Paramos are alpine mountain ecosystems that can be found at high altitudes in the Andean mountain range [1
]. The high annual rainfall, low temperature and unique paramo vegetation, in combination with the Histic Andosols, make them optimal environments for storing and regulating surface and groundwater [2
]. Millions of people are relying on the naturally filtered drinking water coming from these ecosystems, not least the inhabitants of the almost eight-million-people capital city of Colombia, Bogotá, where about 70% of all tap water comes from the nearby located paramo of Chingaza National Park (CNP) (Figure 1
]. Paramo soils in this park have a high hydraulic conductivity, which contributes to a steady streamflow; runoff; and sustained baseflow, which in turn regulates downstream rivers and lakes [2
]. The constant flow of water is used for irrigation, hydropower, drinking water facilities, industries and agriculture, among others. Low temperatures, in combination with the wet soil, limit the organic matter decomposition rate, which in turn benefits the sequestration of carbon [2
Paramos are complex ecosystems under pressure from climate change, land use, and mining. Specifically, regarding greenhouse-gas-emission climate change, the expected intensified warming and changing precipitation intensity and patterns at higher altitudes threaten the thriving of paramos. Current climate change scenarios simulate a likely increase in the long-term average temperature and possible alteration of the precipitation patterns across the paramo regions [5
]. The currently wet and carbon-rich soils could dry up and speed up the decomposition rate, which would convert the paramo soils into sources of CO2
, instead of sinks [5
]. On a large scale, increasing temperatures induce higher evapotranspiration rates from peatbogs (tropical high alpine water bodies) and surface water bodies, lowering water levels and decreasing water storage essential during the dry season [8
]. Prolonged drought periods could be amplified by changing El Niño Southern Oscillation (ENSO) patterns; however, these dynamics are still relatively poorly understood [9
]. Furthermore, the atmospheric stability could shift as a consequence of global warming, diminishing cloud cover at low elevations, reducing occult precipitation, and strengthening evapotranspiration.
In the absence of data for radiation, soil and air moisture and energy fluxes, precipitation and mean and extreme temperatures have been identified as key variables to assess the effects of climate change in these ecosystems [10
]. Temperature and precipitation parameters are well-used indicators when identifying long-term changes in climate, ideally requiring time-series of at least 30 years [10
]. Paramo ecosystems are bound to a specific range of temperature; precipitation; and other forcing variables. When these boundaries shift as a result of climate change, the ecosystem can adapt, move to higher altitudes or disappear [6
], as will happen to many ecosystems worldwide.
If the paramo ecosystem boundaries in CNP are forced to higher altitudes as a result of more extreme precipitation and temperature, then it should be important to establish the range for the new extent of paramos for the sake of water security for Bogotá. Improving knowledge about these environments and related interactions and processes would be an essential step to prevent further deterioration of valuable paramo ecosystems and guarantee water security for megacities such as Bogotá.
This study therefore aims to combine outputs of regional downscaled climate model data with interpolated historical data and actual climatic observations, in order to assess the historical and future hydroclimatic conditions and determine the future extent of paramo ecosystems in CNP. We want to address the following research question: to what extent will change in precipitation and temperature likely affect the suitability of paramo environments? For this purpose, we first calculated mean precipitation and temperature (minimum, mean and maximum) and their ranges at monthly and annual scales in the area of CNP for the historical period (1960–1990) based on available observations. We then validated these observations with a gridded climatic dataset (also used for future climatic predictions) to assess the future values of temperature and precipitation at the location where paramos currently thrive. We finally calculated future extents of the paramo ecosystem in CNP, based on data on maximum and minimum temperatures and precipitation where paramos currently thrive.
1.1. Site Description
CNP (Figure 1
) was declared a National Park in 1977 [14
]; it is located northeast of Colombia’s capital Bogotá, with altitudes ranging between 3000 and 3900 m a.s.l., and a total area of around 770 km² of which approximately 650 km² is paramo (ca. 63% of the total park area). Deep valleys, rough peaks and a varying topography are the result of orogenesis, glacial and volcanic activities, which have been shaping a landscape characterized by abundant lagoons, lakes and rivers. In CNP, there are two main seasons: wet (April to November) and dry (December to March). During the wet season, all water bodies are recharged, allowing both paramo vegetation and soil to store large amounts of water. This water is essential for the ecosystem during the dry season. The mean annual precipitation is around 2000–3000 mm and mean annual temperature is around 11 °C [15
]. At these high altitudes and with the constant input of rain, the evapotranspiration should be potentially higher; however, the dense cloud cover, fog and low leaf area index prevent water from evaporating [9
]. Microclimatological processes create a varying climate across CNP, mainly depending on topography, wind direction, slope and aspect. The ENSO, for example, controls the input of precipitation in the region and varies according to the inter-annual or decadal cycles of El Niño [9
1.2. Data and Methodology
Staff at the research facilities in the park provided raw data for monthly mean temperature (T) and precipitation (P) during the period 1968–2015. We first used these data to calculate the mean, minimum and maximum monthly temperature (Tm
, respectively) and total monthly precipitation (P) from all monthly data available during the period (Figure 1
). In total, 15 precipitation stations and two temperature stations were included in the study. However, stations with data gaps longer than one year were disqualified from the interpolation. Years with missing data were interpolated by using the average of the previous and following year.
Downscaled and interpolated data for Tmin
and P, covering CNP during the period 1960–1990, were retrieved from the WorldClim 1.4 database [17
]. Using the ranges of these three variables, with such a fine spatial resolution (1 × 1 km), can help us understand changes between past and future climatic changes in CNP and some of the implications for paramos. The developers of the WorldClim have conducted a throughout quality control of the dataset, including the correction for topographical variations [18
]. In order to determine the ranges of Tmin
and P, where paramos in CNP currently thrive, we used the paramo extent shapefile from the Sistema de Información Ambiental de Colombia and extracted for each month of the year (n
= 12) the Tmin
and P data for each grid cell of CNP with paramo cover. We determined the distributions for Tmin
and P for each season from all the paramo pixels within CNP and calculated their range, that is, the values of the pixels containing the minimum and maximum values of each. The range represents the overall range of Tmin
and P in which paramo can thrive at each particular location in CNP (grid cell).
Two RCP scenarios were chosen to simulate future climatic conditions for CNP paramo regions for the period 2041–2060. The RCP 4.5 and RCP 8.5 projected data for Tmax
and P were derived from the WorldClim database for each grid cell, as done for the period 1960–1990. In RCP 4.5, the greenhouse gas emissions peak around 2040 before declining, while in RCP 8.5, the emissions continue rising over time [19
]. The simulated data is a part of the Climate Model Intercomparison Project Phase 5 (CMIP5) [10
]. When simulating future climate conditions, a baseline with observed data is required. The interpolated historical WorldClim dataset (1960 to 1990) used in this study was chosen as a baseline for running the Global Climate Model (GCM) that generated the simulated climate conditions. WorldClim has around 20 GCMs with future simulations for monthly Tmin
and P. One of these GCMs, the Community Climate System Model Version 4 (CCSM4), developed by the National Centre for Atmospheric Research (NCAR), was chosen for this study due to its capacity to simulate precipitation and temperature in Colombia [20
]. CCSM4 is a coupled climate model assembled by five diverse models, simulating the Earth’s land, atmosphere, ocean, sea-ice and land-ice. It also has a “coupler” that combines the different models by transitioning information between them [21
]. The estimated amount of atmospheric greenhouse gas concentration influences the outcome of the simulated future climate.
For each grid cell with paramo within CNP, we assumed that paramos could not thrive where the mean of Tm
or P calculated from all the grid cell paramo values within CNP during the period 2041–2060 exceeded the ranges of Tmin
or P during the period 1960–1990 in each particular cell. We performed the calculations for both the wet (April–November) and dry (December–March) seasons. We developed a binary raster dataset in both ArcGIS (GIS software 2015, Environmental Systems Research Institute Inc., Redlands, CA, USA) [22
] and QGIS (Quantum Geographic Information System, Open Source Geospatial Foundation) [23
] for such an assumption, where the specific raster grid cell was assigned a zero for unsuitable areas for paramo, or else a one for suitable areas. Finally, we compared the paramo extent based on the assumption resulting from WorldClim against the current paramo extension (delimited by Sistema de Información Ambiental de Colombia (SIAC) [24
]). For all data, parameters, format, resolution and sources that were used in the processing, see Table S2