Evaluation of Mn concentration provided by soil in citrus-growing regions

https://doi.org/10.1016/j.compag.2009.02.005Get rights and content

Abstract

Manganese (Mn) is an essential nutrient element in citrus growing and Mn deficiency causes some problems related with physiological and morphological structure. Spatial evaluation of Mn obtained from soils in citrus-growing areas is the main objective of this paper. For this purpose, a citrus-growing region in Turkey has been selected and three effective estimation methods: kriging, neural–fuzzy modelling, and fuzzy interval arithmetic have been considered for the spatial evaluations. The model works primarily focus on the model accuracy and smoothing degree of estimations. In addition, error analysis and comparative assessments, which present the advantages and drawbacks of the models, are conducted in the paper. The results and performance evaluations prove the superiorities of soft computing approach in this evaluation.

Introduction

Manganese (Mn) is an essential trace nutrient in all forms of life and it is involved in enzyme activity for photosynthesis, respiration, and nitrogen metabolism. Deficiency in young leaves may show a network of green veins on a light green background similar to an iron deficiency. Intake of this element by the plants is in the form of Mn2+ and it plays crucial roles in oxidation and reduction processes in crops because of being oxidised easily as Mn3+ and Mn4+. In addition to its function in photosynthesis, Mn plays important roles in protein, carbohydrates and lipid metabolisms as well as cell division and extension incidents (Marschner, 1995).

Citrus crops are very sensitive to manganese deficiency which reduces photosynthesis, root growth, sugar production and the disease resistance of citrus trees (Tandon and Roy, 2004). There have been a number of researches carried out with regard to the relationships between Mn deficiency and citrus. One of them, Xiao et al. (2007) found that Mn concentration in old leaves on the spring flush from the “Newhall” (Citrus sinensis Osbeck) navel orange is under the critical concentrations. In addition, Papadakis et al. (2005) determined Mn nutritional status in Washington navel (Citrus sinensis L.) and showed that the average Mn concentration in the leaves of various tree-groups ranged from 6.0 to 8.6 mg kg−1 d.w. They also reported that these concentrations were much lower than the 18 mg kg−1 d.w. which is a suggested critical concentration for Mn deficiency in citrus by Smith (1966).

It is demonstrated that Mn deficiency in orange leaves resulted in decrease in chlorophyll content and an impairment of leaf anatomy (Papadakis et al., 2007a). To solve Mn deficiency in crops, fertilizer should be used. However there are some difficulties in applying external Fe and Mn support from soil for Mn or Fe deficient crops. In this case, foliar application is an effective way of improving Mn and Fe deficiency in crops (Papadakis et al., 2007b). Papadakis et al. (2005) suggested that applying 800 or 1200 mg Mn l−1 from leaves as MnSO4·H2O is successful improving way of Mn deficiency in citrus.

As a result of the former studies, it was showed that Mn is a significant element in citrus growing and Mn deficiency causes some problems related with physiologic and morphologic structure of citrus. Because of these problems, amount of harvest and quality of crops are decreasing. Available means of plant and soil analyses take very long time and bring about considerable costs. Therefore, a reliable and fast method in plant and soil analysis is needed to predict mineral concentrations.

From the methodological view, variation in soil is complex and sampling the soil at a finite number of observations in time yields incomplete pictures (Heuvelink and Webster, 2001). Therefore, the relationship between sampling points must be estimated. These estimations require us to create models of the real world and apply them. Spatial estimations of soil parameters can be carried out using various techniques such as geostatistics (Goovaerts, 2001) and soft computing (McBratney and Odeh, 1997, Bardossy and Fodor, 2004). In the same axis, these methods first use the measured values for model design and then estimate the unknown values between sampling locations with an error.

Mediterranean soils commonly contain high pH, calcareous, and low organic matter (Rashid and Ryan, 2004). These soil properties cause micro-element deficiency in soil and plant. As an important citrus area in Turkey, Adana has Mediterranean soil properties and this area has been selected for the implementation. To realize the study, soil samples 0–30 and 30–60 cm in 37 citrus orchards from Adana-Yüregir district have been collected.

The spatial appraisal of Mn concentrations in the study area is conducted using three effective spatial interpolation methods: kriging, neural–fuzzy modelling, and fuzzy interval arithmetic. The paper mainly focuses on the model accuracy and smoothing degree of estimations. In addition, error analyses and comparative assessments, which describe the superiorities and drawbacks of the models, are presented in the paper.

Section snippets

Statement of problem

The actual value of Mn concentration in soil is never known until it has been measured. Because the measuring process requires labour, time and cost, parameter estimations must be performed using the available data. There are many uncertainties can lead to inaccurate estimates: natural variability, sampling error, estimation error, method selection, etc.

Appraising Mn concentrations in soil is a spatial estimation problem and we handle this using probability and possibility based conventional

Study area and sampling

Soil samples employed in this study were collected by Cakmak et al. (2003) from Cukurova (Adana-Yüregir) region in Turkey to determine Mn concentration in citrus orchards soils. Soils were sampled on November and December in the year 2001. Sampling in soil was realized by collecting surface (0–30) and sub-soil (30–60 cm). Extractable Mn concentration in soils was measured by diethylenetriamine pentaacetate–triethanolamine (DTPA–TEA) method that is suitable for calcareous soils (Lindsay and

Results

Because kriging and neural–fuzzy models produce numerical singletons, the performance comparisons between these methods can be performed together. On the other hand, the interval-based fuzzy model produces a fuzzy number which includes a core interval and therefore it should be evaluated independently of others. Fig. 9, Fig. 10 show the estimation capacities of the kriging and neural–fuzzy model on test data using coefficient of correlation (r), respectively.

In addition to the “r” values,

Discussion

It has been observed from Fig. 9, Fig. 10 that the neural–fuzzy model outperforms more the kriging model. However, high estimation capacities have not been provided for both models (Table 1). This could be resourced from the number of data and natural variability. On the other hand, in addition to relatively good accuracy, the neural–fuzzy model produces more transparent information about the outputs. Particularly, the fuzzy rules identified can be studied and the degree of firing of the rules

Conclusions

We have presented a comparative study for modelling the relation between the spatial positions of soil samples and their Mn concentrations. Appraising Mn concentrations in soil is very important for citrus nutrition. Therefore, various effective modelling tools such as kriging, neuro-fuzzy hybrid modelling and interval-based fuzzy arithmetic have been used for the assessments.

From the applications it has been observed that the neuro-fuzzy hybrid method has produced more accurate outputs than

Acknowledgements

The authors would like to extend their appreciation to the anonymous referees and the editor for their constructive comments.

References (30)

  • C. Carlsson et al.

    Fuzzy Reasoning in Decision Making and Optimization

    (2002)
  • C.V. Deutsch et al.

    GSLIB: Geostatistical Software Library

    (1998)
  • D. Dubois et al.

    Operations on fuzzy numbers

    Int. J. Syst. Sci.

    (1978)
  • M.B. Gorzalczany

    Computational Intelligence Systems and Applications

    (2001)
  • E.H. Isaaks et al.

    An Introduction to Applied Geostatistics

    (1989)
  • Cited by (2)

    • The effects of inarching Citrus reticulata Blanco var. tangerine on the tree vigor, nutrient status and fruit quality of Citrus sinensis Osbeck ‘Newhall’ trees that have Poncirus trifoliata (L.) Raf. as rootstocks

      2019, Scientia Horticulturae
      Citation Excerpt :

      Mn is essential for plant growth and development and acts as a cofactor for enzyme activation in processes such as photosynthesis, lipid biosynthesis and oxidative stress (Socha and Guerinot, 2014; Yasuor et al., 2015). Tutmez et al. revealed that Mn is an essential nutrient element in citrus growth and that Mn deficiency causes problems related to physiological and morphological structure (Tutmez et al., 2009). However, excess amounts of Mn are destructive to chlorophyll and can negatively affect citrus growth (Martinez-Cuenca et al., 2013).

    • Mapping water chemical variables with spatially correlated errors

      2013, Environmental and Ecological Statistics
    View full text