Open access

Land Use and Land Cover (LULC) Change in the Boconó River Basin, North Venezuelan Andes, and Its Implications for the Natural Resources Management

Written By

Joel Francisco Mejía and Volker Hochschild

Submitted: November 21st, 2011 Published: November 7th, 2012

DOI: 10.5772/53579

Chapter metrics overview

3,157 Chapter Downloads

View Full Metrics

1. Introduction

Land Use & Land Cover (LULC) have been historically permanently subject to biophysical and anthropogenic forces which induce changes in different structure-levels and space-time scales, and modify the energy and water exchange of the soil-vegetation-atmosphere system; such modifications become globally significant through their cumulative effects, so it would be particularly hazardous for food production and food security [1] [2] [3]. Thus, Land Use & Land Cover (LULC) changes are simply the most conspicuous changes in cultural landscapes worldwide [4] [5] [6].

Particularly the tropical regions have undergone dramatic Land Use changes in the last few decades, and these changes are the effect of an equally large number of local causes and factors, highlighting a complexity that tends to defy easy generalizations [7] [8] [9] [10] [11].

Many hydrological systems of the tropical regions are relatively densely populated, with relatively high rates of population growth, which has serious implications in the relationships between people and environmental services [12]. In mountainous regions, mostly poor people are settled in steep hillsides (slopes above 15%), usually practicing a smallholder farming system with agricultural production in small parcels for subsistence purposes, as well as shifting cultivation and slash & burn agriculture, which represent a pressure over natural resources in areas which are ecologically fragile and environmentally sensitive. About 25 to 30% of Central America and northern South America consist of mountainous areas where the conditions above mentioned are quite common [13]. Thus, the dynamics of natural resources use in river basins and watersheds across the mountain regions in the tropics are determined by three factors: environmental, social and economical conditions [4] [9] [12] [14] [15].

According to [8], the LULC changes have a notorious impact on climate, at local and regional levels, due to the modifications in the carbon cycle, the local evapotranspiration patterns, as well as precipitation regimes. This fact justifies many concerns about the implications that the LULC changes could have in the water resources, particularly in the hydrological regimes worldwide. These concerns have been motivating the analysis of the relationships between the LULC changes and the hydrological regimes (river flows, runoff dynamic, floodings, water depletion, etc) in a spatial-temporal perspective. Some examples of these includes: [16 - 31].

Certainly, these are valuable experiences to deal with such a complex task; however, there are still many gaps in this process to be solved, and many questions to be answered. Moreover, many of these experiences are all spatially confined to temperate regions, where biophysical as well as socioeconomic conditions are particular. Tropical ecosystems are very different from their counterparts in higher latitudes. They have different geological and evolutionary histories, and different climatic extremes and dynamics. The number of interacting species is typically much higher in tropical ecosystems, including streams networks also, and the interactions are often more complex [9]. Social, economical and political conditions in tropical rural areas are also very complex; thus, the poverty, depressive local economies, instability and lack of plans and investment programs are always current, and usually such complex realities and the collateral relationship has not been well studied so far. Thus, the knowledge remains still weak and the lack of information about local and regional environmental dynamic is remarkable [11] [14] [15] [32] [33] [34] [35].

The river basins are subject to constant processes of change, so the state and the structure of river landscapes and land resources are primarily determined by the type and intensity of the utilisation of the ecological, economic, social or cultural functions provided by the river systems. The new paradigm recognize the river basins as complex, ecological and interactive systems, which means that the integrated water resource management follows the central themes of the ecosystem approach and of adaptative management; in fact, the WFD (Water Framework Directive) of the European Union, has adopted the “ecosystem-oriented river management “ as central approach to be followed by the institution [36].


2. Problem description

The Andean region in Venezuela is considered the most important “water resource-area” in the western part of the country. The source streams of many important river systems in the country are located there, having a complex and intricate channel network, with the “first order streams” (the most important sources of fresh water in many regions worldwide) broadly dominating the landscape system. Due to the biophysical configuration and the attractiveness of the Andean landscapes, the region has been under anthropogenic impact from times before the arrival of the Spaniards. However, in recent decades that pressure has been gradually increasing, which eventually could have significant impacts on the natural resources basis, particularly water, forest and soils.

Located in the northern part of that region, the Boconó River Basin can be considered as a representative case of the complex dynamics characterizing the Andean hydrological systems. Having a total surface area of 1580 km2 and a wide altitudinal range, the River Basin harbor many ecosystems ranging from the Sub-Andean Páramo in the upland areas, to the Savannah ecosystem downstream in the upper plains of the Llanos region. With an annual yield of 2,300 million m3 and a very acceptable chemical quality, the Boconó River was included into the regional planning policies in the Seventies, in order to develop the water resources in the lowlands region, so that the Boconó – Tucupido Dam systems were built in the Llanos region, in order to generate energy, flooding control and for irrigated cropping also [37].

A very significant portion of the area is still under natural Land Cover types, like the Tropical Montane Cloudy Forest (TMCF). This ecosystem has a paramount importance, not only in terms of their ecological richness, but also in terms of hydrological functioning, specifically for water yield. In such forests there is usually a net gain of water that comes from the “horizontal precipitation” or “occult precipitation” in form of wind-driven drizzle and fog [9] [38].

On the other hand, there are also numerous sparse rural settlements representing a huge potential for agricultural production of some crops like coffee and vegetables (potatoes, carrots, onions, beans and others). The high accessibility through an intricate network of rural and local roads makes it easier to promote the sparse settlement in sloping hillsides across the area, being a crucial factor which determines the LULC changes contributing to the intensification of some erosion and land degradation processes [39].

All these conditions acting together in a strongly integrated way, resulting in a complex situation in which the increasingly sparse population is making even more pressure on natural land cover types, particularly on the Tropical Montane Cloudy Forest, so that the conversion of LC into LU appears to be persistent and intense. The River Basin was declared in 1974 “Protected Area” in order to preserve the water resources [40], and the Guaramacal National Park was created in 1988, which cover the southeastern flank of the area. Both figures aimed to guarantee the conservation of the ecosystems, the biodiversity, and to ensure the water production [41].

Nevertheless, the area continues to show a trend respect to the anthropogenic pressure, so the agricultural frontier is even more extended, meanwhile the forested land cover types tends to be decreased, and some land degradation processes like erosion and sediment yield seems to be even more intense. This has severe implications for the biodiversity, but also affects substantially the hydrological dynamic through changes in local microclimates, changes in moisture regimes, that eventually could lead changes in the hydrological regimes, especially the seasonal flows, peak flows, as well as changes in the water quality.

2.1. Main goal

The main goal of this paper is to analyze the spatial dynamic of the Boconó River Basin during the Period 1988 – 2008, in terms of the main LULC changes and systematic transitions that have been occurring in the area under an ecosystem-oriented approach. They were discussed in terms of the implications that such changes and transitions have for the natural resources management at the river basin level (watershed management). The results showed here are only a partial output of the still ongoing PhD project: “Spatial changes and hydrological dynamic of the Boconó River Basin, north venezuelan Andes”, which is actually developed at the Eberhard Karls University – Tübingen, Germany.


3. The geographical context – study area

The study was focused on the upland part of the Boconó River Basin, located in the south- east part of the State of Trujillo, between the coordinates 09°11’40” - 09°31’50” N and 70°04’08” – 70°22’53” W, with a surface area of 537.62 km2. The highest point in the Basin is 3400 m.a.s.l in the Páramo of Cendé, and the lowest point (outlet) is the confluence between the Boconó and Burate river (1100 m.a.s.l) (Fig. 1). The Boconó River drops from the north-east to the south-west, over a distance of approximately 57 km, having a mean runoff about 15, 55 m3/sec [33].

Figure 1.

Location of the Study area

The area has a seasonally humid climate, having a wet period from April to October, and a dry period from November to March. Annual mean rainfall is about 1838 mm, and the annual mean temperature range from 19.7 °C to 21.5 °C [42]. The Basin has a relatively elongated form, and the drainage pattern is dendritic with a tendency to be rectangular, due to the intense tectonic activity [37].

The catchment is located within the tectonic axis formed by the Boconó Fault, which is the most important structural feature of the Venezuelan Andes [37]. The Fault cross longitudinally the river, separating the metamorphosed crystalline rocks in the north portion, from those less metamorphosed in the south part [33]. The basin has a massive and strongly dissected topography, so that the topographic conditions are quite complex and varied, determined by different landforms like: structural risks, erosion risks, structural escarpments, hillsides and alluvial accumulations, and a mean slope which range between 35 – 40% [43].

The lithological framework is generally highly jointed, due the tectonic dynamic, and the rocks basically correspond to the formations: Iglesias Group (gneisses and schist), Sierra Nevada (granites), Mucuchachí (Shale and phyllites) and Palmarito (shale and marl) [44]. Soils are in general relatively deep, with textural classes ranging from clayed to sandy loam, being Ultisols, Inceptisols and Alfisols the most important and representative taxonomic categories in the area [45].

The altitudinal gradient (2300 m.a.s.l) and the climatic conditions, particularly the intense rainfall regime, lead to the existence of the Tropical Montane Cloudy Forest, which cover the 44, 6 % of the total surface. Other important ecosystems in the area are: sub-montane forest, grass, sucessional shrubland, schrub and sub-alpine Páramo. These categories of land cover coexist also with specific land use types, which are very importance not only in economical terms, but in social and cultural perspectives also [46]. Shifting cultivation is located mostly in upland areas, where slash and burning are usual tasks. Conventional agriculture is also developed in lower parts and quaternary landforms, in some cases under irrigation. Coffee plantations are very usual between 800 and 2000 m above sea level, occupying an important portion of the Sub-montane forest. In a small proportion, the extensive grazing shows a moderate development, being usually spatially confined to the low parts and the quaternary landforms [42]. Finally, the 1, 6 % of total surface is occupied by urban use, being the Boconó city the most important urban system in the area.


4. Methodological approach

In order to achieve the purpose of this project, a methodological approach combining remote sensing methods with spatial and multi-temporal analysis in GIS in an interactive way was implemented. At first, the study area was delineated from the SRTM data set (90 m spatial resolution) using the open source GIS software SAGA (System for Automated Geoscientific Analysis), in order to build the Digital Elevation Model (DEM), and also to prepare the basic thematic maps (Topography, Slope, Aspect, Drainage Network). Based on the structure pointed out by [47], the LULC mapping process was done in three main straightforward steps, as follows:

4.1. The pre - processing

Three time-points were defined in order to analyse the LULC dynamic in the river basin: T0 (1988); T1 (1997); and T2 (2008). For each time-point a group of LANDSAT TM scenes corresponding to missions 4, 5 and 7 were compiled from USGS LANDSAT Archive and the Institute of Geography (IGCRN) – ULA (Venezuela), which were considered suitable to the research requirements. The compilation process was quite difficult because the study area is frequently covered by dense clouds, especially during the rainy season. It means that the cloudiness and fog represented a challenge to deal with into the classification process, leading to compile additional scenes for special processing. Thus, the compiled scenes were classified in two groups: “pilot” scenes and “control” scenes. The first group included the main scenes to be classified for each time-point to be considered in the multi-temporal evaluation: 1988, 1997 and 2008, respectively. The second group were used as control images for the optimization of the classification for the first group, in order to improve the clustering processing in those areas covered by cloud, fog and shadows.

All the LANDSAT scenes compiled were pre-processed individually to make the geometric and radiometric correction, as well as the enhancement of some elements like brightness, contrast, haze reduction and equalization, in order to improve the image quality. All these processes were carried out interactively.

4.2. The LULC classification/analysis process

The classification process was developed through a semi – supervised method, following a multi – level clustering for a multi – class segmentation of the scenes. The scenes were separately classified, a procedure considered highly flexible and extensively used in the past, with good results reported [47].

At the first level the scenes were classified through an unsupervised method using the “hyperclustering approach”, a simple and relatively common approach to classify multiple LANDSAT scene mosaics. This classification approach generate many hyperclusters from the image data available by testing for within – cluster heterogeneity; then the hyperclusters can be merged into a smaller number of more reasonable groups which may resemble homogeneous classes, and finally label the resulting classes as spatial features of interest according to a pre-determined map legend or class hierarchy [48]. The process was done using the algorithm K-means available within the ISODATA decision-rule. In this case, the method was applied using 50 clusters to be classified after 24 iterations through the unsupervised approach (previous tests using 80 and 100 clusters, showed not many differences in the effective separation of the classes). The amount was then though reasonable to manage by the interpreter, and appropriate to differentiate the LULC classes in the study area.

Two groups of clusters were then identified: “pure clusters” representing categories with unique spectral signal; and “mixed clusters”, having two or more categories with similar spectral signal, which is normal because LANDSAT imagery for tropical forest regions display minimal band separability among vegetation types, so that different types of categories can be usually difficult to separate [49]. The “mixed clusters” were prone to a second - level classification process. They were separated from the scene through masking process, and after that they were submitted into a second clustering process, using supervised and unsupervised methods. Thus, the classes were correctly separated from the others. During the second – level classification, the clouds, fog and shadows were appropriately separated from other classes. They were used as mask scenes in order to cut the control images through spatial analysis, and finally they were processed like the “mixed clusters”, in the same way above described.

4.3. The product generation process

All the clusters were merged to form twelve final classes using the grouping process. Additionally, a spatial modelling process was done in order to make the altitudinal differentiation of the LC in the river basin, defining the Land Cover categories in an ecological sense, following the ecosystem approach. For this purpose the DEM was combined with the classified images using the ecological criteria from Sarmiento & Ataroff in [50]. Thus, the Land cover categories delineated are virtually “ecosystems units”. The classified scenes were finally filtered and exported to GIS software for the mapping creation and display processes.

The classifications were validated using conventional methods, depending on the availability of the reference ancillary data. For the T0 classification, only a land use map for 1980 was available in a non digital format. This map was then used as a reference source for the validation. A total of 255 validation points corresponding to reference pixels were randomly selected using the “stratified random” sampling method. They were interactively compared with the digital reference map, and the results were stored in the Accuracy Assessment Cell Array (software ERDAS 9,3), which is simply a list of class values for the pixels in the classified image file and the class values for the corresponding reference pixels [51]. The tool finally calculated the error matrix and the corresponding basic statistics, including the Kappa Coefficient, which were listed in the Accuracy report. For T2, a field validation process was driven, combined with validation points defined using Quickbird high resolution scenes available on the “open source” software GOOGLE EARTH, through the same process described for the T0 scenes. Finally, the T1 Classification was validated using the maps for T0 and T2, defining validation points basically in areas considered persistent across the time-period.

4.4. Multi-temporal evaluation of LULC changes in the Boconó River Basin (Post classification)

The multi-temporal evaluation process was conducted through spatial analysis in GIS. Hence, paired overlay was done in order to detect the changes occurred during the time-period considered. The Matrix operation used in this case allows two thematic images or vector files of different years to be compared [52]. This tool allowed to cross two different maps corresponding to the same area, in order to differentiate the changes occurred between the time-points. The resulting class values of a matrix operation are thus unique for each coincidence of two input class values described by rows (input layer 1) and columns (input layer 2) [53]; hence, the process produce two type of results: Maps which can illustrate the changes in a spatial context (land cover change map); and a cross-tabulation matrix containing the differences in area for the different classes.

The cross-tabulation matrix, also denominated “transition matrix” follows the format displayed on Table 1. The rows display the categories of time 1, and the columns display the categories of time 2. Entries on the diagonal indicate persistence in the landscape between the time-period, meanwhile the entries off the diagonal indicate a transition from category “i” to a different category “j” [54].

Time 2Total Time 1Loss
Time 1Category 1Category 2Category 3Category 4
Category 1P11P12P13P14P1+P1+ - P11
Category 2P21P22P23P24P2+P2+ - P22
Category 3P31P32P33P34P3+P3+ - P33
Category 4P41P42P43P44P4+P4+ - P44
Total Time 2P+1P+2P+3P+41
GainP+1 – P11P+2 – P22P+3 – P33P+4 – P44

Table 1.

General cross-tabulation matrix for comparing two maps from different points in time

Starting from the matrix-values, the Gain (Gij) was calculated through the difference between the total value for time 2 (P+j) and the persistence (Pij), using the Eq 1:

Gij= P+j PjjE1

On the other hand, the Loss (Lij) was the difference between the total value for the time 1 file (Pj+) and the persistence, using the Eq 2:

Gij= P+j PjjE2

The swapping (Sj) between the categories was calculated as two times the minimum value of the gains and losses, through the Eq 3:

Sj= 2 x MIN ( Pj+ Pjj, P+j Pjj)E3

The total change for each category (Cj) was the sum of net change (Dj) and the swapping (Sj), or the sum of gain and loss (Eq 4):

Cj= (Dj+ Sj)E4

In order to intend a more detailed analysis of the LULC changes, particularly the systematic inter-category transitions, the methodology proposed by [54] was applied, which analyze the off-diagonal entries to identify systematic transitions of land change for a given landscape´s degree of persistence. For that, the transitions must be interpreted relative to the sizes of the categories, leading to define the gain/loss that would be expected if the gain/loss in each category were to occur randomly [54]. The randomly expected gains for each category were calculated using the Eq 5:

       Gij=P+j-Pjjx Pi+1-Pj+E5

In this case, the gain as well as the proportion for each category at time 2 is considered fixed, distributing the gain across the other categories according the relative proportion of the other categories in time 1. The procedure to calculate the randomly expected losses for each category is quite similar to those explained above, using the Eq 6:

Lij=Pi+ -Piix P+j1-P+iE6

As in the gain, the equation assumes that the loss of each category is fixed, and then distributes the loss across the other categories according to the relative proportion of the other categories in time 2.

Finally, the systematic transitions were identified trough a comparison between the observed and expected values for gain and loss, for each category.


5. Results & discussion

Twelve (12) LULC categories to be analyzed were identified in the Boconó River Basin for T0, T1 and T2 classifications. The Table 2 display the LULC categories, each with the corresponding identity-code, designation, as well as a brief description. The results showing the accuracy and the Kappa Coefficient for the three time-points are displayed on Table 3. Two important clarifications must be here pointed out:

1. - The Category Open-cleared Forest (Oc-F) correspond to the lower sectors of the Tropical Montane Cloudy Forest (Tmc-F), which are prone to a clearcutting process for logging and wood extraction, eliminating partly the canopy of the tallest forest species; the clearing alter greatly the phenological structure of the forest, resulting in a very specific and different spectral signal respect the climax or undisturbed forest. They were conveniently considered separated categories for practical purposes inherent to the research goals.

2.- Coffee plantations constitute an important land use practice in the area; however, during the classification process the plantations (shade coffee) usually showed a very similar spectral signal as the Sub-montane Forest, which is the ecosystem where these plantations are usually located. They couldn’t be effectively separated at this resolution level, and more detailed remote sensing material for the study area was no available. For that reason the coffee plantations were necessarily combined with the Category: Sub-montane Forest (Sm-F).

5.1. General quantification of the change

The corresponding surface values for the time-points analyzed (T0, T1 and T2), are gently resumed on Table 4. An overview of the differences among the period, lead us to set up a basic differentiation between the LULC categories in three main groups as follow:

  1. LULC categories losing surface: basically the natural LC like forest (Tmc-F, Oc-F, Sm-F) and Grass (Gr-L) were included here. All of them show a decreasing trend between T0 and T2 (except Gr-L, which experienced a light increase between T1 – T2). The Tmc-F and Oc-F had a reduction of 3530, 43 ha between T0 – T2, representing the 12, 8 % of the total for the two categories combined in 1988. The reduction of the Sm-F in the river basin was more dramatic, losing the 43, 1% of the surface area respect to 1988, that is, 3244, 59 ha. On the other hand, Gr-L loosed 412, 11 ha between T0-T1, and slightly recovered 85, 05 ha in the next period, losing a total of 327, 06 ha (9 % of the total in 1988).

Table 2.

Land Use / Land Cover (LULC) Categories identified in the Boconó River Basin.

IndicatorT0 (1988)T1 (1997)T2 (2008)
Producers Accuracy87,4685,0291,53
Users Accuracy87,6282,9091,67
Total Accuracy87,3582,5988,80
Kappa Coefficient0,790,790,87

Table 3.

Main results obtained in the Accuracy assessment for the T0, T1 and T2 classifications.

LULC Categories1988 (T0)1997 (T1)2008 (T2)Dif
Dif total
Area (ha)Area (ha)Area (ha)
Tropical Montane Cloudy Forest24573,7823676,1222493,97-897,66-1182,15-2079,81
Open-cleared Forest2973,511648,81522,89-1324,71-125,91-1450,62
Sub-Montane Forest7523,16224,134278,51-1298,97-1945,62-3244,59
Sub-andean Paramo1114,21117,711114,473,51-3,240,27
Grassland (Anthropogenic)1280,341181,162832,03-99,181650,871551,69
Cropping Area2202,842330,462867,4127,62536,94664,56
Eroded Land28,6227,2739,87-1,3512,611,25
Urban Area433,98729,18865,26295,2136,08431,28
Flooding plain234,9391,95293,76157,05-98,1958,86
Sucessional Shrubland8591,6711984,7612840,753393,09855,994249,08

Table 4.

LULC evolution during the considered period

  1. LULC categories gaining surface: they are basically the human-induced types of land cover categories (Gr-An, Cro-L and Ur-U), as well as the categories: Schr and S-Shr. They increased progressively during the period, except Gr-An, which experienced a decrease in T0 – T1; however, the evident increase experienced during T1-T2 justify the inclusion of the category in this group. Gr-An and Cro-L combined, gained 2216, 25 ha, representing an increase of 63, 6 % of the agriculture in the river basin respect 1988. The Urban use (Ur-U) experienced a dramatic increase during the whole period, gaining 99,36 % (431,28 ha) of the surface area that the category occupied in T0. Meanwhile, the LC category S-Shr experienced a big change, gaining almost 50% (49, 5%) of the surface area for T0; so it gained a total of 4249, 08 ha. respect 1988. During the period Schr category gained 135, 36 ha (12%) respect to T0.

  2. Relatively stable LULC categories: here are included the rest of the LC categories: Sa-P, Ero-L and Fl-P. These categories showed a similar pattern during the whole period, in which they loosed and gained surface, but maintaining its proportionality respect the rest of the LULC categories. The Fl-P gained 157, 05 ha (67%) because of the flooding events occurred during the T0-T1. But in the second time-period it loosed 98, 19 ha to other categories.

These basic groups illustrate the general trends for the recent evolution of the LULCC in the river basin. However, they are only the initial framework to understand the spatial dynamic in the study area, so they cannot reflect conveniently the spatial changes in a quantitative/qualitative way. The next section provides a more comprehensive and detailed description of the LULC categorical changes for the two time-periods, in terms of quantification, net change, swapping as well as inter-category transitions.

The Figure 2 show the spatial distribution of the changes in the Boconó River Basin, which occurred within the both periods: T0 - T1 and T1 – T2. In the first period the River Basin experienced a total change of 30,34%, which means that 16309,89 ha were affected by a kind of spatial change processes, meanwhile the 69,66% of the surface area (37452,24 ha) was accounted as persistent landscape or simply persistence. Thus, persistence dominates widely the landscape system of the River Basin, which is considered normal, because the persistence usually dominates most landscapes, including those where authors claim that the change is important and / or large [54].

[55] accounted 92% of persistence for natural land covers in Mexico; in the Atlanta metropolitan area (one of the USA´s fastest growing metropolises), there have been 75% persistence over the last 3 decades (Yang & Lo, 2002) in [54]. [56] determined a persistence of 94, 2% in the community of Madrid – Spain. [57] accounted 93, 3 of landscape persistence in the State of Mexico – Mexico. Finally, [30] also detected a persistence of 80, 5% in the Catamayo-Chira Basin (Ecuador – Peru).

Although the persistence dominates the landscape, as usual, the persistence value of the Boconó River Basin can be considered slightly lower in comparison with those values above mentioned. This fact is important to highlight, considering that the whole river basin is defined as “Protected Area”, with a portion of the surface area also belonging to the Guaramacal National Park.

In the second period the total change was slight higher, with 18464, 7 ha affected by a type of change, representing the 34, 35% of the total area, and the persistence value descended to 65, 65 % of the total surface (35297, 46 ha).

As seen on Figure 2, the change have been occurring in the middle – lower part of the river basin, basically across the sloping dissected areas, the river valley and some extensive quaternary landforms located in the lowest part; in this case, the LULC categories coexist in a very intricate way, showing a very complex and strong patching effect, which is typical of landscapes where the categories are highly fragmented, originating the so – called “chessboard effect” or “chessboard landscape” [58].

5.2. Landscape dynamic: A more detailed view of changes in the River Basin

A more detailed analysis of the transition Matrix derived for the two combined time-periods (T0-T1 and T1 – T2), using the approach proposed by [54], lead to interpret the changes in a more detailed perspective, as follows:

5.2.1. Net change and swapping

The Table 5 resume the landscape dynamic observed for the period T0 – T1. S-Shr was the most dynamic category in the river basin during this period, having a total change which represent the 22,4 % (12037,6 ha), of the total surface. It showed also the highest values for gain and losses respect the rest of LULC. During the period, S-Shr gained 14, 35% of surface area, losing at the same time 8 % to other categories. This category has also the highest value for swapping (16,1 % of the surface area), which means that this LCC constantly experienced changes during the period, losing surface area to other categories and gaining at the same time area from other categories whose changed to this one. Thus, 72% of the change for this category occurred as swapping-change dynamic.

Figure 2.

Persistence and changing area in Boconó River Basin

The second more dynamic category in the area was Sm-F, which experienced a total change of 4485, 51 ha, representing the 8, 3% of the total surface area. In this period Sm-F gained 1593, 27 ha (third highest value), which in many cases could represent an expansion of the shade coffee plantations in the area (included in this category). However, it lost 2892, 24 ha (second highest value) to other categories, representing an important reduction of the forested cover in the area. The category has the third highest value of swapping (3186, 54 ha), which suggest that the Sub-montane Forest also experienced a swapping-change dynamic.

The third category experiencing important changes in the period is the Oc-F, with a total change value of 3721, 41 ha, (7 % of the total area). The Open-cleared Forest gained the fifth biggest portion of surface: 1198, 35 ha, suggesting that the clearcutting and logging in the lowest part of Tmc-F were intense during the period. However, it lost 2523, 06 ha (third biggest amount) to other categories, showing that the clearcutting and logging was also intense within the category. A total of 2396, 7 ha (fifth highest value) were swapping-change dynamic for this category.

GainLossTotal ChangeSwapAbsolute value of net change

Table 5.

Landscape Dynamic in the Boconó River Basin for the Period T0 – T1 (1988 – 1997).

The fourth position in terms of total change (3542, 31 ha), gains (1565, 1 ha), loses (1977, 21 ha) and swapping (3130, 2 ha), is for Grassland; the balance between gains and losses, as well the swapping value, suggest that this category has a strong interaction with other LULCC. The fifth changing category with a total change of 3450, 78 ha (6, 4 % of the total area) is Cro-L, suggesting that the cropping area also experienced important changes during the period. The category gained 1789, 2 ha, which is the second highest value for the period, losing also 1661, 58 ha (sixth value). With the second highest value (3323, 16 ha), Cro-L experienced also a swapping-change dynamic in the area.

Tcm-F is located in the sixth position of total changes, with a total value of 2482, 56 ha (4, 6 % of the total). The category gained 792, 45 ha (seventh value), but lost 1690, 11 ha; meanwhile, 1584, 9 ha were accounted as swapping-change. Finally, Gr-An showed the seventh highest change, with 2270, 16 ha (4, 2 % of the total area). It gained 1085, 49 ha and lost 1184, 67 ha, with a swapping value of 2170, 98 ha.

Despite of the dynamic above described the values for net change shows some differences among the positions between categories. Having the highest net value of 3393, 09 ha, the S-Shr remains as the most dynamic category for the period. The Oc-F had the second highest net change value (1324, 71 ha), and the third position was for Sm-F (1298, 97 ha). The Tropical Montane Cloudy Forest had the fourth highest net change value (947, 66 ha), followed by Gr-L (412, 11 ha), and the sixth position is for the category Ur-U, with a net change value of 295, 2 ha (most of the change in this category is net change, as usual), and a swapping value which tends to be cero. These values lead to affirm that the LCC and particularly the Forested LCC experienced the most important net changes in the river basin during this period.

The Table 6 resume the landscape dynamic for the second period T1 – T2. Some slight differences can be observed respect to the last period. S-Shr remains as the most dynamic category, with a total change value of 11750, 85 ha (22 % of the total area). It gained 6303, 42 ha and lost 5447, 43 ha. The 93% of the total value for this category (10894, 86 ha), occurred as swapping-change dynamic. Sm-F remains in the second position, with a total change of 3638, 88 ha (7 % of the total area). It gained less surface than in the last period (846, 99 ha), which is the seventh observed value for the period. Meanwhile, the losses remained high, having the second highest value for the period (2791, 89 ha). A total of 1693, 98 ha changed in a swapping-change form.

GainLossTotal ChangeSwapAbsolute value of net change

Table 6.

Landscape Dynamic in the Boconó River Basin for the Period T1 – T2 (1997-2008)

The third category experiencing changes in the period is Gr-An, with a value of 3526, 83 ha (6, 6% of total area) for total change. It had the second higher value for gains in the period (2588, 85 ha), meanwhile the losses (937, 98 ha), were lower in comparison to the last period. Of the total value, 1875, 96 ha changed in a swapping-change form. The fourth position in this period is for Cro-L, having a value of 3463, 56 ha (6, 4% of the total area). Cropland gained 2000, 25 ha (the 3rd highest value) during the period, losing 1463, 31 ha (5th value), which can be explained for the type of agriculture applied in the area (small/scale agriculture with shifting cultivation and slash and burn practices). This could explain the high value for swapping (2926, 62 ha) which is the third highest value for the period.

The Gr-L had a total change of 3378, 51 ha (6, 3% of total area), as the fifth changing category. It maintained the same trend as in the last period, gaining 1731, 78 ha, losing 1646, 73 ha, with 3293, 46 ha as swapping-change value. The sixth position in this period was for the Tmc-F, which showed a total change of 2901, 24 ha (5, 4% of the total area). It showed the same trend for gain as in the last period (877, 5 ha), but the losses were quite higher (2023, 74 ha), with 1755, 0 ha as swapping-change dynamic value.

Finally, the Oc-F descended to the seventh position in the period, showing a total change of 2103, 84 ha (3, 9% of the total area). It gained 990, 63 ha, and lost 1113,21 ha, with a swapping value of 1981, 26 ha for the period.

The dynamic showed by the net change values changed slightly respect the last period. The category with the highest net change value was Sm-F (1944, 9 ha), followed by Gr-An (1650, 87 ha); and the Tmc-F reached the third position, with a net change of 1146 ha. S-Shr descended to the fourth position with 855, 99 ha, followed by Cro-L (536, 94 ha) and Ur-U in the sixth position, with a net change value of 136, 08 ha (most of the change occurring as net change).

5.2.2. Systematic Inter-category transitions in the landscape system

Now is possible to derive the categorical trajectory of the changes which have been occurring in the river basin during the considered period. The Table 7 accounts for the most important inter-category transitions for T0-T1 in terms of Losses. The magnitude of the ratio (fifth column) indicates in all cases the strength of the systematic transition between categories [54].

The first thirteen rows on Table 7 indicate spatial patterns or transitions affecting the Forested Land Covers in the River Basin: Tmc-F, Oc-F and Sm-F. These transitions indicate changes associated with deterioration, decrease or disappearance of the Forested areas, depending on the LULC category for which the forested categories have been migrating during the period. For example, the first transition process: Tmc-F – Oc-F indicate that the Tropical Montane Cloudy Forest changed to Open-cleared Forest in 3,764 times more than would be expected If the change were to occur randomly, losing 348,65 ha more than the expected value. This transition, together with the second one, indicate that the TMCF is changing systematically to an intermediate stage (Open-cleared Forest or Successional Shrubland), before it can finally change or migrate to any human–induced types of Land Use categories (Gr-An or Cro-L). No transitions from Tmc-F to Land Use categories were observed. Similar transitional trends were observed in the Highlands of Chiapas – Mexico by [13] and [59], being also described in two different regions in Chile [60][61].

The processes driving the transitions of the Tmc-F are basically associated with: clearcutting, logging, wood extraction and also plants and non-wood extraction. These processes could have been occurring in a successive way, and particularly the logging is probably occurring in a selective form, as observed during the field validation. The selective extraction or harvesting of non-wood products (like Orchids and Bromeliads), has been also reported as a critical problem occurring in this ecosystem [9].

Another example is the transition Sm-F – Ero-L, indicating that in this portion of the surface area, the clearcutting/ logging processes derived in severe land degradation processes like erosion in 6,428 times more than expected, affecting 12, 33 ha. The rest of transitions contribute to explain the other change patterns occurring in the rest of categories, particularly in the human-induced types of Land Cover.

Table 7.

The most systematic transitions occurred in T0-T1, in terms of Losses

As seen on Table 7, Gr-L is basically migrating to Gr-An (174, 33 ha), Cro-L (316, 53 ha) and S-Shr (1224, 45 ha), and with less importance, to Fl-P (31, 32 ha) and Ur-U (31, 32 ha), respectively. Gr-An is basically migrating to Gr-L in 2,146 times more than expected (230, 4 ha). This contributes to explain the high swapping value observed for Gr-L during the period. The category Cro-L migrated to Ur-U in 3,164 times more than expected (98, 1 ha); to Fl-P in 2,874 (49, 05 ha), and to S-Shr in 1,187 times more than expected (846, 63 ha). Particularly the transition Cro-L – Fl-P indicates that the hydrological dynamic of the river, especially the peak flows or flooding events, affected cropping areas. The transition Ero-L – Fl-P suggests an intense hydrological dynamic during the period, which augmented the sediments emission of the river. [62] determined that the yield of sediments in the whole catchment area have increased by 914 % with respect of the estimated value in order to build the Boconó-Tucupido Dam System, located dowmstreams in the lowland region.

The transition Ur-U – Fl-P also suggest that the hydrological events occurred during the period, affected the urban area of Boconó city, which had been expanding across the fluvial plain of the River; it can be corroborated some rows below, with the transition Fl-P – Ur-U, in which the urban area grew up across the Flooding Plain 11,625 times more than expected (6,57 ha). Important flooding events occurred in 1988, 1989, 1991 and 1995 were analyzed by [63]; unfortunately, the historical data for the River Basin is quite deficient and no more reference data exist since 1997.

Finally, the transitions for the category S-Shr suggest a trend for the category to migrate to the human-induced types of Land Cover categories Gr-An (3,070 times more than expected); Cro-L (2,179 times more than expected) and Ur-U (0,114 times more than expected). The rest of the transitions suggest a regeneration process. Shrubland was also observed as a highly dynamic category in the Kalu District-Ethiopia by [64], and also in Central Chile by [60], which can be explained by the forms of cultivation above mentioned, mostly typical in these regions.

The Table 8 shows the most systematic inter-category transitions occurred in the period T0 – T1 in terms of gain. The first twelve transitions are associated to changes in the Forested Land Covers. Particularly the transition Sm-F – Ero-L indicate erosion processes occurring after the clearcutting of the Sub-montane Forest, in 5,489 times more than expected, affecting a total of 12,33 ha. On the other hand, the transitions Gr-An – Gr-l (4,760); Gr-An – S-Shr (2,231), and Gr-An - Sm-F (0,232) suggest a regeneration/revegetation process.

As seen on Table 8, the cropland area in the river basin is growing at the expense of the categories: Oc-F (176, 76 ha), Sm-F (339, 48 ha), Gr-L (316, 53 ha), Gr-An (100, 89 ha), and S-Shr (766, 53 ha). On the other hand, the Gr-An gained surface area migrating basically from: Oc-F (168, 93 ha), Gr-L (174, 33 ha), Cro-L (70, 11 ha), and from S-Shr (497, 34 ha).

The transition Cro-L – Sm-F could to indicate regeneration, or perhaps a change to coffee plantation, or a combination of both scenarios. The transition Cro-L – Gr-L could be explained by the type of cultivation usually practiced in the area, above mentioned.

Table 8.

The most systematic transitions occurred in T0-T1, in terms of Gains

The fact that the urban areas have been growing at the expense of croplands is corroborated again with the transition Cro-L – Ur-U, which indicates that the urban areas grew up from Cropland in 7,028 times more than expected (98,1 ha). The urban areas also grew up at the expense of other categories: Sm-F (61, 92 ha), Gr-L (31, 31 ha), S-Shr (84, 06 ha) and Gr-An (9, 72 ha). On the other hand, the Fl-P grew up at the expense of Cro-L in 5,108 times more than expected, affecting 49, 05 ha.

The transition Ero-L – Tmc-F suggest a regeneration/revegetation process, showing a high level of resilience for the TMCF to be regenerated after such disturbances like landslides, as in this case. The transition Ero-L - Fl-P focuses a source of sediments which were transported by the river during the period. On the other hand, the transition Fl-P – Ur-U confirms the fact that the urban areas (in this case, the urban area of Boconó city) is expanding through the Flooding plain. The last transitions help to confirm the higher swapping-change dynamic associated to the category S-Shr.

The Tables 9 and 10 resume the most systematic transitions occurred in the second period (T1 – T2) in terms of losses and gains, respectively.

As seen on Table 9, the number of rows accounting for changes in the Forested LC was reduced to 9, because of a slight reduction in the transitions of Sm-F, which explains the reduction in the swapping value observed in the category for this period.

The same trend in the transitions for the Tmc-F can be observed in this period, but additionally 5,31 ha of the area covered by the category was affected by erosion processes, particularly landslides. An incipient transition process for the Sa-P occurred during the period, suggesting that some changes derived by anthropogenic pressure have been occurring in the Páramo ecosystems of the river basin. The growing anthropogenic pressure over the Sub-Andean Páramo in the study area was already reported by [65].

The categories Gr-L, Gr-An and Cro-L show the same transitional trends as in the last period. The Urban use continued to growing up at the expense of croplands and the flooding plain, and at the same time, the urban area continued being affected by peak flows or flooding processes. Finally, the S-Shr showed migrating trends to Gr-An (3,168), Cro-L (1,719), to Gr-L (1,507) and to Oc-F (0,910) also.

Respect to the gains in this period, the Table 10 illustrates the trend, where the first ten rows show the changes affecting the Forested LCC. In general, the trends and patterns for the transitions observed on last period remain during the second period.

The category Gr-L showed less intensity in the swapping, meanwhile Cro-L gained surface at the expense of Oc-F (66, 78 ha), Sm-F (269, 37 ha), Gr-L (361, 17 ha), Gr-An (172, 44 ha), Fl-P (36, 9 ha) and S-Shr (1037, 97 ha). Gr-An gained surface migrating from Gr-L in 2,142 times more than expected (502, 83 ha), and also from Cro-L (213, 39 ha), and from S-Shr (1571, 40 ha).

The urban areas continued to growing up in the period, gaining surface area basically from Sm-F (43, 38 ha), Gr-An (3, 78 ha), Cro-L (46, 08 ha), Fl-P (5, 85 ha) and S-Shr (31, 95 ha). Particularly the urban area growing close or into the category Fl-P is vulnerable to the river dynamic. During this period the category S-Shr reported systematic transitions with all the rest of the categories, which explain the high value for swapping-change for the category in this period.

Table 9.

The most systematic transitions occurred in T1-T2, in terms of Losses

Table 10.

The most systematic transitions occurred in T1-T2, in terms of Gains


6. Implications of the observed LULC changes for the watershed management and land use planning

The dynamic of the LULC in the Boconó River Basin for the considered period and through the approach used in this project, lead to establish key elements and a support basis to be considered in the planning processes at the watershed level or even at regional planning level also. Considering that the Boconó River Basin constitute a double “Protected Area”, which has a paramount importance for the development of the water resources in the lowlands, the evaluation of LULC change under the ecosystem approach represent a innovative variation respect the traditional LULC evaluations, in which the LULC are usually considered categories in an abstract sense. In this case, the Land Cover categories are essentially valuable ecosystems which have an ecological richness as well as complementary environmental attributes, being very important to the conservation and sustainability of the three basic land resources: water, soils and biodiversity.

The systematic transitions show the trajectory or directionality of the changes in a categorical sense, leading to identify not only the categories which are more dynamic in a spatial-temporal perspective, but also the possible biophysical and anthropogenic processes driving the transitions. When both interpretations are correctly established, they simply lead to define the key elements to be considered in the land planning processes:

  1. the way how the land resources have been used in the river basin during the last twenty years

  2. the form how the land cover categories as ecosystems have been affected

  3. the trends existing for the different Land Use/Land Cover categories, in a spatial/temporal perspective.

Particularly the spatial visualization (geographical visualization) results in a undoubtedly helpful tool for the planning process, allowing to perceive how these trends are spatially occurring, where are occurring specific processes accounted for problems to be solved, and where these problems are more diverse or intense (hot spots).

As an example, the Figure 3 show the geographic visualization of the transitions for the three main forested Land Cover Categories (LCC) (Tmc-F, Oc-F and Sm-F), for the period T0-T1. The transitions occurred during the Period T1-T2 are displayed on Figure 4. A simple observation of the maps, based on the systematic transitions above described, can lead to the following statements:

1.- The changes affecting the forested land covers, particularly the Tcm-F and the Sm-F tends to be produced in the boundary area between categories. The same trend was observed by [30] in Ecuador. This lead to define belts of clearcutting / logging, which are also called “hot fronts” of deforestation [9], being more evident for the categories: Tcm-F and Oc-F. In the Sub-montane Forest, the belts or “hot fronts” are not clearly defined, because this land cover is highly fragmented among the area. The “Río Negro” Sector located at the upper Boconó River (Figure 4) was severely affected by the changes on the three types of LC, indicating that the processes: clearcutting, logging, wood extraction and non wood & plant extractions were more intense in this sector, during the period. The sector could be defined as “hot spot” or “red flag area”, considering that the deforestation and the LC change is occurring in the sector where the most important streams-sources of the river are located. This sector covers almost the 40% of the stream network area, having therefore the greatest water yield [42].

2.- Observing the two maps, is evident that in the first period, the Open-cleared Forest was systematically reduced among the river basin, meanwhile in the second period, the transition of the Tropical Montane Cloudy Forest was clearly spatially intensified. This lead to corroborate the fact that the dynamic of the TMCF is characterized by a systematic and progressive change, in which the category is migrating to an “intermediate” stage or LCC like Open-cleared Forest or Successional Shrubland, and in other successive stage it can to migrate to another LC or LU categories.

3.- Although the “Guaramacal National Park” was created on 1988, covering the flank south-east of the river basin [41], a “hot front” of deforestation can be observed in the inferior border of this protected area (Figure 3), which clearly increased during the second period (Figure 4). This fact reveals that the creation of the Park has not been completely effective in the protection of the ecosystems included in the protected area.

Figure 3.

Transition area for the Forested Categories in the period T0-T1

4.- The transitions Sm-F – Fl-P; Cro-L – Fl-P and Ur-U – Fl-P suggests a relevant hydrological dynamic occurring during the period studied. The LC Flooding Plain changed actively on last 20 years, accounting for important events like peak flows or even flash-floodings, which expanded the limits of the category among the area, affecting other categories like Sm-F, Cro-L and Ur-U. The dynamic accounted for the Forested LC and the increase of cultivated soils and grass could have been playing a role in the intensification of the hydrological events. The ecological conditions, and particularly the type and density of the land cover play a very important role in the hydrological behaviour and the hydrological response of the landscape. Many authors like: [9] [36] [38] [66] [67] [68] and [69] have been highlighting the importance of the forest ecosystems in the hydrological patterns. Particularly the TMCF is considered as “producer-water forest”, playing a paramount role in the rainfall dynamic, as well as the transpiration, interception, water budget and streamflows [9] [38]. Thus, the systematic reduction of this kind of forest may significantly reduce the rainfall interception, probably leading to an even higher streamflow in the area.

Figure 4.

Transition area for the Forested Categories in the period T1-T2

5.- The transitions Sm-F – Ero-L; S-Shr – Ero-L; and Cro-L – Ero-L, indicate that the area is highly susceptible to soil degradation processes like sheet erosion, rill erosion, landslides and so on, processes which have been activating through the migration of Forested LCC to other categories like Cro-L. Only intense erosive processes like landslides were observed in the classification. However, [39] identified severe erosion processes, especially sheet erosion, in the San Miguel and San Rafael Watersheds (within the study area), which are spatially extended due the high accessibility (intricate road network), the fragile soils and the highly jointed bedrocks.

The accessibility (roads network) has been considered as one of the most important and critical drivers facilitating the LULC change in many regions worldwide [1] [39] [57] [67] [70] [71] [72]. With the exception of the “Río Negro” Sector (See Figure 3), the Boconó River Basin presents a moderately high accessibility [43] [45] [63]. The results obtained by [39] in San Miguel / San Rafael watersheds through a regression tree analysis, revealed that the accessibility had the greatest level of contribution in the occurrence of soil erosion in the area, being the sectors where the cropland have been progressively expanding during the last decades. The occurrence of erosion processes was directly associated to the distance to the road network. This suggest that the accessibility could play a determinant role explaining the intensity and spatiality of the changes that the LULC have been experiencing in the River Basin, as demonstrated by studies conducted in other regions [1] [57] [70-72]. Due the nature and complexity of the variables usually involved, a rigorous analysis of the drivers of LULC change in the area was out of scope of this project, so that further research in this subject is strictly necessary in the near future, in order to comprehensively determine the causal relationship of the factors influencing the changes that affect the River Basin.

The Figures 5 and 6 show the transitions occurred in the Land Use Categories during the first and the second period, respectively. It can be clearly observed where the LUC grow up more intensively in the two periods. The superior window show the San Miguel – San Rafael Watersheds, the sectors where the croplands and the grass anthropogenic grew up more intensively for both periods. These are the sectors which have the most relevant problems related with land degradation in the area, as studied by [39]. The inferior window

Figure 5.

Transition area for the Land Use categories in the period T0-T1

show the expanding process that the Boconó city experienced during the two periods, showing how the city has been expanding among the flooding plain, in areas susceptible to be flooded. The transition Ur-U – Fl-P clearly indicates that some urban sectors have been damaged during the two periods analyzed.

All these interpretations constitutes important tools having practical importance for the institutions or stakeholders involved with the environmental and land planning at local/regional level, being a rational basis to design new plans, or even to improve those which already exists, in order to guaranty the optimization of the natural resource uses in the river basin. This is very important to encourage the effectiveness of the protective figures defined for the whole river basin, accounting for a more sustainable evolution of the LULC in this important “water resource area”.

Figure 6.

Transition area for the Land Use categories in the period T1-T2


7. Conclusions

The methodological approach combining the multitemporal LULC evaluation, together with the ecosystem approach and the inter-category transitional method, represented a very useful tool to define, to describe an to analyze the LULC system in the Boconó River Basin and the changes occurred in the last 20 years. The study demonstrated that the categories: Successional Srhubland (S-Shr), Sub-montane Forest (Sm-F), Open-cleared Forest (Oc-F) and Cropland (Cro-L) were the most dynamic among the two considered periods, accounting for the highest total change value, as well as gains, losses, swapping and net change.

The study also demonstrated that the changes and the reduction showed by the Tropical Montane Cloudy Forest in the area, cannot be directly associated to the expansion of land use categories like Cropland or Grass Anthropogenic. At least on the last 20 years, the TMCF have been changing to an intermediate condition for LC, basically to Open-cleared Forest (Oc-F) and Sucessional Srhubland (S-Shr). Even when the TMCF is under anthropogenic pressure, it can be only associated with logging, wood and timber extraction, as well as the extraction of non wood products and plants.

The systematic transitions that have been occurring in the LULC categories reveal that the land uses Cropland (Cro-L) and Grass Anthropogenic (Gr-An) have been growing, gaining surface basically from Sucessional Shrubland (S-Shr), Sub-montane Forest (Sm-F), and Grassland (Gr-L).This justify the higher values for swapping-change, observed in these categories. On the other hand, the urban areas (Ur-U) have been growing basically at the expense of Cro-L, Gr-L and Fl-P.

The systematic transitions Sm-F – Fl-P; Cro-L – Fl-P and Ur-U – Fl-P, as well as the variation of the category Fl-P during the period, suggest an intense dynamic of the river, and the occurrence of high peak flows and important flooding events during the period, which have been affecting the urban expanding area, as well as croplands. Probably, the decrease of the forested areas, and particularly the TMCF, as well as the increase of the croplands and the grass-anthropogenic, could be directly affecting the hydrological dynamic in the river basin, particularly the behavior of the seasonal flows.

Finally, the systematic transitions helped to focus specific processes that suggest the existence of problems which need to be solved into the land use planning or the watershed management processes. The “hot fronts” of deforestation could be considered as critical areas or priority areas in order to promote the conservation/preservation of the valuable ecosystems as the TMCF, helping to define “area-oriented policies” to ensure the water resources management in the river basin.

Further rigorous research about the associated drivers for LULC change in the area is strictly necessary, in order to reach a comprehensive understanding of the dynamic and transitions of the LULC categories identified and characterized in this project, seeking to encourage the future decisions for land use planning within the watershed management at regional and local level.


We thank to the following institutions which helped in the development of this project: The United States Geological Survey (USGS) and the Institute of Geography (IGCRN) – Universidad de Los Andes, Venezuela, which were the provider of the LANDSAT scenes used in the classification process. The Centre of Geoinformatic and GIS (GIS Zentrum) of the Eberhard Karls University – Tübingen - Germany, which provided the technical support, software and personal who helped during the development of the project. Finally, thanks to DAAD (German Academic Exchange) and FUNDAYACUCHO (Fundación Gran Mariscal de Ayacucho), which have been providing the financial support for the development of the PhD program.


  1. 1. VerburgP.OvermarsK.WitteN.2004Accessibility and land-use patterns at the forest fringe in the northeastern part of the Philippines. The Geographical Journal, 1703238255
  2. 2. KrishnaV.BadarinthK.2004Land use changes and trends in Human Appropriation of Above Ground Net Primary Production (HANPP) in India (1961-98). The Geographical Journal, 17015163
  3. 3. TurnerB.LambinE.ReenbergA.2007The emergence of land change science for global environmental change and sustainability. PNAS, 104522066620671
  4. 4. MannionA.2002Dynamic World. Land cover and Land Use Change. London. Hodder Headline Group.
  5. 5. BormannH.BreuerL.GräffT.HuismanJ.CrokeB.2009Assessing the impact of land use change on hydrology by ensemble modelling: IV. Model sensitivity to data aggregation and spatial (re-) distribution. Advances in Water Resources, 32171192
  6. 6. HouetTh.VerburgP.LovelandTh.2010Monitoring and modelling landscape dynamics. Landscape Ecology, 252163167
  7. 7. LambinE.1997Modelling and monitoring land-cover change processes in tropical regions. Progress in Physical Geography, 213375393
  8. 8. LambinE.GeistH.LepersE.2003Dynamics of Land-Use and Land-Cover Change in Tropical Regions. Annual Review of Environmental Resources, 28205241
  9. 9. BonellM.BrujinzeelL. .Edit2005Forest, water and people in the humid Tropics. Past, present and future hydrological research for integrated land and water management. Cambridge. Cambridge University Press. and UNESCO.
  10. 10. ArmenterasD.RudasG.RodriguezN.SuaS.RomeroM.2006Patterns and causes of deforestation in the Colombian Amazon. Ecological Indicators, 6353368
  11. 11. HadguK.2008Temporal and spatial changes in land use patterns and biodiversity in relation to farm productivity at multiple scales in Tigray, Ethiopia. Wageningen University. Wageningen, The Netherlands. PhD thesis.
  12. 12. PoudelK.2003Watershed Management in the Himalayas. A Resource analysis approach. Delhi. ADROIT Publishers. 361 pp.
  13. 13. CayuelaL.BenayasJ.EcheverríaC.2006Clearance and fragmentation of tropical montane forests in the Highlands of Chiapas, Mexico (1975-2000). Forest Ecology and Management, 226208218
  14. 14. RichterD.SaplacoS.NowakP.1985Problemas de gestión de las cuencas en las tierras altas tropicales húmedas. La Naturaleza y sus Recursos, 2141021
  15. 15. BarriosA.1994Diagnosis del enfoque de manejo de cuencas y el continuo deterioro de los recursos naturales en las cuencas altas de Venezuela. Memorias del 11 Congreso Latinoamericano de Manejo de Cuencas. Mérida- Venezuela.
  16. 16. LorupJ.RefsgaardJ.MazvimaviD.1998Assessing the effect of land use change on catchment runoff by combined use of statistical tests and hydrological modelling: Case studies from Zimbabwe. Journal of Hydrology, 205147163
  17. 17. BrownS.DeeD.2000Using a GIS-Based solution for watershed analysis and automation: a case study. Conference on water resource Engineering and water resources planning and management. Minneapolis.
  18. 18. SteuerJ.HuntR.2001Use of a Watershed- Modeling Approach to Assess Hydrologic Effects of Urbanization, North Fork Pheasant Branch Basin near Middelton, Wisconsin. Water-Resources Investigations Report 014113Middleton. U.S. Geological Survey, in cooperation with the City of Middleton and the Wisconsin Department of Natural Resources.
  19. 19. KrauseP.2002Quantifying the impact of land use changes on the water balance of large catchments using the J-2000 model. Physics and Chemistry of the Earth, 27663673
  20. 20. LegesseD.Vallet-CoulombCh.GasseF.2003Hydrological response of a catchment to climate and land use changes in tropical Africa: case study South Central Ethiopia. Journal of Hydrology, 2756785
  21. 21. BäseF.HelmschrotJ.MüllerH.FlügelW.2006The impact of land use change on the hydrological dynamic of the semi arid Tsitsa catchment in South Africa. Göttingen. Procedure 2nd Göttingen GIS and Remote Sensing Days.
  22. 22. PizarroR.ArayaS.JordánC.FaríasC.FloresJ. y.BroP.2006The effects of changes in vegetative cover on river flows in the Purapel river basin of central Chile. Journal of Hydrology, 327249257
  23. 23. SiriwardenaL.FinlaysonB.Mc MahonT.2006The impact of land use change on catchment hydrology in large catchments: The Comet River, Central Queensland, Australia. Journal of Hydrology, 326199214
  24. 24. MarshallE.RandhirT.2008Spatial modeling of land cover change and watershed response using Markovian cellular automata and simulation. Water Resources Research, 44: W04423.
  25. 25. BrunsD.FetcherN.2008CITY green Watershed Analysis of Toby Creek: An American Heritage River Tributary. Journal of Contemporary Water Research & Education, 1392937
  26. 26. MialheF.GunnellY.MeringC.2008Synoptic assessment of water resource variability in reservoirs by remote sensing: General approach and application to the runoff harvesting systems of south India. Water Resources Research 44: W05411.
  27. 27. BreuerL.HuismanJ.WillemsP.BormannH.BronstertA.CrokeB.FredeH.GräffT.HubrechtsL.JakemanA.KiteG.LaniniJ.LeavesleyG.LettenmaierD.LindströmG.SeibertJ.SivapalanM.VineyN.2009Assessing the impact of land use change on Hydrology by ensemble modelling (LUCHEM). I: Model intercomparison with current land use. Advances in Water Resources, 32129146
  28. 28. HuismanJ.BreuerL.BormannH.BronstertA.CrokeB.FredeH.GräffT.HubrechtsL.JakemanA.KiteG.LaniniJ.LeavesleyG.LettenmaierD.LindströmG.SeibertJ.SivapalanM.VineyN.WillemsP.2009Assessing the impact of land use change on hydrology by ensemble modelling (LUCHEM). III: Scenario analysis. Advances in Water Resources, 32159170
  29. 29. ShillingK.ChanK.LiuH.ZhangY.2010Quantifying the effect of land use land cover change on increasing discharge in the Upper Mississipi River. Journal of Hydrology, 387343345
  30. 30. OñateF.BosqueJ.2010Application of GIS and remote sensing techniques in generation of land use scenarios for hydrological modeling. Journal of Hydrology, 395256263
  31. 31. LiuT.FangH.WillemsP.BaoA.ChenX.VeroustraeteF.DongQ.2012On the relationship between historical land-use change and water availability: the case of the lower Tarim River region in northwestern China. Hydrological Processes. DOI:hyp.9223.
  32. 32. HauckF.1985Soil erosion and it´s control in developing countries. In El Swaify et al. (Ed). Soil Erosion and Conservation. Soil Conservation Society of America. Ankeny, Iowa. USA.
  33. 33. CornielesM.1997Estudio y Modelización del transporte de agua, de elementos disueltos y de particulas en una Cuenca torrencial. Caso del Río Boconó en los Andes venezolanos. Resumen de Tesis presentada ante la Universidad de Avinon y de los Países de Vaucluse para obtener el Título de Doctor en Hidrogeología. Caracas. Ministerio del Ambiente y de los Recursos Naturales Renovables. Servicio Autónomo de Geografía y Cartografía Nacional.
  34. 34. MejiaJ.2000Un modelo suelo-paisaje para la evaluación automatizada de tierras con fines conservacionistas en cuencas altas. Caso: microcuenca del río Zarzales, Edo. Mérida. Centro de Estudios Forestales y Ambientales de Postgrado. Facultad de Ciencias Forestales y Ambientales. Universidad de Los Andes. Mérida. Tesis MSc.
  35. 35. ThapaG.2001Changing approaches to mountain watersheds management in mainland south and southeast Asia. Environmental Management, 275667679
  36. 36. DehnhardtA.PetschowU. .Ed2008Sustainability in River Basins. A question of Governance. Münich, OEKOM Verlag.
  37. 37. OstosO.1975Estudio integral de la Cuenca alta del Río Boconó (Estados Trujillo, Portuguesa y Barinas Venezuela). Estudio preliminar del uso actual de la tierra en el área San Miguel- Boconó- Páramo de Guaramacal. Guanare. División de Edafología del Ministerio de Obras Públicas.
  38. 38. HölscherD.2008Hydrology of natural and anthropogenically altered tropical montane rainforests with special reference to rainfall interception. Biodiversity and Ecology Series, 2129136
  39. 39. CaraballoN.2011Aproximación al estudio de procesos de erosión en cuencas altas tropicales. Caso: Subcuencas de San Miguel y San Rafael, Municipio Boconó. Escuela de Geografía, Facultad de Ciencias Forestales y Ambientales. Universidad de Los Andes, Mérida- Venezuela. Tesis de Grado.
  40. 40. MendezG.RiveroJ.DiazM.2004Actualización del Plan de Ordenamiento y Reglamento de Uso de la Zona Protectora de las Cuencas hidrográficas de los Ríos: Guanare, Boconó, Tucupido, La Yuca y Masparro. Barinas, Edo. Barinas. IV Congreso Forestal Venezolano.
  41. 41. MuñozD.CastilloR.SalasV.2006Estado de Conservación del Parque Nacional Guaramacal. Caracas. BIOPARQUES.
  42. 42. MaciasM.2002Estimación de la producción de sedimentos en la Cuenca alta del Río Boconó, usando regresiones y curvas de duración de caudal. Municipio Boconó, Estado Trujillo. Trujillo. Universidad de Los Andes- Núcleo Universitario Rafael Rangel. Departamento de Ingenieria. Tesis.
  43. 43. CuestaJ.1984Estimación del Indice de Erosividad de la Lluvia (Factor R), en la Cuenca alta del Río Boconó. Universidad de Los Andes. Facultad de Ciencias Forestales y Ambientales. Escuela de Ingeniería Forestal. Trabajo Especial de Grado.
  44. 44. GilJ.2000Estudio Geográfico del Municipio Boconó. Boconó. Fundación La Salle de Ciencias Naturales.
  45. 45. BizzarroC.1985Determinación de los Factores “C” de Cobertura Vegetal y “P” de Prácticas Conservacionistas en la Cuenca Alta del Río Boconó, Estado Trujillo. Mérida. Universidad de Los Andes. Facultad de Ciencas Forestales y Ambientales. Escuela de Ingeniería Forestal. Trabajo de Grado.
  46. 46. BoneJ.HidalgoP.VelasquezF.1985Diagnóstico de la Cuenca del Río Boconó. Informe de avance del Plan de Desarrollo. Boconó, Centro de Ecología.
  47. 47. CihlarJ.2000Land cover mapping of large areas from satellites: status and research priorities. International Journal of Remote Sensing, 21 (6&7): 1093- 1114.
  48. 48. WulderM.FranklinS.WhiteJ.2004Sensitivity of hyperclustering and labelling land cover classes to Landsat image acquisition date. International Journal of Remote Sensing, 252353375344
  49. 49. BaldygaT.MillerS.DrieseK.GichabaCh.2007Assessing land cover change in Kenya´s Mau Forest region using remotely sensed data. African Journal of Ecology, 464654
  50. 50. BurgaC.KlötzliF.GrabherrG.2004Gebirge der Erde. Landschaft, Klima Pflanzenwelt. Stuttgart. Wissenschaftliche Buchgesellchaft.
  51. 51. LEICA GEOSYSTEMS.2007ERDAS Field Guide. Volume Two. Norcross- USA.
  52. 52. ERDAS.2008Change Detection. White Paper. Online Material, available: 13/09/2011]
  53. 53. RamoeloA.2007An innovative method to map land cover changes at a country level utilising hyper-temporal satellite images. A case of study of Portugal. International Institute for Geo-information Science and Earth Observation- ITC. Enschede, The Netherlands. Thesis for the degree of Master of Science in Geo-information Science and Earth Observation for Environmental Modelling and Management.
  54. 54. PontiusR.ShusasE.Mc EachernM.2004Detecting important land changes while accounting for persistence. Agriculture, Ecosystems and Environment, 101251268
  55. 55. VelázquezA.MasJ.DíazG.MayorgaS.AlcántaraC.CastroR.FernándezT.BoccoG.EzcurraE.PalacioJ.2002Patrones y tasas de cambio de uso del suelo en México. Gaceta Ecológica (62): 21- 37.
  56. 56. PlataR.2007Descripción del crecimiento urbano en la comunidad de Madrid en el período 19872000y una aproximación al análisis de factores explicativos. Tutelado del Doctorado en Cartografía, SIG y Teledetección, Universidad de Alcalá.
  57. 57. PinedaN.BosqueJ.GómezM.PlataW.2009Análisis de cambio del uso del suelo en el Estado de México mediante sistemas de información geográfica y técnicas de regresión multivariantes. Una aproximación a los procesos de deforestación. Investigaciones Geográficas, Boletín del Instituto de Geografía, UNAM (69): 33- 52.
  58. 58. LangS.BlaschkeTh.2007Landschaftanalyse mit GIS. Stuttgart. Verlag Eugen Ulmer Stuttgart.
  59. 59. Ochoa-GaonaS.2001Traditional land-use systems and patterns of forest fragmentation in the Highlands of Chiapas, Mexico. Environmental Management, 274571586
  60. 60. SchulzJ.CayuelaL.EcheverríaC.SalasJ.Rey-BenayasJ.2010Monitoring land cover change of the dryland forest landscape of Central Chile (1975-2008). Applied Geography, 30436447
  61. 61. EcheverríaC.NewtonA.NahuelhualL.CoomesD.Rey-BenayasJ.2012How landscapes change: Integration of spatial patterns and human processes in temperate landscapes of southern Chile. Applied Geography, 32822831
  62. 62. GásperiT.2006Condición de sedimentación del Embalse Boconó. Propuesta para su estimación futura. Mérida. Centro Interamericano de Investigación Ambiental y Territorial (CIDIAT). Tesis MSc.
  63. 63. QuevedoM.1997Análisis de la relación precipitación-escorrentía en la Cuenca Alta del Río Boconó, Estado Trujillo. Trujillo. Universidad de Los Andes- Núcleo Universitario Rafael Rangel. Departamento de Ingenieria. Tesis.
  64. 64. TekleK.HedlundL.2000Land Cover Changes Between 1958 and 1986 in Kalu District, Southern Wello, Ethiopia. Mountain Research and Development, 2014251
  65. 65. HidalgoJ.2007Análisis Geográfico para una propuesta de Area Protegida en el Páramo La Cristalina, Estado Trujillo. Universidad de Los Andes. Facultad de Ciencias Forestales y Ambientales. Escuela de Geografía. Tesis de Grado.
  66. 66. NaimanR. .Edit1992Watershed Management. Balancing Sustainability and Environmental Change. New York. Springer Verlag Inc.
  67. 67. ChangM.2003Forest Hydrology. An Introduction to Water and Forest. Florida- USA. CRC PRESS. 373 pp.
  68. 68. WardA.TrimbleSt.2004Environmental Hydrology. II Edition. Boca Ratón- USA. LEWIS PUBLISHERS. 475 pp.
  69. 69. DavieT.2008Fundamentals of Hydrology. Second Edition. London. Routledge Taylor and Francis Group.
  70. 70. SerneelsS.LambinE.2001Proximate causes of land-use change in Narok District, Kenya: a spatial statistical model. Agriculture, Ecosystems and Environment, 856581
  71. 71. BraimohA.VlekP.2005Land-Cover Change Trajectories in Northern Ghana. Environmental Management, 363356373
  72. 72. EtterA.Mc AlpineC.WilsonK.PhinnS.PossinghamH.2006Regional patterns of agricultural land use and deforestation in Colombia. Agriculture, Ecosystems and Environment, 114369386

Written By

Joel Francisco Mejía and Volker Hochschild

Submitted: November 21st, 2011 Published: November 7th, 2012