Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Volume estimation models for avocado fruit

  • Mulugeta Mokria ,

    Roles Conceptualization, Data curation, Formal analysis, Writing – original draft, Writing – review & editing,

    Affiliation World Agroforestry (ICRAF), C/O ILRI Campus, Gurd Shola, Addis Ababa, Ethiopia

  • Aster Gebrekirstos,

    Roles Conceptualization, Funding acquisition, Writing – review & editing

    Affiliation World Agroforestry (ICRAF), United Nations Avenue, Nairobi, Kenya

  • Hadia Said,

    Roles Data curation, Writing – review & editing

    Affiliation World Agroforestry (ICRAF), C/O ILRI Campus, Gurd Shola, Addis Ababa, Ethiopia

  • Kiros Hadgu,

    Roles Conceptualization, Writing – review & editing

    Affiliation World Agroforestry (ICRAF), C/O ILRI Campus, Gurd Shola, Addis Ababa, Ethiopia

  • Niguse Hagazi,

    Roles Conceptualization, Data curation, Writing – review & editing

    Affiliation World Agroforestry (ICRAF), C/O ILRI Campus, Gurd Shola, Addis Ababa, Ethiopia

  • Workneh Dubale,

    Roles Data curation

    Affiliation International Livestock Research Institute (ILRI), Addis Ababa, Ethiopia

  • Achim Bräuning

    Roles Conceptualization, Supervision, Writing – review & editing

    Affiliation Institute of Geography, Friedrich-Alexander-University Erlangen-Nuremberg, Erlangen, Germany


Avocado (Persea americana Mill.) is an important horticultural crop and proved to be a very profitable commercial crop for both local consumption and export. The physical characteristics of fruits are an important factor to determine the quality of fruit produced. On the other hand, estimation of fruit volume is time-consuming and impractical under field conditions. Thus, this study was conducted to devise cultivar-specific and generalized allometric models to analytically and non-destructively determine avocado fruit volume of five wildly distributed avocado cultivars. A significant relationship (P ≤ 0.01) was found between fruit diameter, length, and volume of each cultivar. Our best models (VM2 –for cultivar specific, and VM7-generalized model) has passed all the rigorous cross-validation and performance statistics tests and explained 94%, 92%, 87%, 93%, 94% and 93% of the variations in fruit volume of Ettinger, Fuerte, Hass, Nabal, Reed, and Multiple cultivars, respectively. Our finding revealed that in situations where measurements of volume would be inconvenient, or time-consuming, a reliable volume and yield estimation can be obtained using site- and cultivar-specific allometric equations. Allometric models could also play a significant role in improving data availability on avocado fruit physical appearance which is critical to assess the quality and taste of fresh products influencing the purchase decision of customers. Moreover, such information can also be used as a ripeness index to predict optimum harvest time important for planned marketing. More importantly, the models might assist horticulturists, agronomists, and physiologists to conduct further study on avocado production and productivity through agroforestry landuse system across Ethiopia.


Avocado is a highly variable species and classified into three ecological races (i.e., the West Indian (WI), Guatemalan (G) and Mexican (M) "races") [1, 2]. It is an evergreen tree species and the most economically important species of the Lauraceae family. It is grown commercially in America, Africa, Europe, Asia and Oceania. In 2019, the estimated world’s total avocado production was about 7.2 million tonnes from 726, 660 hectares [3]. The major avocado-growing countries are Mexico, USA, Colombia, Indonesia, Chile, the Dominican Republic, Kenya and South Africa [3]. In Africa, Kenya and South Africa are leading in the production and export of avocado to the global market [4]. In Ethiopia, it was first introduced around 1938 in the eastern and southern parts of the country and it is now being widely distributed throughout the country, and mainly used for household consumption and local market [57]. Currently, Ethiopia is one of the top five avocado-producing countries in sub-Saharan Africa (SSA) and 20th in the world [3].

Avocado is recognized as a source of energy and vitamins. It also provides specific non-nutritive physiological benefits that may enhance health [8], thus, it can be considered as a "functional food" [9]. It is one of the top important commercial crops to be traded at a global scale [10, 11], and becoming one of the most promising fruit crops for both food and nutrition security and earning a considerable amount of financial return from export and domestic market [1214]. Due to government initiatives in promoting investment in horticulture sector as well as combating climate change through land diversification and agroforestry practices, the plantation and the production of avocado is considerably increasing over the last few years across different parts of Ethiopia. Despite the expansion of avocado tree plantations, the physical characteristic of avocado fruits and productivity of small-scale avocado farming are not well studied in Ethiopia. On the other hand, information on fruit size is critical factor to determine the quality of the avocados and has been used to describe the fruit’s growth curve, predict yield, and conduct physiological studies [15]. More importantly, on-tree and non-destructive volume estimates can be used as a ripeness index to predict maturity and optimum harvest time, and can be used while deciding packing material (tray insert) purchasing and marketing arrangements [16]. For physiological studies, measurement of the size of individual fruit over time allows monitoring fruit expansion rate and its response to physiological disorders and agronomic conditions [16, 17]. Moreover, in the context of postharvest operations, fruit size determination is important for several reasons, such as to determine packing material, fruit classified into batches of uniform size, assign market and price differentials of large and small produce, to match consumer preferences [18]. Thus, the availability of reliable fruit size information is critical in horticultural crop processing and marketing. Fruit volume is also a good measure of size, but direct measurement of fruit volume using water displacement approach is time-consuming as well as impractical under field conditions. On the other hand, length and width measurements of avocado fruit are quick and easy in the field or indoors and can be used to numerically represent fruit volume and weight. Therefore, this study was conducted to: (a) determine fruit volume of five avocado cultivars, (b) determine which biometric parameter of the avocado fruit best correlates with volume; (c) derive various cultivar-specific and mixed-cultivar allometric equations to predict fruit volume and (d) to evaluate the predictive performances of the equations and to identify the best allometric equation for the study region.

Materials and methods

Study area and climate characteristics

The study was conducted in the Upper Gana (7° 34’ 24” N, 37° 46’ 4”E) and Jewe (7° 30’ 35” N, 37° 47’ 1”) Kebeles of Limu district, situated in Hadiya zone, in the Ethiopian Southern Nations, Nationalities, and Peoples’ Region (SNNPR) [19] (Fig 1). The study area is located at 223 km South of Addis Ababa. The altitudinal ranges of Jewe and upper-Gana Kebles were between 2000 and 2400m (Fig 1). Based on the data from the Central Statistical Agency of Ethiopia (CSA), the district has an estimated total population of 153,783 and 93% of the people are living in the rural areas and practicing subsistence farming depending on rainfed production system [20]. The average farm size per family head was estimated to be 0.5 ha [19].

Fig 1. The location of the study area in Hadiya zone, Ethiopia.

The shape files were accessed from open data sources of the open AFRICA, (Open access).

In the study area, the annual rainfall ranges between 1300 and 1400 mm with a bi-modal rainfall seasonality, occurring from February to April and from June to September [19]. The average annual minimum and maximum temperatures were 18°C and 23°C [19]. The typical land use system of the district is characterized by Agroforestry (i.e., mixed crop-tree-livestock production) [21]. The district has a favourable climate and agroecological condition for multi-strata agroforestry and home garden intensive farming system.

Tree selection for sample fruit collection

In this study, five different avocado cultivars including Ettinger (race—Guatemalan (G) X Mexican (M) hybrid), Fuerte (race–G X M), Hass (race—G X M), Nabal (race—G), and Reed (race—G) were considered (Fig 2, S1 File) [22, 23]. From each cultivar, 30 avocado trees, representing 30 small-scale farmlands, were selected for fruit sample collection. Fruit samples were collected considering each radii of the crown [2427]. A total of 360 fruit samples were randomly harvested from each cultivar (i.e., three fruits from each radius or 12 sample fruits per tree = 12 fruit X 30 trees = 360 fruits). Then, fruit length (FL in mm), diameter (FD in mm) and weight (FW in g) were measured using digital calliper and weighing scale, respectively. For fruit volume measurement, the collected sample fruits were first classified into three sizes (small, medium, and large). Then, 15 fruits were randomly selected from each size class (total = 45 fruits from each cultivar). For each fruit, actual fruit volume (AFV) was measured using water displacement method [15, 28], in the Biotechnology Laboratory of Wachemo University (WCU), Hossana, Ethiopia.

Fig 2. Photographic documentation of the studied avocado cultivars during fruit sample collection.

Model development, performance evaluation and cross-validation test

Cultivar-specific and mixed-cultivar generalized avocado fruit volume (FV) estimation models were developed using linear and non-linear regression equations based on either fruit diameter, fruit length alone or both fruit length and diameter at the same time as independent variables. Moreover, using two predictors (i.e., length and diameter) may introduce potential problems of co-linearity, resulting in poor precision in the estimates of the corresponding regression coefficients [29]. Thus, we conducted a multicollinearity analysis using Variance Inflation Factors (VIF = 1/(1-r2) and the Tolerance Values (T = 1/VIF) [30]; where r is the correlation coefficient between length and diameter of fruit. VIF value exceeding 10 or if T value was smaller than 0.10 then co-linearity may have a considerable impact on the prediction of the parameters, and consequently, one of those should be excluded from the model [31, 32].

Model performance was checked using various goodness-of-fit statistics, such as the Coefficient of Determination (R2), Standard Error of Estimate (SEE), Index of Agreement (D), Mean Absolute Bias (MAB), Percent Bias (PBIAS), Root Mean Square Error (RMSE), Prediction Residuals Sum of Squares (PRESS), Reduction of Error (RE), and Coefficient Efficiency (CE), [3335]. The estimation models with higher R2 may sometimes also have unstable parameters estimates. Thus, we further calculated Percent Relative Standard Error (PRSE) of the coefficients and Weighted Akaike information criterion (AICiw) to check the stability of model parameter estimates [29]. Outlier and influential diagnostic test statistics, including Cook’s distance, Leverage point, Studentized Residuals and DFFITS were analysed to examine the accuracy of model fit [29, 36]. Finally, models were evaluated and ranked based on all goodness-of-fit statistics, outlier, and influential diagnostic statistics [35].

To validate the best fitting equation for volume estimation, model cross-validation was conducted following a split-sample approach in which 45 measured fruit sample were partitioned into two sets, 33 for ‘‘training” (i.e., to develop the equations) and the remaining 12 fruit samples for ‘‘testing” the equations. The partitioning was performed according to the following procedure. From each size class (small, medium, and large) of the samples, four samples were randomly selected to perform the ‘‘test” dataset and the remaining sample fruits were used to form the ‘‘training” dataset. The goodness-of-fit statistics and equation coefficients of the ‘‘training” equations were compared with those derived using the full dataset, and the estimated and measured volume of ‘‘test” sample fruit was compared [35, 37, 38]. Finally, the full dataset was used to build fruit volume estimation models.

Statistical analysis

Pearson correlation tests were conducted between actual AFV-FD and FL to be able to identify which fruit biometric variables were most strongly correlated with FV. A correlation analysis was also conducted between independent variables (i.e., FL vs FD). The differences among avocado cultivars in AFV was assessed using one-way analysis of variance and the significance of differences were tested using the least significant difference test (LSD) with P < 0.05 [35].

Results and discussion

The relationship of physical characteristics of avocado fruits

The physical characteristics of sample avocado fruits and their statistical attributes are presented in Table 1. The AFV ranged from 125–480 cm3, FW ranged from 129 – 595g, FL ranged from 64.5–129.9 mm, and FD ranged from 53.8–99.8 mm (Table 1). These findings are in line with other reports [2, 39, 40]. Information on fruit physical appearance is important for the customer who is used to assess the quality and taste of fresh product influencing the purchase decision [16].

Table 1. Summary of volume, and fruit size characteristics of five avocado cultivars.

The relationship between FV and other pomological traits (i.e., FL, FD,) were significant (P < 0.01) (Fig 3) and that is in agreement with other studies that have shown strong relationships between different pomological traits of avocado fruit [2]. Moreover, the correlation between FD and FV were consistently stronger compared to the relationship between FL and FV (Fig 3), indicating the FD is a very good predictor of fruit volume and highlighting those changes in diameter, therefore, affects the fruit volume more than does a change in fruit length. More importantly, knowing the relationship between fruit physical characteristics is critical in the horticultural sector because these pomological traits are sometime used as fruit maturity index [2, 41, 42]. Several indices have been used to determine avocado fruit maturity, hence there is no single factor can be considered the most important; however, it can be stated that from a postharvest standpoint, quality begins at harvest with physiological maturity. This indicates that understanding the stage of physiological maturity is critical for the development of a successful avocado fresh fruit industry while assuring the quality to the consumer [43].

Fig 3. Fruit volume as a function of fruit length, diameter, and regression of fruit diameter as a function of fruit length.

On Fig 3, the letter E, F, H, N, R and M refers Ettinger, Fuerte, Hass, Nabal, Reed and Mixed cultivar data.

Allometric equations and model cross-validation

As a preliminary step to model calibration, the degree of collinearity among fruit length and diameter was analysed. The VIF was ranged from 1.1 to 4.6 and T values ranged from 0.21 to 0.95, depending on cultivar type, respectively (S1 Table). Hence, for all selected genotypes, VIF was < 10 and T was > 0.10, showing that the co-linearity between predictors (fruit length and diameter) is negligible, thus both predictors (FL, FD) were considered during model formation [31, 32] (S1 Table).

Cultivar-specific and mixed cultivar generalized volume model was developed to approximate the shape of avocado fruits using either fruit length (FL) or width (FD) separately and using both predictors at a time. We found that the predictive performance of tested model form were varied within cultivar (Table 2, S1 Table). This might be attributed to differences in the equation forms and predictors included in the models [35]. Allometric model performance analysis and cross-validation test results showed that the Linear Regression Model (MV2) which includes only FD as predictor was ranked the best model given the set of nine (9) candidate model forms for all cultivar-specific models, while the Multiple Linear Regression Model (VM7) was the best for generalized mixed cultivar model (Table 2), and detailed model performance statistics for all tested model forms are presented in S1 Table).

Table 2. Equations and goodness-of-fit performance statistics for estimating avocado fruit volume of five different cultivars and multiple cultivars grown in Limo district, Hadiya, zone.

Our best model explained 94%, 92%, 87%, 93%, 94% and 93% of variation in fruit volume in Ettinger, Fuerte, Hass, Nabal, Reed and mixed-avocado cultivar model, respectively (Fig 4, Table 2, in S1 Table for more details). The three best-performing models for each avocado cultivar and mixed-cultivar are shown in Table 2, where the influence of coefficients was significant (P < 0.01) (Table 2). Our best model (VM2 for cultivar specific) and VM7 –for mixed cultivar models have passed all the rigorous verification and cross-validation statistical test and produced the lowest average relative error (PBIAS%), implying that fruit diameter is reliable predictors of cultivar specific fruit volume, while using both FL and FD might increase the predictive performances of generalized allometric models (S1 Table). Moreover, the performance of our best models (VM2, and VM7) to make an accurate prediction is not an artifact of overfitting, because the parameter values were stable across the subset of the cross-validations “training data set” and full data set (S1 Table). Moreover, the volume of the twelve ‘‘test” fruit volume estimated with the cross-validation equations (i.e., training dataset) differed little from the values estimated with equations produced with the full dataset (Fig 5, S1 Table). The deviations (PBIAS%) in the volume estimates between the two sets of equations were less than 1% for all and mixed cultivar equations (Fig 5). Moreover, the PRSE value of < 20% and outliers and influential points of less than 10% of the total observation as well as higher positive value of CE, and RE provides evidence that the parameter estimates were reliable in the selected best models [29, 4446] (S1 Table). Thus, VM2 (i.e., cultivar-specific model) and VM7 (i.e., generalized model) are reliable to determine the fruit volume based on their easily measurable fruit pomological traits (i.e., FL or/and FD). Furthermore, measuring fruit length and diameter are easy in the field, thus site-and cultivar-specific allometric model would enable researchers to make non-destructive or repeated measurements on the same fruits. Our allometric models could provide accurate estimate of avocado fruit volume and may reduce required time and financial resources while using common method of volume measurements like water displacement, gas displacement and expensive instruments, e.g., image processing software or machine vision techniques. In line with this, review literature showed that there are different non-destructive volume estimation models developed from easily measurable parameters for different type of fruits such as pepper [15, 17], Babassu (Attalea speciosa) fruit [47], Karanda (Carissa carandas) fruit [31], and Apple fruit [48]. Our findings confirmed that non-destructive allometric models based on easily measurable morphometric dimensions can be more accurate and practical in field conditions than destructive methods used in traditional growth curves [18].

Fig 4.

Relationships between FV and FD (left panel) and corresponding residual plots (right). Figs A-F refers Ettinger (A), Fuerte (B), Hass (C), Nabal (D), Reed (E) and Mixed species (F).

Fig 5. Relationship between estimated and measured fruit volume of the 12 cross-validation test fruits.

Circles are the volume estimates calculated using the full data set equations and the crosses are the estimates calculated using the cross-validation training dataset equations. On Fig 5, the letter E, F, H, N, R and M refers Ettinger, Fuerte, Hass, Nabal, Reed and Mixed cultivar data.

Conclusions and recommendation

This study provided the first avocado cultivar-specific and mixed-cultivar generalized allometric equations to estimate avocado fruit volume non-destructively. Among tested model forms, VM2 (for cultivar-specific model), and VM7 (generalized model) have passed all rigorous verification and cross-validation statistical tests, confirming that the models have sufficient skill to estimate fruit volume from easily measurable parameters. Our best allometric models (VM2 and VM7) explained > 87% of the variation in measured fruit volumes of each cultivar. A high degree of correlation (R2 > 0.93) between measured and estimated fruit volume provided quantitative evidence of the validity of the selected volume estimation models. The allometric equation developed in this research could be practical in the estimation of avocado fruit volume and applicable under field conditions. Besides, the generalized mixed-cultivar model can reliably be used to estimate avocado fruit volume when cultivar type is unknown. Our finding revealed that in the situations where fruit length and diameter measurements are possible and/or where measurements of volume would be inconvenient, or time-consuming, site- and cultivar-specific allometric equations can be used to estimate fruit volume while it is on the tree. Therefore, the allometric equations generated in this study could play a considerable role in improving data availability on avocado fruit physical appearance which is critical to assess the quality and taste of fresh products, which in turn, influences the purchase decision of customers. It can also potentially assist horticulturists, agronomists, and physiologists to estimate fruit volume of avocado accurately and to carry out yield estimation before harvesting. Finally, we also recommend conducting a similar study using a large dataset collected from different agro-ecological regions, which would help to have a robust generalized and species-specific avocado volume estimation model that can be used across regions.

Supporting information

S1 File. Colored photograph of five Avocado cultivars.



We thank the local community and local administration offices in the study area for their support during fieldwork. We are also grateful to Mr. Eyuel Tesfaye, and Mr. Demeke Beyen for their facilitation of the fieldwork and organizing the team of experts for data collection. The contents of this document are solely the responsibility of the author/s and do not necessarily represent the official views of USAID or the U.S. Government or that of the institutional position of ICRAF (World Agroforestry).


  1. 1. Schaffer B, Whiley AW. Environmental regulation of photosynthesis in avocado trees–A mini Review. Actas V Congr Mund del Aguacate. 2003; 335–342.
  2. 2. Bayram S, Tepe S. Determination of some physicochemical properties in fruits of some avocado (Persea americana Mill.) cultivars during the harvesting periods. Derim. 2019.
  3. 3. FAOSTAT. Agricultural production data. 2020 [cited 25 Dec 2020]. Available:
  4. 4. Altendorf S. Major tropical fruits market review 2018. Rome, FAO. The; 2019. Available:
  5. 5. Tadesse A. Market Chain Analysis Of Fruits For Gomma Woreda, Jimma Zone, Oromia National Regional State. Haramaya University. 2011.
  6. 6. Faris A. Review on Avocado Value Chain in Ethiopia. Ind Eng Lett. 2016;6: 33–40. pmid:2225
  7. 7. Megerssa B. The role of avocado production in coffee based farming systems of South Western Ethiopia: The case of Jimma zone. African J Agric Res. 2013;8: 2285–2296.
  8. 8. Dorantes L, Parada L, Ortiz A. Post-harvest Operations. Mejía AD, editor. 2004.
  9. 9. Palou A, Serra F, Pico C. General aspects on the assessment of functional foods in the European Union. Eur J Clin Nutr. 2003;57: S12–S17. pmid:12947445
  10. 10. USDA. Avocados, Raw, all commercial varietie. In: [Internet]. 2020 pp. 1–43.
  11. 11. Fulgoni VL, Dreher M, Davenport AJ. Avocado consumption is associated with better diet quality and nutrient intake, and lower metabolic syndrome risk in US adults: Results from the National Health and Nutrition Examination Survey (NHANES) 2001–2008. Nutr J. 2013;12: 1–6. pmid:23282226
  12. 12. Shumeta Z. Avocado Production and Marketing in Southwestern Ethiopia. Trends Agric Econ. 2010;3: 190–206.
  13. 13. Elias S. Supply Chain Analysis of Avocado and Mango Fruits in Gedeo Zone. 2016;6: 46–55.
  14. 14. Terheggen A, Chains GV, Trade C, Livelihoods L. Ethiopia ‘ s Potential Role in the Global Avocado Market. 2019.
  15. 15. Kadri Bozokalfa M, Kilic M. Mathematical Modeling in the Estimation of Pepper (Capsicum annuum L.) Fruit Volume. Chil J Agric Res. 2010;70: 626–632.
  16. 16. Wang Z, Walsh KB, Verma B. On-tree mango fruit size estimation using RGB-D images. Sensors (Switzerland). 2017;17: 1–15. pmid:29182534
  17. 17. Ngouajio M, Kirk W, Goldy R. A simple model for rapid and nondestructive estimation of bell pepper fruit volume. HortScience. 2003;38: 509–511.
  18. 18. Moreda GP, Ortiz-Cañavate J, García-Ramos FJ, Ruiz-Altisent M. Non-destructive technologies for fruit and vegetable size determination—A review. J Food Eng. 2009;92: 119–136.
  19. 19. Riga FT. Biomass yield and nutritive value of sweet lupine in mid altitudes of Lemo District, Hadiya Zone, Southern Ethiopia. Wolaita Sodo University. 2017. pmid:16333192
  20. 20. CSA-Ethiopia. Population and Housing Census 2007 Report. Cent Stat Agency, Addis Ababa. 2007.
  21. 21. Mekonnen K, Yasabu S, Gebremedhin B, Woldemeskel E, Tegegne A, Thorne P, editors. Promoting productivity and market access technologies and approaches to improve farm income and livelihoods in Ethiopia: Lessons from action research projects. Proceedings of a Workshop, ILRI, Addis Ababa, 8–9 December 2016. Nairobi, Kenya: ILRI. Nairobi, Kenya: ILRI,; 2019. Available: http://africa-rising.
  22. 22. Ali S, Plotto A, Scully BT, Wood D, Stover E, Owens N, et al. Fatty acid and volatile organic compound profiling of avocado germplasm grown under East-Central Florida conditions. Sci Hortic (Amsterdam). 2020;261.
  23. 23. Beyene K, Advisor: Characterization of Hass and Fuerte Avocado (Persea americana) Varieties Grown in Wondogenet and Optimization of Some Oil Extraction Parameters. Addis Ababa University. 2017.
  24. 24. Schaffer B, Baranowski RM. Sample size estimates for avocado yield experiments. J Am Soc Hortic Sci. 1986;111: 985–987.
  25. 25. Zilkah S, Klein I. Growth Kinetics and Deremiantion of Shape and Size of Small and Large Avocado Fruits Cultivar’ Hass’ On the Tree. Sci Hortic (Amsterdam). 1987;32: 195–202.
  26. 26. Marini RP. Estimating mean fruit weight and mean fruit value for apple trees: Comparison of two sampling methods with the true mean. J Am Soc Hortic Sci. 2001;126: 503–510.
  27. 27. Marini RP, Schupp JR, Baugher TA, Crassweller R. Sampling apple trees to accurately estimate mean fruit weight and fruit size distribution. HortScience. 2019;54: 1017–1022.
  28. 28. Erickson LC, Kikuta Y. Determination of Surface Area and Volume of Avocado Fruits. In: California Avocado Society Yearbook. 1965 pp. 103–106.
  29. 29. Sileshi GW. A critical review of forest biomass estimation models, common mistakes and corrective measures. For Ecol Manage. 2014;329: 237–254.
  30. 30. Gill JL. Outliers, residuals, and influence in multiple regression. J Anim Breed Genet. 1986.
  31. 31. Mahanti KK, Rathan SK, Ravanappa V, Vadivel S, Ramakrishnaiah JB. A Regression Model for Nondestructive Fruit Volume Estimation in Karonda (Carissa Carandus L.). Suppl Agron. 2016;11: 659–662.
  32. 32. Johnston R, Jones K, Manley D. Confounding and collinearity in regression analysis: a cautionary tale and an alternative procedure, illustrated by studies of British voting behaviour. Qual Quant. 2018;52: 1957–1976. pmid:29937587
  33. 33. Seong C, Herand Y, Benham BL. Automatic calibration tool for hydrologic simulation program-FORTRAN using a shuffled complex evolution algorithm. Water (Switzerland). 2015;7: 503–527.
  34. 34. Mokria M, Gebrekirstos A, Abiyu A, Van Noordwijka M, Braüning A. Multi-century tree-ring precipitation record reveals increasing frequency of extreme dry events in the upper Blue Nile River catchment. Glob Chang Biol. 2017;23: 2222–2249. pmid:28712116
  35. 35. Mokria M, Mekuria W, Gebrekirstos A, Aynekulu E, Belay B, Gashaw T, et al. Mixed-species allometric equations and estimation of aboveground biomass and carbon stocks in restoring degraded landscape in northern Ethiopia. Environmeantal Res Lett. 2018;00: 00.
  36. 36. Rahman SMAK, Sathik MM, Kannan KS. Multiple Linear Regression Models in Outlier Detection. Int J Res Comput Sci. 2012;2: 23–28.
  37. 37. Negash M, Starr M, Kanninen M, Berhe L. Allometric equations for estimating aboveground biomass of Coffea arabica L. grown in the Rift Valley escarpment of Ethiopia. Agrofor Syst. 2013;87: 953–966.
  38. 38. Burnham KP, Anderson DR. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach (2nd ed). Ecological Modelling. 2002.
  39. 39. Bayram S, Alper Arslan M, Turgutoǧlu E, Erkan M. The performance of some avocado cultivars under Mediterranean coastal conditions in Turkey. J Food, Agric Environ. 2012;10: 588–592.
  40. 40. Janice DA, John A, Jemmy FT. Morphological characteristics of avocado (Persea americana Mill.) in Ghana. African J Plant Sci. 2018;12: 88–97.
  41. 41. Lee SK, Young RE, Schiffman PM, Coggins CW. Maturity Studies of Avocado Fruit Based on Picking Dates and Dry Weight. J Amer Soc Hort Sci. 1983;108: 390–394.
  42. 42. Woolf A, Clark C, Terander E, Phetsomphou V, Hofshi R, Arpaia L, et al. Measuring avocado maturity; ongoing developments Allan. Orchard. 2003.
  43. 43. Barmore CR. Avocado Fruit Maturity. Proc First Int Trop Fruit Short Course Avocado Sauls JW, Phillips RL Jackson LK (eds) Gainesv Fruit Crop Dept, Florida Coop Ext Serv Inst Food Agric Sci Univers. 1976; 103–109.
  44. 44. Ammann CM, Wahl ER. The importance of the geophysical context in statistical evaluations of climate reconstruction procedures. Clim Change. 2007;85: 71–88.
  45. 45. Krause P, Boyle DP, Base F. Comparison of different efficiency criteria for hydrological model assessment. Adv Geosci. 2005;5: 89–97.
  46. 46. Moriasi DN, Arnold JG, Liew MW Van, Bingner RL, Harmel RD, Veith TL. Model evaluation guidelines for systematic quantification of accuracy in watershed simulations. Am Soc Agric Biol Eng. 2007;50: 885–900.
  47. 47. Vanderlei RS, Silva JV, Moura F de BP. Biometry and non-destructive allometric model for estimating Babassu (Attalea speciosa) fruit volume and densities. Rev Ouricuri. 2019;9: 1–10.
  48. 48. Ziaratban A, Azadbakht M, Ghasemnezhad A. Modeling of volume and surface area of apple from their geometric characteristics and artificial neural network. Int J Food Prop. 2017;20: 762–768.