Master’s thesis
Physical Geography and Quaternary Geology, 45 Credits
Department of Physical Geography
Payment for Ecosystem Services (PES) and Water Resource
Management of the tropical mountain ecosystem páramo
A case study in the northern parts of Ecuador
Ellinor Hallström
NKA 178
2017
Preface
This Master’s thesis is Ellinor Hallström’s degree project in Physical Geography and Quaternary Geology at the Department of Physical Geography, Stockholm University. The Master’s thesis comprises 45 credits (one and a half term of full-time studies).
Supervisor has been Steve Lyon at the Department of Physical Geography, Stockholm University. Examiner has been Stefano Manzoni at the Department of Physical Geography, Stockholm University.
The author is responsible for the contents of this thesis.
Stockholm, 15 June 2017
Steffen Holzkämper Director of studies
Photos on front page: Ellinor Hallström
ABSTRACT
Latin America has pioneered the concept of Payment for Ecosystem Services (PES) as a strategy to improve the management of ecosystem services. Ecuador is not an exception, where many PES schemes have been implemented to protect the tropical mountain ecosystem “páramo” and the water resources these areas are generating for downstream societies. A successful PES scheme needs to achieve both targeted bio-physical objectives and at the same time benefit local conditions while not risking to sacrifice the local demand for ecosystem services. This balance is explored here in a case study focusing on the Río Grande watershed in the highlands in the northern parts of Ecuador by exemplifying community participation in the public PES scheme Socio Bosque (PSB) starting in 2009. The water resource distribution (precipitation, discharge, actual evapotranspiration and potential evapotranspiration) in the watershed was evaluated over the last decades. The local perception of the PSB and its impacts on local and regional water resources were also studied and characterized. The results showed that the annual discharge in the Río Grande watershed has decreased significantly from 1967-2014 and that the annual discharge was significantly lower
between 1997-2015 compared to 1979-1997. Since precipitation did not decrease significantly during this period, the changes of the annual discharge are more likely depended on factors controlling the seasonal distribution of discharge and evapotranspiration in the watershed. For example, large scale land use changes coupled with a significantly warmer climate in the region could be a possible driver.
Of course, this would not exclude other important factors such as changes in water demand and the supply of freshwater from the Río Grande watershed to downstream societies. The results of this case study showed that it is likely too early to see any impacts in the water balance components as a direct response to the implemented PSB scheme. Clearly, this motivates a need for continued evaluation of the local perception and the water resources to ensure that the need and demand for ecosystem services in a long-term perspective are maintained.
RESUMEN
Latinoamérica ha sido pionera en el concepto de Pago por Servicios Ecosistémicos (PSE o PES en las siglas en inglés) como estrategia para mejorar la gestión de servicios ecosistémicos. En Ecuador, se han implementado muchos PSE para proteger el ecosistema montañoso tropical de El Páramo así como los recursos acuíferos que dichas áreas generan para las sociedades que habitan cuenca abajo.
Un esquema de PSE exitoso requiere alcanzar los objetivos biofísicos y respetar las necesidades locales de servicios ambientales. Este equilibrio se ha analizado tomando como objeto de estudio la cuenca hidrográfica del Río Grande en las tierras altas del norte de Ecuador y la participación comunitaria en el programa de PSE denominado Socio Bosque (PSB) iniciado en 2009. Se estudiaron la distribución del agua (precipitación, descarga del agua, evapotranspiración actual y
evapotranspiración potencial) en la cuenca hidrográfica durante las últimas décadas. También se estudiaron los impactos locales y regionales del PSB en los recursos hídricos y los percepción local con respecto a la implementación de PSB. Los resultados muestran que la descarga anual de la cuenca hidrográfica ha decrecido significativamente durante el período comprendido entre 1967 y 2014, particularmente, señalan un decrecimiento considerablemente mayor entre 1997 y 2015 con respecto al período 1979-1997. La precipitación no se redujo durante el período estudiado y, en consecuencia, los cambios en la descarga anual dependen presumiblemente de factores que controlan la distribución estacional de la descarga y la evapotranspiración en la cuenca. Como
ejemplo, los intensos cambios en el uso del suelo junto a un clima regional marcadamente más cálido pueden ser dos condicionantes del fenómeno. Esto no excluye otros factores como los cambios en la demanda y abastecimiento de agua potable en la cuenca del Río Grande en las comunidades que se encuentran distribuidas a lo largo del río. Los resultados muestran que es aún temprano para observar impactos en los componentes del balance hídrico como resultado directo de la
implementación de un esquema de PSB. Esto motiva la necesidad de una evaluación continua de la percepción local y un monitoreo los recursos hídricos para garantizar que las necesidades y
demandas de servicios ecosistémicos en la región se mantengan a largo plazo.
ACKNOWLEDGEMENT
There are several incredible people who I have met during the time I have worked on this project and each have contributed to make this project possible in different ways. I therefore would like to warmly thank each of them. First of all, the community La Esperanza and all of you who participated and were interviewed in this study. Thanks Edison Semanate for being an excellent driver and assistant during the entire field visit. I would also like to thank Pablo Lloret, at the Empresa Pública Metropolitana de Agua Potable y Saneamiento (EPMAPS) for welcoming me to Ecuador, being very supportive throughout the whole project, and together with Ximena Fuentes, for giving me the opportunity to explore and learn more about the different páramo ecosystems in Ecuador. I am also grateful for the support I have received from Altropico, and the data and information provided from the Instituto Nacional de Metrologia e Hidrologia (INAMHI) and the Gobierno Autónomo
Descentralizado de la Provincia Del Carchi. At the Department of Physical Geography at Stockholm University, I want to thank Steve Lyon for your optimism, overall support throughout the whole project and for being quick in response and assistance when needed. I am also very grateful for the support received from Lowe Börjesson, at the Department of Human Geography. Thanks also to Carol Hunsberger at the department of Geography University of Western Ontario in Canada for inspiration to undertake the project. Thanks to the Department of Ecology, Environment, and Plant Science at Stockholm University and the Swedish International Development Cooperation Agency (SIDA) for financial support. I would also like to thank all the friends and families who have helped and encouraged me along the way, especially Camilla Hallström, Stefan Johansson, Bengt Hallström, Geoff Penhorwood, Imanol Rubio, Mariana Semanate, Maritza Cevallos, and Johanna Lundberg.
TABLE OF CONTENTS
1. INTRODUCTION ... 8
1.1. Payment for Ecosystem Services as a management approach of natural resources ... 8
1.2. What is the “páramo”? ... 9
1.3. A case study of Payment for Ecosystem Services in the páramo ... 10
2. DATA AND METHODS ... 11
2.1. Site description ... 11
2.2. The characteristics of Río Grande watershed and the páramo vegetation ... 13
2.3. Climatological and Hydrological Data Considered and Data Quality Assessment ... 14
2.4. Long-term water balance observations ... 15
2.5. Potential impacts of the PSB scheme on the hydrological and climatological parameters ... 16
2.6. Statistical hypothesis testing ... 17
2.7. Qualitative evaluation of the local people’s observations of the bio-physical conditions and the PSB ... 17
3. RESULTS ... 19
3.1. Assessing Data Quality ... 19
3.2. Long term water balance observations ... 22
3.3. Short term water balance observations and impacts of the PSB ... 25
3.4. Local observations of the bio-physical condition and the PSB scheme ... 28
4. DISCUSSION ... 34
4.1. Water balance analysis and local observations of the bio-physical conditions in the watershed ... 34
4.2. Potential páramo and land use impacts on hydrological parameters in the watershed .. 34
4.3. The PSB scheme as a land-water resource management strategy ... 35
4.4. Uncertainties and limitations of the study ... 37
5. CONCLUSIONS ... 38
6. REFERENCES ... 39
APPENDIX
APPENDIX A: THE RÍO GRANDE WATERSHED AND THE PÁRAMO VEGETATION ... 44
APPENDIX B: LIST OF RESPONDENTS... 47
APPENDIX C: INTERVIEW GUIDE ... 48
APPENDIX D: MATRICES USED FOR THE QUALITATIVE ANALYSIS ... 49
APPENDIX E: DATA QUALITY ASSESSMENT; BOXPLOTS ... 51
APPENDIX F: DATA QUALITY ASSESSMENT; MEAN MONTHLY AVERAGES ... 52
APPENDIX G: ANNUAL OBSERVATIONS OF THE HYDRAULIC AND CLIMATOLOGICAL PARAMETERS; BOXPLOTS ... 53
APPENDIX H: STATISTICAL ANALYSIS; WATER BALANCE OBSERVATIONS (18 YRS) ... 55
APPENDIX I: STATISTICAL ANALYSIS; MONTHLY WATER BALANCE OBSERVATIONS (18 YRS) ... 56
APPENDIX J: STATISTICAL ANALYSIS; WATER BALANCE OBSERVATIONS (12 YRS) ... 58
APPENDIX K: STATISTICAL ANALYSIS; WATER BALANCE OBSERVATIONS (6 YRS) ... 60
1. INTRODUCTION
1.1. Payment for Ecosystem Services as a management approach of natural resources
Payment for Ecosystem Services (PES) have been the subject of increasing global popularity since the early 1990s as a strategy to improve the management of ecosystem services (Grima, et al., 2016). PES schemes are often referred to as a volunteer based management strategy, where an investor is paying a seller to sustain the provision of a specific ecosystem service, often (but not always) through conservation practices (Wunder, 2005). The concept of ecosystem services and PES schemes have emerged as a response to the market and policy makers failure to value and consider non-monetary resources of for example, air, water, forests, wetlands and local knowledge (McMichael, 2012;
Muradian, 2013). With economic compensation, the interests of landowners and external actors regarding ecosystem services were expected to increase (Wunder, 2007; Grima, et al., 2016).
Even though economic preconditions have been developed for PES schemes (Börner, et al., 2010) and many PES schemes are in place (Martin-Ortega, et al., 2013). Information and knowledge barriers exist for implementing PES schemes because of long geographical and/or social distances between the providers and investors (Muradian, 2013). An example of this would be highlighted in watershed based PES schemes between upstream providers and downstream beneficiaries of ecosystem services related to the supply of fresh water (Muradian, 2013). Other impediments for PES schemes include institutional preconditions, land grabbing, insecure tenure, overlapping claims, and the lack of information on private tenure (Börner, et al., 2010). The results and outcomes from a PES scheme should therefore not be taken for granted and are typically considered to not be able to replace regular command-and-control strategies of ecosystem services and natural resources (Wunder, 2007;
Börner, et al., 2010; Muradian, 2013). There is thus a risk that the “wrong” land managers or land areas can be targeted due to a poor design of a PES scheme and that the desired hydro-ecological or conservation benefits will not be received consequently (Porras, et al., 2013). Many small-scale landowners, for example, in the Brazilian Amazon are struggling with soil fertility in traditional slash- and-burn systems (Börner, et al., 2010). Due to unequal distribution of land, they would not benefit from regular command-and-control environmental management strategies without compromising with their welfare. Targeting instead the large-scale land owners, who contribute the most to deforestation with PES schemes would be more beneficial (Börner, et al., 2010). However, many ecosystem services are often treated as common-pool or public goods, and this could present a social dilemma in terms of how best to manage (Muradian, 2013). In this case, PES schemes could be an incentive for collective actions, where the effectiveness of the program depends much on the social meanings (context and culture) of the PES scheme (Muradian, 2013). Most of the PES schemes that are in place have only operated for less than 10 years and take shape only through a learning by doing process (Martin-Ortega, et al., 2013). As such, the science (bio-physical and local
improvements), practices and theory behind PES schemes need to be explored (Martin-Ortega, et al., 2013).
Latin America has pioneered the concept and the implementation of PES schemes (Martin- Ortega, et al. 2013; Grima, et al. 2016). Costa Rica is a good example of this as it was the first country in 1997 to implement a public PES scheme (FONAFIFO) (Pagiola, 2008). Since then, Colombia, Bolivia and Brazil have all adopted PES strategies into their public conservation and environmental strategies (Grima, et al., 2016). Ecuador is not an exception and many different types of PES schemes have been implemented during recent years to protect the tropical mountain ecosystem “páramo” and the water resources these areas generate to downstream societies (Southgate & Wunder, 2010). The community based PES scheme in the Pimampiro society is one example where economic
compensation to upstream land owners to protect the water resources in the area has been
conducted (Southgate & Wunder, 2010). In the Ecuadorian capital, Quito a larger PES scheme (FONAG) works to sustain the quality of the fresh water provided from surrounding páramo areas (Southgate & Wunder, 2010). The Program Socio Bosque (PSB) is another public PES scheme that the state of Ecuador implemented in 2008. This PSB scheme is compensating land owners, both
individuals and different collectives such as communities and indigenous groups, with legal rights to land areas of tropical forest, cloud forests, dry forest, high altitude forests and other ecosystems including the páramo (Ministerio del Ambiente Ecuador, 2016). The objectives with the PSB are to conserve the ecological, economic and cultural values of these ecosystems (like water resources);
significantly decrease the deforestation and consequences of global warming; and to improve the life for the people living in rural areas, specifically indigenous and peasant communities (Ministerio del Ambiente Ecuador, 2016) (Ministerio del Ambiente Ecuador, 2017). By 2015 in total 2775 contracts had been signed, covering 1´500´000 ha of different vegetation types (Ministerio del Ambiente Ecuador, 2016). Approximately, 273 contracts are covering 61’000 ha of páramo (Ministerio del Ambiente Ecuador, 2015).
1.2. What is the “páramo”?
The word “páramo” has many different meanings and is typically used to describe an ecosystem, geographic area or a climate condition as much as it is used to describe a zone of life and a
productive site (Hofstede, 2014). However, it is often used to refer to the vegetation zone above the tree limit and below the snow limit in tropical mountain ecosystems that typify South America. This area is dominated by shrubs, herbs and different grass types that are growing in tufts (Beltrán, et al.
2009; Sklenár, et al. 2005). Tropical mountain conditions, similar to the páramo, exist also in other continents such as in the tropical eastern parts of Africa (Hedberg, 1964). Typically, the climate variations over the year are small in these zones, but the temperature shifts can be large between night and day (up to 20 °C) (Buytaert, et al., 2006; Hedberg, 1964). Such conditions are described as
“being summer during the day, but winter during the night” and days with high solar radiation and cold nights with frost are common (Hedberg, 1964). In South and Central America, the páramo forms an interrupted belt along the Andes mountain ridge from Peru to Venezuela with two divergent parts in Panama and Costa Rica (Hofstede, 2014).
Despite the extreme climate conditions in the páramo, the plants have adopted and developed a variety of different characteristics to survive. The genus Espeletia (Appendix A), for example, have evolved super-cooling mechanism of the adult leaves to protect them from freezing and dead leaves retained along the stem to give isolation (Rada, et al., 1987; Goldstein & Meinzer, 1983). The long stems protect buds from the cold at ground level during the night (Smith, 1980). This evolution, also improves the pith water storage capacity and allows for better survival during long periods of water stress (Goldstein, et al., 1985). Due to this adaptation capacity, the flora of the páramo have formed one of the World’s most unique high elevation mountain ecosystem with a rich biodiversity and endemism. In total 3595 species have been found in these biotopes (Sklenár, et al., 2005), including a total of 126 species of the genus Espeletia up to an elevation of 4500 m in the páramo of Ecuador, Colombia and Venezuela (Sklenár, et al., 2005). Only in Ecuador, where the páramo covers
approximately 5 % of the land surface (1’337’119 ha) (Beltrán, et al., 2009) in total 404 genera and 1524 species of plants have been found and is to its size the country where the flora is most diverse (Sklenár, et al., 2005). Apart from the biodiversity, the soils in the páramo (mostly Andisols and Histosols) serve an important ecosystem function since they regulate (store) incoming precipitation to generate a uniform base flow to connecting rivers, like the Amazon basin, throughout the year. In addition, there are many urban areas downstream from the páramo that are dependent on the fresh
water supplies from the páramo’s base flow sustaining soils, among those the cities Quito, Cuenca, Bogata and Mérida (Buytaert, et al., 2004; Sklenár, et al., 2005).
Many of the people living in the páramo in Ecuador have marginalised income sources and are dependent on the land for the cultivations of vegetables and potatoes for the household and/or to sell on local and regional markets. Alternatively, sheep and cattle are kept and allowed to graze in the páramo (Mena Vásconez & Hofstede, 2006). Current research agrees that intensification of these types of land use practises increase the bare land and subsequently the erosion problems in the páramo, this in turn affects the water retention capacity (Poulenard, et al., 2001; Podwojewski, et al., 2002; Buytaert, et al., 2004; Buytaert, et al., 2005; Buytaert, et al., 2006), the slow hydraulic response and the uniform distribution of water to connecting streams in the páramo. As a consequence this could result in a smaller base flow and larger peak flows (Buytaert, et al., 2005). These research efforts have taken place at such small scales that the extent to which large scale land use changes affect the hydrodynamic nature of the páramo is still not completely understood. There are few discharge stations in these high-altitude regions (Mosquera, et al., 2016). The dynamics of runoff and the interactions with rainfall is therefore not well studied and far from completely understood (Mosquera, et al., 2016). In addition, the topography in the tropical Andes is highly irregular and the weather is influenced both by winds from the Atlantic and the Pacific Ocean. Evaluating changes in the climate and its impacts on water resources is therefore difficult (Buytaert & De Bièvre, 2012).
However, due to these uncertainties there is a concern that the agriculture zone is expanding in the páramo and the deforestation increasing (Vanacker, et al., 2003; Mena Vásconez & Hofstede, 2006), while the potential impact of these changes are largely unknown for downstream populations.
The PES schemes like the public PSB scheme have been argued to be a good and effective conservation strategy to protect natural ecosystems in Ecuador from deforestation (de Koning, et al., 2011). The success with PES and PWS schemes in Latin America and in Ecuador have, however, not been quantified to a large extent (Martin-Ortega, et al., 2013; Grima, et al., 2016). Most of the criticisms of PES and PWS have been focusing on the schemes’ neoliberal approach of making nature and ecosystem services a commodity for the benefit of commercial purposes in downstream areas (Rodríguez de Francisco, et al., 2013; Boelens, et al., 2014). People in upstream areas, who are dependent on the land and the ecosystem services from the páramo for their livelihood have been argued to not benefit economically from participate in PES schemes (Boelens & Rodríguez de Francisco, 2014; Boelens, et al., 2014). In several watersheds, for example, PES schemes in
combination with other general environmental laws have potentially made the life more restricted for families living in upstream areas (Boelens & Rodríguez de Francisco, 2014; Boelens, et al., 2014).
The population size and poverty level of the people and communities participating in the PSB are not considered in the PSB contracts, and it could as such be a challenge to combine poverty reduction targets with the conservation objectives of the PSB (Krause & Loft, 2013).
1.3. A case study of Payment for Ecosystem Services in the páramo
Despite (or perhaps due to) the complexity whereby a PES scheme includes both bio-physical conditions and social inference, few studies of the PES schemes in Ecuador have considered the bio- physical aspects. Such as, for example, impacts on water resources, in combination with the
underlying conservation motivations for a PES scheme while assessing the participant’s perspective.
This current study aims to contribute to fill this lack of information by the evaluation of a páramo land targeted with a PES scheme and by focusing on to evaluate potential impacts the PES scheme could have on the provision of water the páramo land is generating to downstream societies. An interdisciplinary approach has been used, which is advantageous when considering problems that are increasingly complex and intertwined. Such an approach allows a problem to be studied from
different angles (e.g. both the physical and the human dimension) and thereby can both contribute to and challenge different perspectives (McNeill, et al., 2001). The study focuses on the Río Grande watershed in the northern parts of Ecuador. This study site was chosen after an evaluation of
available hydraulic observations in the Ecuadorian páramo in combination to the location of different implemented PES schemes. In the Río Grande watershed, the local community has legal rights to a mountainous area with a páramo ecosystem. In August 2009, this community was among the firsts to participate in the PSB. The distribution of water in the watershed over the last decades is evaluated from a water balance approach. This was accomplished by analysing general trends and long-term step changes in the precipitation (P), discharge (Q), actual evapotranspiration (AET), potential evapotranspiration (PET) and temperature (T) in the watershed. Secondly, the study considers potential impacts of the PSB scheme on the water balance components (P, Q, AET and PET) of the Río Grande watershed. In addition, semi-structured interviews were conducted with a selected group of people from the community to assess local perceptions of the PSB, and the impacts of the PES scheme on both local and regional water resources. Taken together, such an interdisciplinary approach will allow for a better understanding of the bio-physical conditions in the watershed, the páramo and the PSB scheme as a land-water management strategy.
2. DATA AND METHODS
2.1. Site description
The river Río Grande in the northern parts of Ecuador (Figure 1), has its headwaters in the páramo of the community “La Esperanza”, on the eastern side of the active stratovolcano Chiles (4748 m.a.sl) (Instituto Geofísico, EPN, 2017). At lower elevations, the river meanders through a mosaic landscape with potatoes cultivations and grasslands mixed with pastures. It then passes by the village Tufiño with a population of approximately 2300 people (GAD Tufiño, 2015) before it converges with the river Játiva o Alumbre and form the river Río Carchi. This is the river that makes up the border between Colombia and Ecuador in this region. The watershed of Río Grande has an area of 54 km2 (INAMHI, 2016) and is the focus of this study both when it comes to the quantitative hydrological analysis and the qualitative data collection. The water of Río Grande is lead through an open channel system to one of the local power plants before the river converges with the river Játiva o Alumbre.
The Río Grande watershed is also supplying the communities downstream in the province with freshwater, among those Tulcán, (Riascos, 2016) a city with a population of approximately 77 175 people (Gobierno Autónomo Descentralizado Provincial del Carchi, 2013). The community of La Esperanza has about 350 families and received legal status on August 01, 1938 when it was registered in the Registro Nacional de Comunas in the Ministerio de Previsión Social y Comunas. It has legal rights to a mountainous land area of 14163 ha in the canton (Cantón) Tulcán between the parishes (parroquias) Tufiño y Maldonado (Vásquez Narváez, 2008). Most of the people within the community live downhill from the páramo land both on the western and eastern side of the mountain ridge and the active stratovolcanoes Chiles and Cerro Negro. On the western side, in the parish Maldonado, the people live in small villages in valleys. In these villages, the climate is temperate to sub-tropical and the people are mainly dedicated to the production of granadilla, blackberries and tamarillo fruit. On the eastern side, where this study is focused, most of the people of the community live in the village Tufiño, located at an elevation of about 3120 m.a.sl. Here the climate is colder and the people are mainly dedicated to the production of potatoes and milk
(Appendix A), which are sold at the local markets. Also, beans and “mellocos” (Ullucus tuberosus) are
cultivated, but mainly for home/personal consumption (Vásquez Narváez, 2008). The temperature shift is high between night and day in the village and the only available source to heat up their houses is burning firewood. Since August 2009, the La Esperanza community has been participating in the national conservation program Socio Bosque (PSB), with 8622 ha of land set aside for
conservation (Figure 1), most of which is páramo (Puetate, 2016). Approximately 40 % of the PSB area is located within the watershed of Río Grande (Figure 1). A PSB contract covers a period of 20 years, which can be renewed after that time. For the first 50 ha registered, the PSB gives $ 30 US/ha/year. For larger land areas of 51-100 ha, the rate is $ 20 US/ha/year (Minesterio del Ambiente de Ecuador, 2011).
Figure 1. Reference system: WGS 1984 UTM Zone 18N. Projection: Transverse Mercator. Data sources: Gobierno Autónomo Descentralizado de la Provincia Del Carchi, Instituto Nacional de Metrologia e Hidrologia (INAMHI). Illustration of the sub catchment of the rivers Río Grande (54 km2) and Játiva o Alumbre (72 km2) to the catchment Río Carchi. Of the total area of the páramo La Esperanza included in the program Socio Bosque (PSB) (8622 ha) approximately 40 % is within the watershed of Río Grande and Játiva o Alumbre. In the map the neighbour watersheds Río Carchi, Río Napo and Río Mira are highlighted and also the metrological and hydrological stations of Precipitation (P), Discharge (Q) and Temperature (T) considered in this
study.
Content may not reflect National Geographic's current map policy. Sources: National Geographic, Esri, DeLorme, HERE, UNEP-WCMC, USGS, NASA, ESA, METI, NRCAN, GEBCO, NOAA, increment P Corp.
2.2. The characteristics of Río Grande watershed and the páramo vegetation
Between the highest point of the volcano Chiles (4723 m.a.sl) (Instituto Geofísico, EPN, 2017) of the watershed of Río Grande and the lowest point at the discharge station (H0091) (3120 m.a.sl) there is a difference of 1603 m in altitude. The vegetation of the páramo changes spatially with different climate conditions (Table 1). By Boada, et al. (2008) in total five different vegetation characteristics of the páramo of La Esperanza were studied: high mountain evergreen forest, herbaceous páramo, páramo of frailejones, dry páramo and montane grassland lakes (Appendix A). The high mountain evergreen forest is the transition zone between forest and páramo vegetation and is generally found at an elevation of around 3000 m.a.sl (Boada , et al., 2008). In the páramo of La Esperanza, 23 species of trees, with a mean high of 3.51 m were found by Salgado (2008) in this zone including Escallonia myrtilloides (Grossulariaceae), Polylepis sericea (Rosaceae), Oreopanax seemanianus (Araliaceae), Miconia latifolia (Melastomataceae), Gynoxys sp. (Asteraceae) and Polylepis incana (Rosaceae). In the herbaceous páramo of La Esperanza (3400-4000 m.a.sl), a total of 88 species of herbs were found by Salgado (2008). These were dominated by grass types that are growing in tufts like Calamagrostis intermedia (Poaceae), Carez muricata (Cyperaceae), Cortaderianitida (Poaceae) y Paspalum sp.
(Poaceae). In the vegetation zone páramo of frailejones (3500-3700 m.a.sl.) a total of 21 species of plants were found by Salgado (2008) including Puya clava-herculis (Bromeliaceae), Loricaria thuyoides (Asteraceae), Hyperium lancioides (Clusiaceae), Gynoxys fuliginosa (Asteraceae) and “El Frailejón” Espeletia pycnophylla ssp. angelensis (Asteraceae) (Appendix A). “El Frailejón” forms big fields of forests in the páramo of La Esperanza and is a plant that has only been found in these northern parts of Ecuador (apart from a divergent population in Tungurahua) and in the Andes of Colombia and Venezuela (Boada, et al., 2008). In the vegetation zone Montane grassland lakes (>2100 m.a.sl.), lakes and lagoons are found and the flora are restricted to the related banks. Laguna Verde is one example in the páramo of La Esperanza (Figure 1). Appendix A, illustrate some of the plants identified at this site during the field visit in August 2016.
Table 1. Different vegetation zones of the páramo of La Esperanza (El páramo del Artesón) studied by Boada, et al. (2008).
Altitude (m.a.sl)
T min (°C)
T max (°C)
Annual P (mm)
Annual PET
(mm) Vegetation characteristic High mountain
evergreen forest 3000-3400 6 17 922 882 Transition zone between forest and páramo
Páramo of
herbaceous 3400-4000 4 13 722 820 Herbs growing in tufts
(Calamagrostis and Festuca) Páramo of
frailejones 3500-3700 5 13 983 805 “El Frailejon” Espeletia
(Asteraceae)
Dry páramo ≥4200 3 12 754 766 Islands of grass and shrubs, a few
mosses and lichens Montane
grassland lakes >2100 - - - - Flora restricted to the banks of
lakes and lagoons.
In total 39 species of birds of 17 families were found by Buitrón (2008) in the páramo of La Esperanza. Among those are many of the families Trochilidae (hummingbirds) and Thraupidae (tangaras). Few individuals of the condor (Vultur gryphus), which is a species nearing extinction in Ecuador, have been seen in the páramo around the volcano Chiles but, but in the study by Salgado (2008) none were found.
The páramo also constitutes the home for many mammals. According to Boada (2008) a total of 19 species were found in the páramo of La Esperanza. Among those were the “puma” Puma
concolor, “Lobo de páramo” Lycalopex culpaeus, “Zorrillo” Conepatus semistriatus, “Oso de anteojos”
Tremarctos ornatus, “Coatí andino” Nasuella olivacea, “Comadreja andina” Mustela frenata and the
“Gato de las pampas” Leopardus pajeros.
The soil of the páramo in Ecuador is composed of a parent material of ash from the Holocene (younger than 10 000 years). Around Quito and on slopes of active volcanos the soil is younger, coarser and richer in primary minerals. In contrast to these northern parts of Ecuador, where the Andisols are older (300 BP) and richer in non-allophanic poorly ordered constituents. These soils are more associated with organic matter (up to more than 15 %) (Poulenard, et al., 2001). For most Andisols it is typical with an enrichment of organic matter averaging about 8 %. The degree of organic matter can be higher if the volcanic soil develops a low pH and Al-humus complexes start to form (Eriksson, et al., 2005). The permeability and water retention capacity is normally good in this type of soil, since the bulk density (the total mass of the material of the total volume) is normally low (less than 0.9 kg/dm3) (Eriksson, et al., 2005).
2.3. Climatological and Hydrological Data Considered and Data Quality Assessment
The input data, namely precipitation (P) from 1975-2015, discharge (Q) from 1967-2015 and temperature (T) from 1963-2015 used in this study are all primary data generated by the Instituto Nacional de Metrologia e Hidrologia (INAMHI) (Table 2), delivered as monthly averages with a minimum of 20 days of measurements for each month. The discharge station (H0091) used is located before the river Río Grande converges with the river Játiva o Alumbre, and before the water of Río Grande is led through an open channel system to one of the local power plants. The discharge observations at this station are based on daily measurements of the water level, every morning and afternoon, and has been conducted by the same observer for the last 23 years (Trujillo Tupe, 2016).
This data has then been reported to INAMHI who have made the estimates of the discharge (m3/s) in the river Río Grande (INAMHI, 2015).
There are several rain gauges in this northern highland region of Ecuador, installed according to the norms of the World Metrological Organisation (WMO) (Figure 1). The precipitation at these stations has been observed at three specific times per day (07, 13 and 19 h) and reported to INAMHI (INAMHI, 2015). Only one of the rain gauges (M0308) is located in the watershed of Río Grande at the altitude of 3418 m.a.sl. The annual P of this station was compared both graphically and with a Kruskal Wallis statistical test to the average annual P calculated from five stations in the region (Table 2). This was done to assess potential variability due to the highly irregular topography in the region and the potential for the added value of considering the P data available from stations that are located in the neighbour catchments Río Napo and Río Mira. Based on this initial comparison of regional rain gauge data and the uncertainties associated with wind patterns and the distribution of P in the region, only data from the M0308 station were used for further analysis in this study.
For the temperature estimates in the region of the Río Grande watershed, the data from two meteorological stations were used: M0102 (3000 m.a.sl.) and M0103 (2860 m.a.sl.). While these stations are located in the neighbouring watershed of Río Míra, they are considered representative since they are located at more or less the same altitude as the discharge station (H0091) (3120 m.a.sl.). As such, these were used on the assumption that the T in the region does not vary in the same way the P patterns might do.
The input data was organised into hydrological years from dry period to dry period (August to August) and the quality of the data was assessed graphically. This allowed for a better understanding of the data and to identify possible patterns and trends. With the use of boxplots potential outliers was identified. In all the original data series of P, T and Q, some months of data were missing. In the discharge data, some of the gaps were larger. Over the entire period of observation, the years with more than two months of monthly data missing were marked, including the hydrological years 1984-
1986, 1992, 1994 and 2003-2004. The impact of including the years with the remaining gaps (≤ 2 months) in the measured observations of the P, Q and T data was assessed with a non-parametric Wilcoxon Rank Sum test.
Annual actual evapotranspiration (AET) was estimated using the water balance approach (Equation 1), where the change in storage (ΔS) was assumed to be zero for the analysis on an annual scale:
P = Q + (AET + ΔS) (Equation 1)
The potential evapotranspiration PET was estimated using the Langbein (1949) equation (Equation 2). Where PET is the estimated potential evapotranspiration (mm/yr) and T is the annual average temperature.
PET = 325 + 21T + 0.9T2 (Equation 2)
Table 2. Meteorological and hydrological stations measuring P, T and Q in the northern highland region of Ecuador, considered in this study.
Type Station Name Watershed Province Latitude Longitude Altitude (m.a.sl)
Data (years) P M0102 El Angel Río Mira Carchi 0G 37' 8.2" N 77G 56'41.4"
W 3000 1964-
2015 P M0305 Julio
Andrade Río Mira Carchi 0G 39' 23.1"
N
77G 43'13.2"
W 2890 1964-
2015
P M0103 San
Gabriel Río Mira Carchi 0G 36' 15" N 77G 49'10" W 2860 1964- 2015
P M0101 El
Carmelo Río Napo Carchi 0G 41'3" N 77G 36'42" W 2955 1964- 2015 P M0308 Tufiño Río Carchi Carchi 0G 48' 1.1" N 77G 51'19.7"
W 3418 1975-
2015 Q H0091 Grande
Aj Jativa Río Carchi Carchi 0G 48' 15" N 77G 50' 46" 3120 1967- 2015 T M0102 El Angel Río Mira Carchi 0G 37' 8,2" N 77G 56'41,4" 3000 1963- 2015
T M0103 San
Gabriel Río Mira Carchi 0G 36' 15"N 77G 49'10" W 2860 1963- 2015
2.4. Long-term water balance observations
To get a better understanding for the water resource distribution in the Río Grande watershed the water balance components P, Q, AET, PET and T were evaluated over a longer period of time. The long-term general trend of the parameters was analysed using a regression analysis. All trends were tested assuming the data was following a linear trend. As such, the resulting slope for each data series was statistically tested using a t-test against the slope of 0 (the null hypothesis) and a significant level of 5 %.
For the trend analysis for these years, the data series of P and Q were interpolated in the statistical software JMP from SAS and all monthly gaps in the data was filled using a linear
relationship between the two surrounding monthly values around the missing value (Equation 3).
The impact of this interpolation method on the trend analysis was also explored to avoid any spurious correlations.
𝑥𝑛𝑒𝑤 = 𝑥𝑡2𝑡 −𝑥𝑡1
2−𝑡1 × (𝑡𝑛𝑒𝑤− 𝑡1) + 𝑥2 (Equation 3)
xnew:new value
xt1: monthly average at time t1
xt2: monthly average at time t2
t1: month before xnew
t2: month after xnew
In addition to the regression analysis, step change analysis of the P, Q, AET, PET and T data were conducted for the 18-years periods from 1979-1997 and 1997-2015. The break point of 1997 was selected to allow for an even amount of data in both periods and to simplify comparison between the different short-term step change analysis conducted to evaluate the impacts of the PSB. The years with more than two months of data missing in the Q data (1984-1986, 1992, 1994 and 2003- 2004) were excluded in all the step change analyses. Considering the remaining years with missing monthly data and the uncertainties these gaps bring to the interpretation of the results, in these step change analysis, the data series of the hydraulic parameters including the years with gaps were analysed. Additionally, the gaps in the data of the hydraulic parameters were filled using two different strategies; a linear relationship between the two surrounding monthly data values around the gap (Equation 3); and the average of the four surrounding monthly values around the gap. These two data sets with filled gaps were also analysed. Overall, the goal of considering these variations in gap filling vs. not gap filling of the data was to ensure a robust estimate of potential step-shift in either forcing or response components of the water balance parameters.
Monthly averages of the P, Q, AET and T data were calculated and compared between the 18- years periods 1979-1997 and 1997-2015. This was done to analyse whether or not the distribution of the hydrologic parameters had changed over the seasons rather than on an annual scale. For these analysis the data series with gaps filled (Equation 3) was used.
2.5. Potential impacts of the PSB scheme on the hydrological and climatological parameters The PSB was implemented in August 2009 in the páramo ecosystem of the local community La Esperanza. To monitor whether or not the PSB scheme has had an impact on the climatological and hydrological parameters in the watershed, short-term step change analysis using 6-years periods and 12-years periods were conducted. In addition, this analysis tested for the robustness of the results for the time periods considered when assessing change in the system. For these analysis, the years of 1979-2014 were used for the precipitation and evapotranspiration data. The discharge and
temperature data looked further back in time. As such the analysis of these parameters was extended to include the years 1967-2014. The same data series (including gaps and gaps filled) that were used in the 18-years step change analysis were also considered in this analysis. A
complementary data series of the annual estimates of each of the parameters was conducted from the sum of the monthly averages of the P, Q, AET and T.
2.6. Statistical hypothesis testing
The null hypotheses (H0), that the annual estimates of the hydraulic parameters P, Q, AET, PET and annual average T were the same in the different time periods considered in the step change analysis was statistically tested using the JMP software from SAS. The distributions of the data in the different periods were checked using histograms and normal quantile plots. A normal distribution could not be guaranteed for all periods. The non-parametric test Wilcoxon/Mann Whitney for two independent samples was therefore used. In the short-term step change analysis, in the case with more than two periods of data to test the nonparametric Kruskal Wallis test was used. From this test, it is only possible to tell whether or not there is a significant difference between the different populations, and not which populations are different. The multi comparison test Wilcoxon Method for Each Pair was therefore used as a complement to these analyses if the H0 from the Kruskal Wallis test could be rejected.
Non-parametric rank-based tests are often less powerful than a parametric approach, meaning that it is easier to make a Type II error by not rejecting the H0 when it is false. The power of the tests could be increased by increasing the significance level deemed acceptable, but that would in turn increase the chance of making a Type I error of rejecting H0 when it is true (Weiss, 2014). For these tests a significance level of 5 % was used. If a p-value resulted to be ≤ 0.05 in the tests, one of the populations could be concluded, with 95 % confidence, to have a significant higher or lower annual observation.
2.7. Qualitative evaluation of the local people’s observations of the bio-physical conditions and the PSB
To get a better understanding of the physical conditions of the watershed of Río Grande and how they were perceived, qualitative data and information were collected by key informant interviews (Mikkelsen, 2005) with people in the community La Esperanza. The aim with these interviews was also to characterize the local perception about the PSB as a land/water management strategy. The target group of participants were middle age or older with experience in the land use of the
community and how it has changed over time, and with knowledge of the implementation of the PSB scheme. Semi structured interviews were chosen, since this type of interview can be organised into structured, but flexible interviews and gives an advantage for the researcher to identify issues that could be unknown in advance (O'keeffe, et al., 2015; Mikkelsen, 2005). The interviews were completed in August 2016, seven years after the implementation of the PSB.
In total, 13 interviews were completed that lasted between 15-45 minutes (Appendix B). Due to confidentiality and ethical considerations all respondents participating in the study are presented anonymously. The interviews were conducted with a volunteer approach in an undisturbed
arrangement, most of them (Nr. 1-7, 9-10, and 13) in connection to community meetings. The other interviews were conducted with people that were met in the field. A pre-tested interview guide (Appendix C) of questions was used during the interviews to structure the interviews and to be sure that all topics were covered. Interviews started with a presentation of the purpose of the study and by informing and checking if it was allowed to record the interview. The interviews covered a primary section with questions about the area, its natural resources, ecosystem services and disservices and a second section about the conservation program Socio Bosque. The interviews were flexible in their construction and opened to add, skip or change the order of the questions depending on how the interview was progressing and the different themes covered. All interviews were recorded except for one, which took place outside in the páramo. Brief notes were taken during all interviews, but more detailed notes were taken during the interview that was not recorded. These notes were reviewed
and summarized the same day the interview was completed. A Spanish speaking assistant was present during all interviews, who helped to record the interviews and to translate or sort out potential misunderstandings. The interviews were transcribed word for word in Microsoft Word.
Each question and answer from this material was then sorted into larger themes covered in the interview guide (Appendix D). This material was in turn sorted a second time and organised into a matrix with more detailed themes, questions, and positive/negative/neutral responses (Appendix D).
The underlying material representing a summary of the results was translated into English and is presented in this study. During this qualitative material work, it was decided that two of the interviews would not be included in the final summary of the results. The location, where these interviews took place was considered to have a negative effect on the structure of the interviews and thereby on the resulting data. One of these interviews took place during a morning shift where the respondent was milking cows. While potentially biased, this interview setting did provide important information about the land use practises and life in Tufiño.
Additionally, two semi structured interviews were conducted as a follow up to the initial 13 interviews. The first was with Trujillo Tupe (2016), who is the individual that has been recording the water level of the river Río Grande. The other interview was with Santiago Levy, who is working with environmental issues in the province of Carchi. The same interview guide (Appendix C), was used as the underlying structure for these interviews. These interviews are considered to target regional experts with the water flows and environmental health in the watershed respectively.
3. RESULTS
3.1. Assessing Data Quality
In Figure 2 of the annual P data from station M0308 (both data sets including and excluding years with gaps) for the period 1975-2014 it is possible to recognize that there are highs and lows in the precipitation that regularly repeat in the data series. For example, years with more than 1600 mm seem to repeat every 6-10 years and years with less than 888 mm of precipitation every 8-13 year. It also seems that the precipitation can shift drastically from year to year in the area. Between the years 1988 (1606 mm) and 1991 (718 mm), for example, there was a drop of 888 mm, also between the years 1998 (1671 mm) and 2001 (785 mm) there was a drop of 886 mm.
Of the P data that were covering the period Aug 1979 – Jul 2015 (used in the step-change
analysis) the data missed in total 1.6 % of the monthly observations (Table 3). In total, six of the years were missing one month of data spread over the year, and one year was missing two months of data.
The data set of the annual P including the years with gaps was compared to a set of data excluding these years. The statistical hypothesis test of this comparison resulted in that the H0 could not be rejected. As such, no significant difference could be recognised between these data sets on the significance level of 5 % (Table 3). In addition, the distribution of the monthly average P for the period 1979-2014 of the two P datasets, including and excluding years with gaps, did not show any difference over the year (Appendix F). June to September were the months with a lower monthly average P (41 – 71 mm), while October until May, had a higher monthly average P (92-167 mm).
The annual P estimated from station M0308 and the mean annual P estimated from the five stations in the region followed similar patterns over the period 1975-2014 (Figure 3). The mean annual P at the station M0308 (incl. gaps; 1257 mm) was, however, higher than the regional mean annual P (1106 mm) for the period 1975-2014. A significant difference was also found in the non-parametric Kruskal Wallis statistical test of the annual P from the two data sets and on the significant level of 5 % (Table 4). As such, there is a difference between the annual P estimated at the station located in the watershed of Río Grande, compared to the regional mean annual P. The regional mean monthly average P had a slightly lower distribution most notable during October to December and March to May compared to the P from M0308 (Appendix F).
Table 3. Data quality assessment of the P data from station M0308 (Boxplots in Appendix E), using a Wilcoxon Rank Sum test of the annual P (2-Sample Test, Normal Approximation) (α = 0.05).
Period Data Data
missing Min Median Max Mean Std
Dev S Z P-value
Year % Mm mm mm mm mm
1975- 2014
P (excl.
gaps) 1.6 785.4 1217.4 1762.3 1250.4 235.7 P (incl.
gaps) 784.6 1250.4 1762.3 1257.5 242.4 1205 -0.1774 0.8592 1979-
2014
P (excl.
gaps) 1.7 785.4 1202.3 1762.3 1256.4 236.8 P (incl.
gaps) 784.6 1225.0 1762.3 1257.0 244.4 1205 -0.1774 0.8592
Figure 2 of the annual Q (incl. gaps and excl. years with gaps) from 1967-1979 shows that there have been highs and lows in the annual Q that have returned regularly. It is also notable that those years with high flow (1974, 1982, 1988, 1999 and 2014) have decreased over the time. The data used for the regression analysis and step change analysis of the annual Q (1967-2014) missed 12.7 % of all monthly observations (Table 5). In total, 13 years of this data missed one or two months of
observations, and eight years missed more than two months of observations. In total, 20 years with gaps appeared after 1979 and resulted in that the data used for the estimate of the annual AET (1979-2014) for the step change analysis is missing 16.9 % of its data. No significant difference was found between the annual Q calculated from the data including gaps and the data excluding the years with gaps (Table 5). The distribution/patterns of the monthly average Q over the year was similar for the different data sets (with and excluding the gaps) for the period 1966-1979, with a peak monthly average Q in May (111-113 mm), and the lowest monthly average Q in September (39-46 mm) (Appendix F).
Table 5. Data quality assessment of the Q data from station H0091 (Boxplots in Appendix E); using Kruskal- Wallis Tests of the annual Q (1-way Test, ChiSquare Approximation) (α = 0.05).
Period Data Data
missing Min Median Max Mean Std Dev
Chi
Square DF P- value
Year % mm Mm mm mm mm
1967- 2014
Q (excl.
gaps) 12.7 651.0 1045.2 1678.5 1062.2 269.4 Q (excl.
gaps > 2 months)
496.3 995.3 1678.5 1005.0 286.0
Q (incl.
gaps) 99.7 941.0 1678.5 944.4 327.7 2.356 2 0.3079
1979- 2014
Q (excl.
gaps) 16.9 651.0 924.5 214.4 1398.9 946.1 Q (excl.
gaps > 2 months)
496.3 996.2 286.3 1678.5 1012.2
Q (incl.
gaps) 99.7 859.4 295.4 1471.0 856.4 4.7469 2 0.0932
Table 4. Mean annual P at the station M0308 and the mean annual average P in the region, and the results from the comparison of the annual P of these observations using a Kruskal Wallis test (α = 0.05).
DATA PERIOD MEAN
(mm)
STDV.
(mm)
P Region Aug 1975 - Jul 2015 1105.8 203.2
P M0308 (incl. gaps) Aug 1975 - Jul 2015 1257.5 242.4 P M0308 (excl. gaps) Aug 1975 - Jul 2015 1250.4 235.7 Kruskal-Wallis test
p-value
One way, chi-square
approximation 0.0017
Figure 2. Precipitation data from the station M0308 from 1975 - 2014 (A); Annual Q at the station H0091 from 1967 – 2014 including and excluding years with monthly gaps of data missing (B); annual average T (°C) of the two metrological stations
M0103 and M0102 from 1963 – 2014 (C).
Figure 2 shows that the annual average T at both stations (M0103 and M0102) followed similar patterns during the period 1963 – 2014. The T data between 1963-2014 (used for the regression analysis) from the station M0102 missed in total 7.4 % of the monthly observations and 10.3 % of the data for the period 1979-2014 (used for the step-change analysis) (Table 6). The T data from the station M0103 missed 2.4 % of the monthly observations for the longer period, and 0.2 % for the shorter period. The mean annual average T at the station M0103 (12.2 °C) was higher than the mean annual average T at the station M0102 (11.9 °C). A statistically significant difference was also found when the data series of the annual average T were compared (Table 6). The mean monthly average T at the two stations M0103 and M0102 follow similar patterns over the year (Appendix F). The mean monthly average T at both stations vary little over the year with the lowest T in July (11.2 – 11.4 °C) and the highest in November for the station M0103 (12.8 °C) and in October for the station M0102 (12.2 °C) (Appendix F).
0 500 1000 1500 2000
1960 1970 1980 1990 2000 2010 2020
P (mm)
A
P (incl. gaps) P (excl. gaps) Linjär (P (excl. gaps))
0 500 1000 1500 2000
1960 1970 1980 1990 2000 2010 2020
Q (mm)
B
Q (incl. gaps) Q (incl. gaps ≤ 2 months)
Q (gaps excluded) Linjär (Q (gaps excluded))
10 11 12 13 14 15
1960 1970 1980 1990 2000 2010 2020
T(°C)
C
M0103 M0102 Linjär (M0103) Linjär (M0102)
Table 6. Data quality assessment of the T data from station M0102 and M0103 (Boxplots in Appendix E), using a Wilcoxon Rank Sum Test of the annual average T (2-Sample Test, Normal Approximation) (α = 0.05).
Period Data Data
missing Min Median Max Mean Std Dev
Chi
Square DF P-value
Year % ° C ° C ° C ° C ° C
1963- 2014
T M0102 7.4 11.1 11.8 12.7 11.9 0.3
T M0103 2.4 11.6 12.2 13.1 12.2 0.3 1928.5 -4.79 <0.0001 1979-
2014
T M0102 10.3 11.1 11.9 12.7 12.0 0.3
T M0103 0.2 11.6 12.3 13.1 12.3 0.3 1928.5 -4.79 <.0001
3.2. Long term water balance observations
The trend of annual P between the period 1975 – 2014 estimated from station M0308 resulted to be negative (-2.4 mm/year) (Figure 3). The regression analysis also resulted in a negative slope of -3.1 mm/year for the regional mean annual P estimated over the period 1963-2014 (Figure 3). The statistical tests of these slopes resulted in high p-values (>0.05) of 0.5227, and 0.1310 respectively.
The H0 could therefore not be rejected and no significant linear trends of the data series were found.
The mean annual P at the station M0308 (gaps filled) was 1250 ± 213 mm for the 18 years-period Aug 1979 – Jul 1997 and 1305 ± 244 mm for the period Aug 1997 – Jul 2015. In the statistical analysis of the annual P of these two periods no significant difference was neither found in none of the data series (incl. gaps vs. gaps filled) (Appendix G & H). The mean monthly average P had a similar
distribution over the year during the two periods 1979-1997 and 1997-2015 (Figure 5). No significant differences between any of the months were found.
The regression analysis of the annual Q between 1967-2014 resulted in a negative trend of -9.85 mm/yr (Figure 3). The statistical hypothesis test of this slope resulted in a low p-value (<0.05) of 0.000492 and as such the negative trend was significant. The period 1979-2015 (885 ± 201 mm) resulted in a lower mean annual Q compared to the period 1979-1997 (1109 ± 288 mm). The results from the statistical testing of the annual Q of these two 18-years periods varied between the
different data sets (Appendix G & H). For the data series where the gaps had been filled the H0 could be rejected, while not for the data series where the gaps had not been filled. The results from the monthly analysis of the Q (gaps filled) (Figure 5) showed that all months had a lower mean monthly average Q during the period 1997-2015 compared to the period 1979-1997. For the months between August to October the differences of the monthly average Q were significant. In the 12-years analysis of the annual Q, the two later periods 1991-2003 and 2003-2015 had a significantly lower annual Q compared to the periods 1967-1979 and 1979-1991 (Appendix G & J). These results were consistent for all the different tested data series (incl. gaps vs. gaps filled). As such, these results indicated that the annual Q in the Río Grande watershed started to decrease significantly between 1991-2003.
Figure 3. Annual P of the station M0308 from 1975-2014 and the mean annual average P in the region from 1964 - 2014 (A).
Annual Q of the station H0091 from 1967-2014 (B); Estimation of the annual AET from 1975 – 2014 and the annual PET from 1963 - 2014 in the Río Grande watershed (C).
P Region: y = -3.0876x + 1210.1 R² = 0.0459
P M0308: y = -2.3889x + 1353.3 R² = 0.0108
0 500 1000 1500 2000
1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020
BEFORE PSB PSB
P (mm)
A
P Region P M0308 interp. Linjär (P Region) Linjär (P M0308 interp.)
y = -9,8585x + 1316,9 R² = 0,2342 0
500 1000 1500 2000
1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020
BEFORE PSB PSB
Q (mm)
B
Q interp. Linjär (Q interp. )
AET: y = 5.7124x + 121.84 R² = 0.0714 PET: y = 0.5336x + 693.32
R² = 0.3447
-500 0 500 1000 1500
1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020
BEFORE PSB PSB
AET (mm)
C
AET interp. PET Linjär (AET interp.) Linjär (PET)
