Four corn varieties were grown at one location, a village in the north of China, in 2012, 2013 and 2014. All these varieties were conventional varieties, and representative varieties widely planted in the north of China. According to the differences of hardness, the corn were classified as soft (LS1), hard (LS3), and two intermediate hardness (LS2; LS4). The harvest was carried out in early October in each of the three years. Table 1 is the meteorological index over three growing seasons at the planting area. During each year, four corn varieties were sun dried to about 14% moisture content.

After the corn was dried (early November), twenty-four barrows (Duroc×Landrace×Yorkshire; 2012: initial body wight [BW] = 33.27±4.30 kg; 2013: initial BW = 31.88±2.93 kg; 2014: initial BW = 34.21±3.81 kg) were individually housed in stainless-steel metabolism crates (1.4×0.7×0.6 m) at the Fengning Swine Research Unit of China Agricultural University (Hebei, China). A feeder and a nipple drinker were installed in each pen. The crates were located in a room with temperature maintained at 22°C±1°C. In each of three years, pigs were allotted to one of four diets according to a completely randomized design, and each diet was fed to six pigs.

In 2012, 2013, and 2014, four diets were formulated to contain 96.8% of each corn variety and 3.2% other ingredients including vitamins and minerals (Table 2). All diets were fed in mash form. Corn was included as the sole energy and protein source in the diet, and vitamins, minerals and salt were added to meet or exceed the estimated requirements for growing pigs (NRC, 1998).

Feed was provided in equal amounts twice daily, at 0800 and 1700 h. Water was available continuously for each pig. During the adjustment period to the metabolism crates and diets, average daily feed intake was gradually increased until it was estimated to supply 4% of the BW determined at the initiation of each adaptation period. During the collection period, all fresh fecal samples were collected as often as possible throughout the day and were stored at −20°C. The collection and sample preparation of feces and urine were conducted according to the methods described by Li et al. (2015).

At the end of the animal experiment, fecal samples were dried at 60°C in a forced-air oven for 72 h, and then were finely ground. Diets and feces were analyzed for dry matter (DM) (AOAC, 2000), ether extract (EE) (Thiex et al., 2003) and ash (AOAC, 2000). Kjeldahl N was determined according to the method of Thiex et al. (2002). Diet and fecal samples were analyzed for neutral detergent fiber (NDF) and acid detergent fiber (ADF) by using filter bags and fiber analyzer equipment (Fiber Analyzer, Ankom Technology, Macedon, NY, USA) following a modification of the procedure of Van Soest et al. (1991). Starch was determined after converting starch to glucose using an enzyme assay kit (Megazym International Ireland, Wicklow, Ireland). The gross energy (GE) of urine was measured by applying 4 mL of urine sample onto 2 filter papers in a special crucible manufactured by Parr Instrument Company and then dried for 8 h in a 65°C drying oven. The GE of feces and diets were measured using an isoperibol oxygen bomb calorimeter (Parr 6400 Calorimeter, Moline, IL, USA).

Energy values determined from the GE in the feces and urine were subtracted from the intake of GE to calculate DE and metabolizable energy (ME) for each diet. The apparent total tract digestibility (ATTD) of chemical constituents determined from the excretion of chemical constituents in the feces were first subtracted from the intake of the chemical constituents then were divided by the intake of the chemical constituents to calculate the ATTD of the chemical constituents for each diet and corn (Kong and Adeola, 2014). The DE and ME in the corn diets were divided by 0.968 to calculate the DE and ME of the corn itself.

Data were analyzed as a randomized complete block design by using PROC general linear model of SAS (SAS Inst. Inc., Cary, NC, USA) with pig as the experimental unit. Main effects of variety and planting year interactions were tested. Means were separated using the least squared means procedure and the PDIFF option of SAS. The alpha level of 0.05 was used to assess the significance of difference between means. The proximate chemical composition, GE, DE, and ME of corn, sunshine duration, and rainfall were correlated using PROC CORR of SAS.

The chemical compositions of four corn varieties over three consecutive years are presented in Table 3. The average concentration of DM in 12 corn samples pooled in 2012, 2013, and 2014 were 86.46%, ranging from 85.14% to 88.49%. On a DM basis, the concentrations of starch, EE, NDF, ADF, crude protein (CP), ash, and GE averaged 73.06% (71.41% to 74.42%), 4.15% (3.33% to 5.53%), 13.14% (10.72% to 15.34%), 2.52% (2.05% to 2.97%), 8.66% (8.13% to 9.31%), 1.32% (1.19% to 1.51%), and 18.48 MJ/kg (18.19% to 18.62 MJ/kg), respectively. The coefficient of variation (CV) of 12 corn samples for starch, CP, ash, and GE were within 10%, but wide variations in the content of EE (CV: 16.60%), NDF (CV: 10.68%) and ADF (CV: 14.43%). In this experiment, variety and planting year are the major sources of variations in the chemical composition.

The DE and ME contents of the 12 corn samples and ATTD of nutrients are presented in Table 4. The interaction between variety and planting year was not significant except the ATTD of GE (p<0.05), ADF (p<0.01) and CP (p<0.01). Variety LS1 had a higher (p<0.05) DE than other corn varieties (LS2, LS3, and LS4); no differences among LS2, LS3, and LS4 were observed. The ATTD of organic matter (OM), DM, GE, NDF, and ADF was also greater (p<0.05) in LS1 than in LS2, LS3, and LS4. The DE and ME contents in 2012 was greater (p<0.05) than in 2013 and 2014; no differences between 2013 and 2014 were found. The ATTD of NDF, ADF in 2014 was lower (p<0.01) than that in 2012 and 2013; however, the ATTD of CP in 2014 was greater (p<0.01) than that in 2012 and 2013. The ATTD of EE in 2012 was greater (p<0.01) than in 2013, and the ATTD of EE in 2013 was greater (p<0.01) than in 2014.

Correlations of parameters that determine DE and ME are presented in Table 5. The content of NDF had a significant negative correlation with DE and ME contents (r = −0.63; p<0.05; r = −0.81; p<0.01, respectively). The content of EE had a significant negative correlation with ME (r = −0.63; p<0.01). The sunshine duration (July, August, and September) was significantly positively correlated with ME (r = 0.58; p<0.05). And the content of GE was also significantly positively correlated with ME (r = 0.57; p<0.05).

Among four varieties of corn, variety LS1 had the highest DE and ME contents over three consecutive years. The ATTD of OM, DM, GE, NDF, and ADF was also greater in LS1 than in LS2, LS3, and LS4. When compared among three planting years, the DE and ME contents in 2012 was greater than in 2013 and 2014. And no interaction between variety and planting year in DE and ME contents was found. All these results showed that although the DE and ME contents significantly varied from year to year, the LS1 had higher DE and ME contents than all other corn in these three years. The results provide a probability that the rank of varieties in DE and ME contents is going to be stable in one area in subsequent years. It should be possible to grow certain varieties of conventional corn containing greater DE and ME when fed to growing pigs in that location.

From Table 4, the DE and ME contents in 2012 were greater than in 2013 and 2014; while no differences between 2013 and 2014 were observed. According to Table 3, the starch, EE, NDF, ADF, CP, and GE contents were variable in these three years. The lower DE and ME in 2013 and 2014 than in 2012 may be the result of higher NDF and lower GE in 2013 and 2014 than in 2012 (Noblet and Perez, 1993; Li et al., 2014a). One reason for the variation in chemical composition of corn is the difference in climatic conditions. According to Table 1, the sunshine duration, average temperature, relative humidity and rainfall were variable in the three growing seasons. Some reports indicate that starch, CP and EE contents are significantly influenced by sunshine duration, rainfall and average temperature in growing seasons (Genter et al., 1956; Thompson et al., 1973; Leeson and Summers, 1976). Another reason for the variable available energy of corn from year to year is the different moisture content of corn at harvest. Although corn was harvested at the same time in each of planting years, the initial moisture varied significantly from year to year due to different weather conditions in both growing and harvesting periods. In 2012 corn was harvested at 20% to 25% moisture content, in 2013 at 27% to 37% moisture content, in 2014 at 26% to 33% moisture content. The moisture levels of corn at harvest could represent the stage of kernel maturity (Leeson and Summers, 1976; Leeson et al., 1977). The ME of corn fed to adult roosters is negatively correlated with the moisture levels of corn at harvest (Leeson and Summers, 1976; Leeson et al., 1977). In addition, another possible explanation for the variable DE and ME of corn from year to year is differences in the agronomic practices used, such as irrigation and N fertilization. Some reports indicate that irrigation and N fertilization could influence the protein and lysine content of normal corn (Cromwell et al., 1983; Kniep and Mason, 1991). The experiment conducted by Cromwell et al. (1983) suggested that N fertilization of normal corn, while increasing its protein and lysine content, has no beneficial effect on its nutritive value for chicks.

Correlations of parameters that determine DE and ME contents of 12 corn samples are presented in Table 5. From Table 5, the sunshine duration and GE contents were significantly positively correlated with ME, but the NDF content was significantly negatively correlated with DE and ME contents. These results are in agreement with those reported by Li et al. (2014a). The EE content was significantly negatively correlated with ME content, which is not in line with the data of Li et al. (2014a). According to Table 3, the EE content in 2013 was greater than in 2012 and 2014. However, from Table 4, the DE and ME contents in 2013 were lower than in 2012; while no differences between 2013 and 2014 were observed. The factors determining the DE and ME were not only the EE content, since other chemical composition differences, such as NDF, ADF, CP, starch, also correlated with the DE and ME contents. These differences may be the reason for the disagreement between our findings and those reported by Li et al. (2014a). Interestingly, the sunshine duration (July, August and September) was positively correlated with DE, ME and starch contents while the rainfall (July, August, and September) was negatively correlated with DE and ME contents, but positively correlated with NDF content. Such data provides the possibility that we can predict the DE and ME contents according to the rainfall and sunshine duration in a specific location.