**Satellite-Based Monitoring of Ecosystem Functioning in Protected Areas: Recent Trends in the Oak Forests (***Quercus pyrenaica* **Willd.) of Sierra Nevada (Spain)**

M.A. Dionisio, D. Alcaraz-Segura and J. Cabello *Andalusian Center for the Assessment and Monitoring Global Change (CAESCG) Dept. Plant Biology and Ecology, University of Almería Spain* 

#### **1. Introduction**

The implementation of monitoring and early warning programs on the ecological status of natural areas is increasingly recognized as an environmental priority (Lovett et al., 2007). However, the development of such programs faces important challenges derived from the many requirements that ecological indicators should fulfill to achieve effective monitoring and alert systems (Oyonarte et al., 2010). Nowadays, ecosystem functioning characterization has become crucial for the monitoring and management of ecosystems due to several reasons (Cabello et al., 2008). First, the evaluation of functional features of ecosystems, such as the carbon gains dynamics, complements the traditional description of ecosystems based solely on vegetation structural features (like physiognomy, dominant species, or floristic composition) derived from few plot observations (Mueller-Dombois & Ellenberg, 1974; Stephenson, 1990; Alcaraz-Segura et al., 2009a). Second, ecosystem functional attributes show a much quicker response to environmental changes than structural ones (Milchunas & Lauenroth, 1995; Wiegand et al., 2004; Alcaraz-Segura et al., 2008a). Third, functional traits are related to key ecological processes that provide a direct measurement of key ecosystem services (Oyonarte et al., 2010; Paruelo et al., 2011; Volante et al., In press). Finally, remote sensing tools can be used to monitor ecosystem functional attributes over extensive areas, in different regions, and with a fast-revisiting frequency (Paruelo et al., 2005; Pettorelli et al., 2005; Baldi et al., 2008; Cabello et al., 2008; Alcaraz-Segura et al., 2009a). The use of satellitederived information allows for tracking the integrity of key ecological processes and their spatial and temporal variability with the advantage of using common protocols throughout the Earth (Dale & Beyeler, 2001). In this sense, several works have shown the ability of timeseries of satellite images to assess the existence of long-term ecosystem functional changes both at the regional (Baldi et al., 2008; Alcaraz-Segura et al., 2010b) and local (Alcaraz-Segura et al., 2008a; Alcaraz-Segura et al., 2008b; Alcaraz-Segura et al., 2009b; Cabello et al., Accepted) scales.

Satellite-Based Monitoring of Ecosystem Functioning in Protected Areas:

**2.1 The Pyrenean oak forests of Sierra Nevada National Park** 

ecological and evolutionary factors determining its high biodiversity.

**2.2 Monitoring forest ecological status with EVI** 

(García & Mejías, 2009).

cost but effective information.

**2. Methodology** 

al., 1997).

Recent Trends in the Oak Forests (*Quercus pyrenaica* Willd.) of Sierra Nevada (Spain) 357

peripheral populations in Sierra Nevada National Park (Molero et al., 1992; Bonet et al., 2010). *Quercus pyrenaica* is a winter semi-deciduous tree with high water demand during the summer. Hence, the predicted lengthening of the summer dry period associated to a reduction in the annual precipitation and the increase in the mean annual temperatures (Bonet et al., 2010) could impose a serious challenge for the regeneration of these forests (Molero et al., 1992; Blanca & Mendoza, 2000). Unfortunately, compared to the wide availability of studies of forest ecology in Europe, there is an enormous lack of knowledge of the conservation status and ecology of Pyrenean oak woodlands in the Iberian Peninsula

Our objective in this study was to use a satellite-based approach to monitor changes in ecosystem functional attributes of the oak forests of the Sierra Nevada National Park (Figure 1). This approach is based on the characterization of the seasonal dynamics and the interannual variability and trends of the Enhanced Vegetation Index (EVI). From the mean annual curve of EVI of each forest patch, we derived functional attributes related to primary production, seasonality, and phenology of the forests. Finally, by contrasting the baseline conditions of each forest patch with the long-term observed trends for the period 2001-2009, we identified processes of functional changes happening in these forests that could guide management actions. We propose this satellite approach as a near-real-time tool to provide managers with ecologically meaningful assessments of the ecosystem status based on low-

Sierra Nevada National Park is located in the southeast of the Iberian Peninsula (Figure 1). This National Park protects the best samples of high and medium Mediterranean mountainous ecosystems (MMARM, 2004). This park is a hot spot for plant species richness (Blanca et al., 1998; Blanca, 2001) and invertebrate biodiversity. Its altitude (several summits over 3000 m.a.s.l.), its proximity to Africa, and steep altitudinal gradient constitute the main

The Pyrenean oak forests (Figure 1) of Sierra Nevada represent a conservation priority for the Park managers. There are nine locations distributed on siliceous soils both in the northwestern and southern slopes of the mountain range. In general, they are associated to

Our monitoring approach was based on the characterization of ecosystem functional attributes derived from the seasonal dynamics of the Enhanced Vegetation Index (EVI). The EVI calculates the normalized difference in reflectance between the red light that is absorbed in photosynthesis and the strong reflection of near infra-red light caused by the cell structure of the leaves. It also includes a third wavelength (blue) that is used to correct the influence of the atmosphere and the soil. EVI is defined according to equation 2 (Huete et

1 2

(2)

*NIR C R C B L* 

major river valleys and within an altitudinal range of 1200 to 1900 m.a.s.l. (Table 1).

*NIR R EVI G*

Remote sensing tools can be used to detect both evident functional changes produced by land-use transformations (Volante et al., In press), and other subtle and less noticeable changes including insect outbreaks (Kharuk et al., 2003), wind (Yuan et al., 2002), droughts (Tucker & Choudhury, 1987) or floods (Sanyal & Lu, 2004), fires (Riano et al., 2002), pollution (Chu et al., 2003), etc. These impacts may derive in significant changes in key ecological processes, for instance, carbon balance, microclimate, and biodiversity patterns (Turner, 2005; Lovett et al., 2006; Perry & Millington, 2008). Remote sensing has been proved to be useful for monitoring this kind of "within-state" changes (Vogelmann et al., 2009). In particular, satellite-derived spectral vegetation indices, such as the Enhanced Vegetation Index (EVI) and the Normalized Difference Vegetation Index (NDVI), are considered the most useful approach to monitor ecosystem responses to environmental changes (Pettorelli et al., 2005). Vegetation indices constitute the most feasible approach to estimate primary production at the regional scale (Paruelo et al., 1997) since they show a linear response to the intercepted fraction of photosynthetically active radiation (FPAR) (Hanan et al., 1995), which represents the conceptual basis to relate vegetation indices with net primary production (NPP) through Monteith's model (Monteith, 1972) (equation 1).

#### NPP = PAR \* FPAR \* RUE (1)

Where NPP is the Net Primary Production, PAR is the amount of incident Photosynthetically Active Radiation, FPAR is the fraction of that PAR that is intercepted by vegetation green tissues, and RUE is the Radiation Use Efficiency that plants have to transform that radiation into organic carbon compounds. Given this direct relationship with NPP, the most integrative descriptor of ecosystem functioning (McNaughton et al., 1989; Virginia & Wall, 2001), vegetation indices are frequently used to derive indicators of ecosystem functioning such as the annual amount of carbon absorbed by vegetation, or the seasonality and phenology of the carbon gain dynamics (Pettorelli et al., 2005; Alcaraz-Segura et al., 2006).

To evaluate the usefulness of satellite-derived vegetation indices for monitoring functional changes within protected areas, we focused on the Sub-Mediterranean Pyrenean oak forests (*Quercus pyrenaica* Willd.) of the Sierra Nevada National Park (Spain). These forests are considered as a Natural Habitat of Community Interest (*Quercus pyrenaica* oak woods and *Quercus robur* and *Quercus pyrenaica* oak woods from Iberian northwestern, Directive 92/43/CEE) (García & Mejías, 2009). The Pyrenean oak forests are a *quasi*-endemic habitat of the Iberian Peninsula. The only non-Iberian representations are in the Central West of France and in the Rif Mountains of northern Morocco. In the South of Spain, the Pyrenean oak is considered as a vulnerable species (Blanca & Mendoza, 2000). Sierra Nevada oak populations are considered of great biogeographical importance since they constitute the southernmost Iberian representation of these forests (Molero et al., 1992) and they are considered relict deciduous forests in the Southern Mediterranean region (Blanca & Mendoza, 2000; Blanca, 2001). Several stands of these forests in the Sierra Nevada National Park have an unfavorable conservation status (Molero et al., 1992; Bonet et al., 2010). Multiple global change drivers have an impact on these southernmost woodlands of *Quercus pyrenaica* in the Iberian Peninsula. Historically, these populations have been subjected to intense human disturbances (logging, fires, grazing, agriculture, etc). As a result, these forests are highly fragmented and display low ecological maturity (García & Mejías, 2009) that threatens their long-term conservation. Currently, trends towards temperature rises and precipitation decreases have been hypothesized as the main constraining factor reducing peripheral populations in Sierra Nevada National Park (Molero et al., 1992; Bonet et al., 2010). *Quercus pyrenaica* is a winter semi-deciduous tree with high water demand during the summer. Hence, the predicted lengthening of the summer dry period associated to a reduction in the annual precipitation and the increase in the mean annual temperatures (Bonet et al., 2010) could impose a serious challenge for the regeneration of these forests (Molero et al., 1992; Blanca & Mendoza, 2000). Unfortunately, compared to the wide availability of studies of forest ecology in Europe, there is an enormous lack of knowledge of the conservation status and ecology of Pyrenean oak woodlands in the Iberian Peninsula (García & Mejías, 2009).

Our objective in this study was to use a satellite-based approach to monitor changes in ecosystem functional attributes of the oak forests of the Sierra Nevada National Park (Figure 1). This approach is based on the characterization of the seasonal dynamics and the interannual variability and trends of the Enhanced Vegetation Index (EVI). From the mean annual curve of EVI of each forest patch, we derived functional attributes related to primary production, seasonality, and phenology of the forests. Finally, by contrasting the baseline conditions of each forest patch with the long-term observed trends for the period 2001-2009, we identified processes of functional changes happening in these forests that could guide management actions. We propose this satellite approach as a near-real-time tool to provide managers with ecologically meaningful assessments of the ecosystem status based on lowcost but effective information.
