.
Acknowledgments
The results of this study contribute to the project Rapid Urban Growth in the Andes (2012–2015) that is gratefully financed by the Austrian Science Fund (FWF) [Project No. P24692]. Field work and data acquisition in 2011 were supported by the Tyrolean Regional Government's science fund (TWF). Further, I would like to thank the University of Innsbruck for having awarded a doctoral start-up fellowship. Special thanks go to my mentor Axel Borsdorf – who has promoted my interest in the fields of urban and mountain research – for his valuable advices and continuous support, as well as to the anonymous reviewers for their very constructive suggestions.
References
Citations & impact
Impact metrics
Article citations
The "sowing of concrete": Peri-urban smallholder perceptions of rural-urban land change in the Central Peruvian Andes.
Land use policy, 38:239-247, 01 May 2014
Cited by: 0 articles | PMID: 25844006 | PMCID: PMC4375667
Similar Articles
To arrive at the top five similar articles we use a word-weighted algorithm to compare words from the Title and Abstract of each citation.
The "sowing of concrete": Peri-urban smallholder perceptions of rural-urban land change in the Central Peruvian Andes.
Land use policy, 38:239-247, 01 May 2014
Cited by: 0 articles | PMID: 25844006 | PMCID: PMC4375667
Rural Land Use Change during 1986-2002 in Lijiang, China, Based on Remote Sensing and GIS Data.
Sensors (Basel), 8(12):8201-8223, 11 Dec 2008
Cited by: 4 articles | PMID: 27873983 | PMCID: PMC3791014
Land use as a driver of soil fertility and biodiversity across an agricultural landscape in the Central Peruvian Andes.
Ecol Appl, 27(4):1138-1154, 24 Mar 2017
Cited by: 1 article | PMID: 28117908
Numerical simulation of the impact of reforestation on winter meteorology and environment in a semi-arid urban valley, Northwestern China.
Sci Total Environ, 569-570:404-415, 25 Jun 2016
Cited by: 0 articles | PMID: 27348705
What are the Andean Colombian anurans? Empirical regionalization proposals vs. observed patterns of compositional dissimilarity.
PeerJ, 11:e15217, 13 Jun 2023
Cited by: 0 articles | PMID: 37334120 | PMCID: PMC10274619
Funding
Funders who supported this work.
Austrian Science Fund FWF (2)
Grant ID: P 24692
Grant ID: P24692