INFORM October 2024

QUALITY ANALYSIS

inform October 2024, Vol. 35 (9) • 21

In 2024, worldwide demand for soybeans is expected to reach nearly 383 million metric tons, increasing in parallel with global GDP growth. As global soybean trade grows there is greater attention placed on the compositional differences associated with trade origins. In 1990, a research team analyzed the quality of soybean shipments from Argentina, Brazil, Paraguay and the US to five western European and three east Asian ports between 1985 to 1989 (doi.org/10.1007/BF02540483). They collected data on weight, damaged seeds, foreign matter, splits, free fatty acids (FFA) of degummed oil, moisture, oil and protein con tent, phosphorus and the total oxidation value of the oil. The researchers found that soybeans from the US consistently weighed more than South American soybeans. Soybeans from Brazil had a higher percentage of damaged seeds than those from the US. Argentina had the lowest percentage of dam aged seeds. US seeds had more foreign matter than Argentina seeds. However, Argentine soybeans had the most split seeds.

Finally, the researcher concluded that oil quality from US soy beans was highest and contained the least amount of FFA. The review paper revisited past quality studies by performing a meta-analysis on their results. To address the insufficient data regarding soybean oil composition or quality differences based on origin, our team collected and tested samples of crude degummed soybean oil from shipping vessels or domestic soybean crushing facilities between 2020 and 2022. COUNTRY OF ORIGIN AND MEAL COMPOSITION Several soybean meal quality studies have been conducted in the past two decades. Between 2007 and 2015, a research team from Spain analyzed 515 meal samples from Argentina, Brazil, and the US. They found that US and Brazilian meal had higher crude protein. A team at the University of Illinois reported a similar result in the paper they published a year later, with the added note that US meal contained more indis

Table 1. Chemical composition plot of soy meal based on studies conducted on US and Brazilian soybeans. (KOH – KOH soluble protein). US BR Std. Mean Difference Std. Mean Difference Study or Subgroup Mean SD Total Mean SD Total Weight IV, Random, 95% CI IV, Random, 95% CI

3.1.1 Sucrose Cámara 2017

8.29 0.837 32

6.24

0.837 26 11.1%

2.42 (1.73, 3.11] 3.01 [0.91, 5.11] 2.02 [1. 76, 2.28] 2.98 [0.89, 5.07] 2.71 [1.05, 4.36] 3.11 [0.79, 5.43] 1.96 [0.98, 2.94] 2.11 [1.88, 2.34]

Galkanda-Arachchige 2020 García-Rebollar 2016

8.64 0.93

5

5.54 0.93

5

1.2%

8.43 1.07 180

6.43 5.52 3.72

0.89 165 78.1%

Lagos 2017

8.59 0.93 5.94 0.8 8.594 0.97 8.29 1.01

5 6 5

0.93 0.73

5 7 4

1.2% 1.9% 1.0% 5.5%

Li 2015

Lopez 2020 Ravindran 2014 Subtotal (95% CI)

5.347 0.87

16

6.3

0.94

10

249

222 100.0%

Heterogeneity: Tau² = 0.00; Chi² = 3.90, df = 6 (P = 0.69); I² = 0% Test for overall effect: Z = 17.98 (P < 0.00001) 3.1.2 Raffinose Cámara 2017 1.21 0.218 32

1.64 0.218 26 20.7%

-1.95 [-2.58, -1.31] -0.32 [-1.58, 0.93] -1.78 [2.03, -1.53] -0.31 [1.56, 0.94] -0.46 [-1.57, 0.65] -0.04 [-1.36, 1.27] -1.00 [-1.67, -0.32]

Galkanda-Arachchige 2020 García-Rebollar 2016

1.46 0.252 1.09 0.28 180 5

1.55

0.252

5 13.5%

1.58 0.27 165 24.5%

Lagos 2017

1.45 0.26 0.82 0.12 1.447 0.25

5 6 5

1.54 0.26

5 13.5% 7 15.0% 4 12.9% 212 100.0%

Li 2015

0.9

0.19

Lopez 2020

1.458 0.22

Subtotal (95% CI)

233

Heterogeneity: Tau² = 0.47; Chi² = 20.55,df = 5 (P = .0010); I² = 76% Test for overall effect: Z = 2.90 (P = 0.004) 3.1.3 Stachyose Cámara 2017 5.68 0.598 32 4.79

0.598 26 29.9%

1.47 [0.88, 2.06] 2.97 [0.89, 5.05] 2.15 [1.88, 2.41] 2.91 [0.86, 4.97] 2.81 [1.12, 4.50] 2.89 [0.68, 5.11] 2.08 [1.64, 2.52]

Galkanda-Arachchige 2020 García-Rebollar 2016

6.51 0.614 6.39 0.53 180 5

4.49

0.614

5

4.2%

5.25 0.53 165 51.8%

Lagos 2017

6.47 0.62 2.66 0.48 6.471 0.66

5 6 5

4.47 1.41

0.62 0.35

5 7 4

4.3% 6.1% 3.7%

Li 2015

Lopez 2020

4.414 0.592

Subtotal (95% CI)

233

212 100.0%

Heterogeneity: Tau² = 0.08; Chi² = 6.98, df = 5 (P = 0.22); I² = 28% Test for overall effect: Z = 9.26 (P < 0.00001) 3.1.4 KOH Cámara 2017 80.4 4.486 32 79

4.486 26 28.8%

0.31 [-0.21, 0.83] 0.95 [0.73, 1.17] 1.58 [0.51, 2.65] 1.34 [0.46, 2.23] 0.92 [0.45, 1.38]

García-Rebollar 2016

86.1

4.3

180 82

4.3 1.53 3.17

165 40.8% 7 13.3%

Park 2002

86.45 1.96

13 83.43 16 72.5

Ravindran 2014 Subtotal (95% CI)

77.2

3.52

10

17.1%

241

208 100.0%

Heterogeneity: Tau² = 0.13;Chi² = 7.69, df = 3 (P = 0.05); I² = 61% Test for overall effect: Z = 3.86 (P = 0.0001)

-4 -2 0 4 Higher in United States Higher in Brazil 2

Test for subqroup differences: Chi²= 87.37. df = 3 (P < 0.00001), 1² = 96.6%