Application of the MODIS MOD 17 Net Primary Production product in grassland carrying capacity assessment

https://doi.org/10.1016/j.jag.2018.09.014Get rights and content

Highlights

  • A new, easy to implement, globally applicable approach for grassland carrying capacity assessment.

  • Above ground biomass derived from MODIS NPP accurately predicted grassland biomass in the field.

  • Stocking rates reported for summer pastures in Azerbaijan exceeded predicted carrying capacity.

Abstract

Remote sensing based grassland carrying capacity assessments are not commonly applied in rangeland management. Possible reasons for this include non-equilibrium thinking in rangeland management, and the costliness of existing remotely sensed biomass estimation that carrying capacity assessments require. Here, we present a less demanding approach for grassland biomass estimation using the MODIS Net Primary Production (NPP) product and demonstrate its use in carrying capacity assessment over the mountain grasslands of Azerbaijan. Based on publicly available estimates of the fraction of total NPP partitioned to aboveground NPP (fANPP) we calculate the aboveground biomass produced from 2005 to 2014. Validation of the predicted aboveground biomass with independent field biomass data collected in 2007 and 2008 confirmed the accuracy of the aboveground biomass product and hence we considered it appropriate for further use in the carrying capacity assessment. A first assessment approach, which allowed for consumption of 65% of aboveground biomass, resulted in an average carrying capacity of 12.6 sheep per ha. A second more realistic approach, which further restricted grazing on slopes steeper than 10%, resulted in a stocking density of 6.20 sheep per ha and a carrying capacity of 3.93 million sheep. Our analysis reveals overgrazing of the mountain grasslands because the current livestock population which consists of at least 8 million sheep, 0.5 million goats and an unknown number of cattle exceeds the predicted carrying capacity of 3.93 million sheep. We consider that the geographically explicit advice on sustainable stocking densities is particularly attractive to regulate grazing intensity in geographically varied terrain such as the mountain grasslands of Azerbaijan. We further conclude that the approach, given its generic nature and the free availability of most input data, could be replicated elsewhere. Hence, we advise considering its use where traditional carrying capacity assessments are difficult to implement.

Introduction

Carrying capacity is a concept used to regulate stocking density and avoid overgrazing by livestock (Stoddart et al., 1975). It has been defined as “the density of cattle providing the maximum sustained production of beef” (MacNab, 1985) or more generally “the maximum population of a given species that can be supported indefinitely in a defined habitat without permanently impairing the productivity of that habitat” (Rees, 1996). In grassland management, it relates to the number of livestock per unit area that can be sustained given the amount of available forage and is expressed as the number of animals and days that an area of land may be grazed. It is calculated from estimates of the above ground biomass, the fraction of that biomass that can be grazed sustainably (proper use factor) and the livestock’s daily food requirements.

While simple to define, the implementation of carrying capacity assessments is demanding because the production of forage varies in space and time. Given this, in areas with large year-to-year variation in biomass production, the carrying capacity may need to be adjusted annually. Recommended stocking rates are typically fixed for a whole ranch or an entire country, which may result in under- or overexploitation of available resources in heterogeneous landscapes. In such situations, a geographical differentiation of carrying capacity would allow a better exploitation of the pasture resources (Neudert et al., 2013).

The carrying capacity concept is a Malthusian view (Malthus, 1798) of a grazing system. It assumes a grazing system in which livestock populations are allowed to grow faster than and exceeding equilibrium with the available forage resources, resulting in overexploitation with rangeland degradation as a result (Hardin, 1968). Following theoretical work by DeAngelis and Waterhouse (1987), a paradigm shift occurred towards considering drylands grazing systems as non-equilibrium rather than equilibrium systems (Ellis and Swift, 1988; Behnke and Scoones, 1993; Vetter, 2005). According to this paradigm, drylands grazing systems are non-equilibrium systems, where volatile rainfall and primary production prevents livestock populations overshooting the equilibrium with their forage resources and overstocking and range degradation are unlikely to occur.

As a result, some consider carrying capacity based rangeland management inappropriate for use in drylands grazing systems because of the assumed non-equilibrium nature of these systems (e.g. Scoones, 1995; Niamir-Fuller, 1999). Others however, have suggested that the difficulty to confirm a rangelands’ equilibrium status has resulted in applying the non-equilibrium paradigm to much wider geographies than appropriate (Fernandez-Gimenez and Allen-Diaz, 1999; De Leeuw et al., 2019). Despite this controversy, carrying capacity assessments are nowadays little used in drylands in most parts of the world because of persistent advocacy in favor of the non-equilibrium paradigm. This is a missed opportunity, because carrying capacity assessments remain a conceptually appealing and useful management tool in equilibrium grazing systems.

In such grazing systems, remote sensing could be applied to support carrying capacity assessments with the estimation of biomass that such assessments require. Reeves et al. (2015), who reviewed available remote sensing methods in rangeland management and carrying capacity assessment distinguished empirical models and semi-empirical approaches for estimation of aboveground biomass. The empirical approach was pioneered by Tucker et al. (1983) and Tucker (1985) who predicted biomass through the inversion of statistical models describing the relationship between remote sensing vegetation indices and field based biomass estimates (See also Liu et al., 2017, 2015; Yu et al., 2010). The second approach consists of coupled empirical and process-based models. For example, Hunt and Miyake, (2006) coupled empirical biomass estimates derived from MODIS NDVI to feed into a process-based model for gross (GPP) and net (NPP) primary production to calculate the biomass available to animals. Similarly, the Livestock Early Warning System (LEWS) implemented in Africa (Stuth et al., 2005) and Mongolia (Angerer, 2012) fed biomass estimates derived from MODIS NDVI into the process-based Phytomass Growth Simulation Model to predict available forage.

Obviously, remote sensing has potential to support carrying capacity assessments and it would be ideal if in equilibrium grazing systems the technology could be implemented to support more sustainable livestock and grassland management in dryland ecosystems. Yet, to our knowledge, the above approaches have not been widely adopted and implemented for managing stocking rates and grasslands. Reasons may include the controversy over the non-equilibrium paradigm, the labor intensity and the technical expertise required for remotely sensed carrying capacity assessments and the observation that these methods perform poorly when transferred to other regions (Eisfelder et al., 2012).

Another approach would be to model carrying capacity based on remotely sensed estimates of net primary production (NPP). Eisfelder et al. (2014) for example parametrized the BETHY/DLR model to predict NPP for Kazachstan. While the BETHY/DLR approach predicted NPP for the area for which it was parametrized (i.e. Kazachstan), the MODIS gross (GPP) and net (NPP) primary production products (Running et al., 2004; Zhao et al., 2005) predict primary production for the entire globe. The MODIS MOD17 product calculates GPP, or photosynthesis, based on fAPAR, derived from remote sensing, which is fed into a light use efficiency photosynthesis model (Haxeltyne and Prentice, 1996; Hilker et al., 2008). The algorithm also estimates maintenance and growth respiration of the vegetation, which when subtracted from GPP yields NPP. MOD17 delivers estimates of GPP and NPP on an eight day basis, which are aggregated to monthly and annual time scales. This approach has been used by Reeves et al. (2006) who compared MODIS NPP to field estimates of peak biomass from grasslands in Oklahoma and concluded that MODIS vegetation productivity estimates are suited for regional grassland studies.

Despite the promising results from Reeves et al. (2006), MODIS NPP has never been used to estimate grassland carrying capacity. One of the reasons for this is that the MODIS NPP product measures NPP that is allocated to above and belowground biomass, while an estimate of aboveground biomass is needed for a carrying capacity assessment. A carrying capacity model thus needs to account for the fraction of the total NPP that is partitioned to above (fANPP) and belowground (fBNPP) biomass. Hui and Jackson (2005) analyzed a dataset of above and belowground NPP estimates in grasslands around the world. Their analysis revealed that mean annual temperature (MAT) was the single best predictor of fBNPP in grasslands, while the addition of mean annual precipitation (MAP) did not further improve the results. This knowledge on drivers of grassland fBNPP could be combined with the MODIS NPP product to derive an estimate of aboveground NPP for use in carrying capacity assessments. To the authors’ knowledge, there has been no attempt to assess grassland carrying capacity using this approach.

The objective of this paper is to describe a novel approach to assess grassland carrying capacity based on an aboveground grassland biomass estimate derived from a combination of the MODIS MOD17 Net Primary Production Product and the fraction of NPP allocated to aboveground biomass that was derived from mean annual temperature and demonstrate its application to the mountain grasslands of Azerbaijan.

Section snippets

Modelling aboveground biomass as an input to a carrying capacity assessment

The approach taken to model carrying capacity is presented schematically in Fig. 1. First we predicted above ground biomass (AB) based on the MODIS Net Primary Production (NPP) Product MOD17A3H (Running and Zhao, 2015; Running et al., 1999). The MODIS NPP values, which are expressed in kg C m−2 yr-1, were converted to biomass using a biomass to carbon conversion factor of 0.47 (IPCC et al., 2006). Next, Aboveground Biomass (AB) was derived from NPP, using the fraction of NPP that is allocated

Net Primary Production (NPP)

There is significant variation in average annual NPP across Azerbaijan (Fig. 4). NPP varies from below 200 g C m−2 yr-1 in semi-arid lowlands, 400 to 600 g C in irrigated lands around River Kura to 600–1400 g C in rainfed croplands and forests on the footslopes and slopes of the Greater and Lesser Caucasus. Above 1600 m asl NPP declines with elevation from around 600 g C m−2 yr-1 in the upper sub-alpine zone to below 200 g C m−2 yr-1 in the sub-nival zone.

Forests, which were excluded from

Discussion

We presented a new approach to establish grassland carrying capacity based on the MODIS MOD17 NPP product. This deductive approach depends on an estimate of above ground NPP, which proved remarkably accurate as shown by validation of the ANPP predictions against independent biomass data from the field. The carrying capacity approach is easy to implement and less demanding than the empirical remote sensing approaches that have been used so far. Based on this, we conclude that the deductive

Conclusion

The approach for carrying capacity assessment described here has potential for utilization in grassland areas elsewhere. Overgrazing is a problem worldwide and potentially the approach could be utilized globally. Given its reliance on freely available data the approach is particularly useful for application in areas where data needed for traditional assessment is scarce or difficult to acquire through field sampling. We thus advise to consider using the approach for assessment of carrying

Acknowledgements

The authors declare not conflict of interest. The contributions of JdL, formerly with Baku State University and with the Centre for International Migration and Development (CIM) and AR were supported by CIM and the GIZ IBIS project in the Southern Caucasus and a PhD grant from Baku State University respectively. The collection of biomass data by JE and RN was supported by the research project “Proper utilization of grasslands in Azerbaijan” funded by the Volkswagen Foundation (grant number I/81

References (57)

  • Azstat

    Livestock Numbers 1995-2017. The Statistical Committee of the Republic of Azerbaijan. Baku, Azerbaijan

    (2017)
  • R.H. Behnke et al.

    Rethinking range ecology: implications for rangeland management in Africa

  • A. Bornard et al.

    Agroecological diagnosis of the vegetation on dairy-cow mountain pastures in the French Northern Alps - development and utilization of a simplified typology

    Agronomie

    (1992)
  • P. Borrelli et al.

    An assessment of the global impact of 21st century land use change on soil erosion

    Nat. Commun.

    (2017)
  • Dadashov, V., Undated. Sheep Breeds of Azerbaijan. Accessed 3 July...
  • P.N. De Leeuw et al.

    The Concept of Rangeland Carrying Capacity in Sub-saharan Africa – Myth or Reality

    (1990)
  • J. De Leeuw et al.

    The pastoral farming system: balancing between tradition and transition

  • D.L. DeAngelis et al.

    Equilibrium and non-equilibrium concepts in ecological models

    Ecol. Monogr.

    (1987)
  • C. Eisfelder et al.

    Derivation of biomass information for semi-arid areas using remote-sensing data

    Int. J. Remote Sens.

    (2012)
  • F. Ellis

    Climate Variability and Complex Ecosystem Dynamics: Implications for Pastoral Development. Living With Uncertainty: New Directions in Pastoral Development in Africa. I. Scoones

    (1994)
  • J.E. Ellis et al.

    Stability of African pastoral ecosystems: alternate paradigms and implications for development

    J. Range Manag.

    (1988)
  • J. Etzold et al.

    Monitoring Manual for Summer Pastures in the Greater Caucasus Azerbaijan. Working Paper. Sustainable Management of Biodiversity, South Caucasus

    (2013)
  • M.E. Fernandez-Gimenez et al.

    Testing a non-equilibrium model of rangeland vegetation dynamics in Mongolia

    J. Appl. Ecol.

    (1999)
  • S.E. Fick et al.

    Worldclim 2: new 1-km spatial resolution climate surfaces for global land areas

    Int. J. Climatol.

    (2017)
  • M. George et al.

    Stocking Rate and Carrying Capacity. Module 4: Ranch Operations and Grazing Management. Ecology and Management of Grazing, an Online Course

    (2009)
  • Government of Azerbaijan

    Rules on Allocation and Use of Pastures, Commons and Hayfields

    (2000)
  • G. Hardin

    The tragedy of the commons

    Science

    (1968)
  • A. Haxeltyne et al.

    A general model for light-use efficiency of primary production

    Funct. Ecol.

    (1996)
  • Cited by (42)

    View all citing articles on Scopus
    View full text