Soil health assessment for coffee farms on andosols in Colombia Fatma Rekik a, ⁎, Harold van Es a , J. Nicolas Hernandez-Aguilera b , Miguel I. Gómez b a Soil and Crop Sciences Section, School of Integrative Plant Science, Cornell University, Bradfield Hall, 306 Tower Rd, Ithaca, NY 14850, United States b Charles H. Dyson School of Applied Economics and Management, Cornell University, Warren Hall, 137 Reservoir Ave, Ithaca, NY 14850, United States abstract article info Article history: Received 24 March 2018 Received in revised form 8 May 2018 Accepted 9 May 2018 Available online xxxx Developing local soil health (SH) benchmarks for different ecosystems is important for supporting locally appro- priate management decisions and correct interpretation of soil health results. This study was conducted to de- velop SH scoring functions as benchmarks specific to coffee production in Cauca, Colombia. A total of 223 soil samples were collected from coffee farms in six municipalities and were analyzed for 13 SH indicators including wet aggregate stability (WAS), available water capacity (AWC), respiration rate, pH, contents of active carbon (AC), organic matter (OM), protein, phosphorus (P), potassium (K), magnesium (Mg), manganese (Mn), iron (Fe) and zinc (Zn). A scoring function for each indicator was developed using the cumulative normal distribution (CND) function with parameters based on either the average local conditions for a given indicator (physical and biological indicators), or thresholds found in the literature for coffee systems (chemical indicators). Separate scoring functions by textural group (fine, medium) were necessary for AWC, OM, AC, and respiration. A best sub- sets regression (BSR) using the overall soil health index as the response variable was executed to determine the indicators with highest predictive power of overall soil health. AC was the best single predictor of soil health, and AC combined with protein, P and pH offer additional predictability, suggesting them for a simplified SH test. © 2018 Elsevier B.V. All rights reserved. Keywords: Mollic Andosols Lithosols Soil health assessment Soil health benchmarks Scoring functions Colombia Coffee Smallholder agriculture 1. Introduction Soil health (SH) is critical to sustainable agricultural production, and its quantitative assessment provides a framework for management. Proper interpretation of SH measurements requires benchmarks to as- sess where a sample lies on the SH spectrum (Arshad and Martin, 2002). The Comprehensive Assessment of Soil Health (CASH) approach developed at Cornell University measures biological, chemical and physical soil properties that are key indicators of SH, and converts labo- ratory and field measurements into generally recognized and easily in- terpretable scores that aid in management decisions (Moebius-Clune et al., 2016). In this framework, scores are derived from functions that were developed following the approach of the Soil Management Assess- ment Framework by Andrews et al. (2004) which assesses a soil indica- tor measurement in relation to a set of empirical values and assigns a normalized score. Similarly, scoring in CASH consists of comparing indi- vidual measured data to a standardized dataset of soils from regions in the United States (Fine et al., 2017). The scoring of each individual SH indicator comes in one of three forms - “more is better”, “optimum range”, and “less is better” - and is adjusted for soil texture when it af- fects the SH indicator. CASH scoring functions for the physical and biological indicators generally follow a cumulative normal distribution (CND) curve specific to each indicator. Others are based on thresholds determined in the lit- erature, which are outcome-based in terms of crop response to different levels of an indicator, as in the case of P, K, pH, and minor elements (Moebius-Clune et al., 2016). All scoring functions are scaled between 0 and 100, and indicator scores are grouped into three ranges: “low” (0–30), “medium” (30–70) and “high” (70–100). From all indicator scores an overall SH index score is calculated as their unweighted arith- metic mean and is interpreted as “very low” (b40), “low” (40–55), “me- dium” (55–70), “high” (70–85), and “very high” (N85; Moebius-Clune et al., 2016). Regional, climatic and soil differences generally have a significant impact on SH and require adjustment of scoring and interpretation frameworks (Congreves et al., 2015). In addition to regions in the USA, Moebius-Clune (2010) developed scoring functions for SH assessment in western Kenya from a chrono-sequence experiment on recently deforested agricultural land. It is important to further “test the Test” in other ecosystems which may ultimately serve as a step towards a widely-standardized SH assessment and interpretation protocol. The scoring functions used in CASH were developed for regions in the USA that are characterized by a temperate climate with diverse production systems including grain, livestock, vineyards and vegetable production, and their use in SH assessment in tropical climates, including Colombian coffee smallholder farms, is not appropriate (Congreves et al., 2015; Idowu et al., 2008; Moebius-Clune, 2010; Schindelbeck et al., 2008). Geoderma Regional 14 (2018) e00176 ⁎ Corresponding author at: 1006 Bradfield Hall, 306 Tower Rd, Ithaca, NY 14850, United States. E-mail addresses: fr235@cornell.edu (F. Rekik), hmv1@cornell.edu (H. van Es), jnh79@cornell.edu (J.N. Hernandez-Aguilera), mig7@cornell.edu (M.I. Gómez). https://doi.org/10.1016/j.geodrs.2018.e00176 2352-0094/© 2018 Elsevier B.V. All rights reserved. Contents lists available at ScienceDirect Geoderma Regional journal homepage: www.elsevier.com/locate/geodrs