基于关联度、主成分和聚类分析的西南区玉米新组合评价
2020-05-19范继征石达金吕巨智唐国荣程伟东
范继征, 石达金, 吕巨智, 唐国荣, 程伟东
(广西农业科学院玉米研究所, 南宁 530007)
中国西南区玉米常年种植面积在466.67万hm2,主要分布在广西、云南、贵州、四川和重庆等地,是我国第三大玉米产区[1]。该地区以山地和丘陵为主,地形地貌复杂、生态环境多样,用常规方法进行统计学分析,来评价参试玉米品种的优劣,常会造成一定程度的偏差。此外,由于产量、品质等性状多为数量性状,依据个别性状评价不同品种的优劣,通常会出现顾此失彼的现象。因此,在对玉米品种筛选鉴定过程中,需要采用科学合理的方法结合不同性状的综合表现进行选择。
灰色关联度分析法可以在较少数据或者没有显著相关性条件下对样本数据进行分析,找出各因素与参考数列的关系,该方法具有所需样本少、方法简便、信息量大等优点[2],已在玉米品种鉴定、新组合筛选中取得良好效果[3-7]。主成分分析法主要是采用降维的思想,把一个多属性的复杂问题转化为少数属性的简单问题。具体方法就是把多个指标转化为少数几个综合指标,使这些综合指标尽可能地反映原始数据的信息量[8-9]。聚类分析则是一种将各种事物或现象进行分类的方法,便于对样品进行综合分析,得出较为全面的分析结果。主成分分析结合聚类分析法已应用于多种农作物的综合评价[10-15],在玉米新品种和新组合的评价中也有较多报道[16-20]。
本研究以2017年国家玉米产业体系西南区55个玉米新组合为研究对象,基于14个与产量性状相关的农艺性状数据,综合应用灰色关联度分析、主成分分析和聚类分析来评价不同新组合的优劣,旨在为西南区玉米新组合评价提供科学依据。
1 材料与方法
1.1 试验材料
供试材料为国家玉米产业体系西南区各个单位提供的55个玉米新组合(表1),以及对照品种桂单162和渝单8号。试验于2017年在广西农业科学院明阳基地进行。
1.2 试验方法
试验采用双对照间比法排列,不设重复,每5个参试品种设2个对照。种植密度为5.7×104株·hm-2,小区面积为21 m2,5行区,行长6 m,行距0.7 m。试验田四周设大于5行保护行,田间管理同当地水平。
表1 55个西南区玉米新组合
序号组合T1荣玉1754T2南T931T3绵TX1701T4渝单701T5靖2017-1T6德3239T7黔1701T8川单1751T9恩1077T10CS17-1T11云瑞506T12S183T13HZ1720T141701T15成单658序号组合T16青青711T17苏科1618T18荣玉1755T19南T928T20绵732T21渝单841T22靖2017-2T23德10T24黔1702T25川单1752T26恩1124T27CS17-2T28云瑞007T29桂A115T30HZ1721序号组合T311702T32成单388T33青青722T34苏科1614T35荣玉1756T36南T605T37绵1308T38渝单703T39靖2017-3T40恩912T41黔1703T42川单1753T43恩932T44黔1704T45云瑞1207序号组合T46GD1676T47HZ1722T481703T49成1702T50青青733T51苏科玉1604T52渝单836T53黔1705T54渝单896T55恩1010ck1桂单162ck2渝单8号
1.3 调查项目
调查株高、穗位高、茎粗、穗长、穗粗、秃尖长、穗行数、行粒数、百粒重、出籽率、籽粒含水量、单株粒重、倒折率、空杆率和产量等15个性状。果穗性状按照单穗平均法,收获时取有代表性的20个样穗进行考种,取平均值比较。
1.4 数据分析
采用Microsoft Excel 2016软件和DPS 14.0软件进行描述统计分析、灰色关联度分析、主成分分析和聚类分析。灰色关联度分析依据邓聚龙等方法[21-23],将55个新组合的产量作为参考数列X0,其余14个农艺性状为比较数列,分别为:株高(X1)、穗位高(X2)、茎粗(X3)、穗长(X4)、穗粗(X5)、秃尖长(X6)、穗行数(X7)、行粒数(X8)、百粒重(X9)、出籽率(X10)、籽粒含水量(X11)、单株粒重(X12)、倒折率(X13)、空杆率(X14),对数据进行标准化处理,并计算关联系数和关联度进行比较分析。此外,以产量最高组合为理想品种,计算不同组合的关联度并进行分析比较。
主成分分析依据杨锦越等的方法[17,24],对收集整理的农艺性状数据采用中心化的标准化处理,计算特征值、贡献率、累积贡献率和特征向量数据。然后,利用主成分的特征向量(Z),构建主成分与农艺性状之间的线性方程F。最后,结合主成分的方差贡献率构建玉米新组合综合评价模型λ。
2 结果分析
2.1 玉米不同新组合农艺性状和产量表现
从表2可以看出,55个不同玉米新组合的各个农艺性状和产量表现不同。参试组合的株高范围在229.30~329.10 cm,株高低于2个对照平均值的组合有22个。穗位高在86.60~148.70 cm之间,高于2个对照平均值的组合有7个。穗长在14.90~19.95 cm之间,比2个对照平均值长的组合有32个,大于19 cm的有5个组合,分别是南T 928、荣玉1756、渝单703、荣玉1754和渝单896。穗粗在45.05~55.68 mm之间,高于对照平均值的有37个组合,穗粗最大的组合为青青733。秃尖长在0.13~2.00 cm之间,秃尖长在1.5 cm以上的组合有5个,分别是青青733、绵1308、青青711、HZ 1722和恩1010。穗行数在13.00~20.20行之间,低于对照平均值的有4个组合,分别是成1702、云瑞1207、HZ 1721和德3239。行粒数在32.80~44.90粒之间,大于对照平均值的有11个组合,南T 928的行粒数最高。百粒重在23.52~36.25 g之间,大于对照平均值的有25个组合。单株粒重在0.094~0.197 kg之间,大于对照平均值的有33个组合,单株粒重大于0.180 kg的组合分别是川单1752、1703、成单388、绵1308和渝单836。倒折率在0~76.0%之间,未发生倒折的组合有25个,倒折率大于20%的组合有云瑞007、川单1753、荣玉1754和靖2017-2。空杆率在0~18.31%之间,空杆率为0的组合有14个,空杆率大于15%的组合有云瑞506和靖2017-2。
其中,倒折率、空杆率和秃尖长的变异系数较大,分别达到了202.05%、93.91%和47.18%。对于产量而言,产量表现高于2个对照的有28个组合,介于对照之间的有9个,低于2个对照的有18个。其中,1703的产量最高,为11 472.19 kg·hm-2,靖2017-2的产量最低,为5 495.12 kg·hm-2。
从变异系数大小来看,15个农艺性状的变异系数范围在1.78%~260.71%之间。其中,变异系数大于45%的性状有3个,从大到小依次为倒折率、空杆率和秃尖长;变异系数小于5%的性状仅有1个,即为出籽率;变异系数在10%~15%之间的性状有单株粒重、产量和穗位高;其余性状则介于5%~10%之间。
表2 不同组合的产量和农艺性状
组合X1/cmX2/cmX3/mmX4/cmX5/mmX6/cmX7X8X9/gX10/%X11/%X12/kgX13/%X14/%X0/(kg·hm-2)T1307.60146.5019.1519.1046.040.8015.639.629.3486.6812.850.1422.377.898165.04T2326.10148.7018.6018.5546.820.5716.439.031.0487.7614.400.1510.815.418665.09T3287.00134.9021.3117.7546.751.3015.634.730.4086.5312.550.0920.2716.225495.12T4265.50119.4017.1116.6048.920.7015.836.529.3388.2014.500.152.742.748877.95T5268.30111.0017.3017.0548.340.5514.638.633.9785.9912.900.175.885.888938.54T6271.80120.5019.0718.6549.610.3516.439.328.4188.5215.050.181.457.259969.00T7229.3099.0018.2718.9045.050.3515.039.128.8487.3212.750.145.260.008329.52T8280.80103.5019.5916.9554.980.7020.235.032.2289.3313.100.182.671.3310570.57T9277.60112.7020.1617.2549.371.3515.036.835.5687.1414.250.151.334.008753.77T10286.90123.6020.0217.4547.320.8016.642.326.4687.3013.500.168.115.419257.45T11299.80104.3018.0617.9046.620.6315.434.632.8285.4814.350.130.007.897887.46T12264.60117.3017.2216.7546.870.9316.037.329.2687.4416.150.140.000.008273.99T13276.20125.2018.5016.0546.300.8515.237.428.9086.7412.800.131.354.057642.75T14296.40124.6018.8117.1550.220.7516.435.528.6486.4913.200.142.744.117822.85T15285.20125.2019.8217.1550.680.6718.838.030.0585.2014.650.160.008.828873.76T16291.60115.1018.4817.3046.470.5216.238.923.7785.4714.100.132.700.007684.36T17244.8097.5018.2618.2545.111.3213.036.030.1386.5813.050.122.701.357006.76T18263.5098.2016.3616.8048.410.8018.035.428.3187.7113.500.120.006.257839.99T19273.10100.4018.4917.4548.610.6515.036.335.2687.0114.200.170.002.6710005.75T20258.90116.5019.1919.0046.681.0515.642.029.6486.5315.300.170.006.769755.02T21263.30108.7018.4717.1546.280.6016.034.533.2286.9913.400.160.000.009664.01T22286.20117.2019.1717.6049.181.0016.033.534.5886.2913.200.179.331.3310387.11T23279.80122.1019.6217.1052.921.5016.635.935.7286.8615.050.170.006.769968.41T24286.90123.0020.2417.8549.360.3515.642.134.5087.1813.550.205.480.0011394.58T25274.30103.2020.4617.3550.620.9519.037.830.4488.0914.550.190.001.3011472.19T26245.90121.1018.2217.1549.170.3515.835.733.4388.1714.900.180.000.0010580.64T27248.10119.2021.2616.2545.251.2515.237.331.6286.8412.800.1176.001.336398.66T28283.80125.8020.1919.4549.300.9515.641.231.4188.3014.750.185.196.4910769.43T29239.6097.9020.3916.9045.981.2515.840.931.9387.2912.250.141.4118.317638.49T30248.70109.8019.6916.9547.930.4015.841.228.8188.2613.150.140.000.008708.21T31271.00135.0017.1416.1052.560.7516.036.535.4982.9614.750.175.631.419629.65T32320.80137.2020.6017.9551.270.6515.036.336.2587.2512.300.1626.323.959604.18T33258.00110.5019.2718.2548.881.4015.640.628.8586.3214.200.145.638.457772.61T34281.40118.1017.0416.5049.930.9516.435.832.2684.2614.050.150.001.338835.87T35234.70122.5020.0015.9047.210.9014.432.833.9583.1113.000.121.356.766979.28T36259.7098.5018.3418.3550.810.3817.640.930.7588.0613.950.170.0010.149296.14T37247.9098.9018.5819.9552.091.2516.044.926.7886.9712.850.130.0010.007449.14T38272.30111.2020.7916.4048.970.1316.439.130.2489.6914.450.160.001.329794.59T39285.40125.9017.7017.5045.411.0215.437.328.5886.6812.800.140.000.008244.72T40275.40103.7018.6118.3550.741.8518.636.330.4389.5415.950.180.001.3310842.15T41262.60120.5017.6617.2055.682.0018.439.033.5586.4614.550.160.000.009163.70T42318.00144.0019.0518.0547.500.7515.439.529.8188.2117.150.170.004.119632.69T43254.40104.1016.6619.3048.620.6516.040.929.6789.3913.900.150.001.339113.90T44312.90122.6019.4117.4549.740.4015.439.430.3888.7514.500.140.005.198797.48T45284.10123.0016.0316.9047.370.4018.633.427.9486.4713.350.161.320.009350.56T46283.70110.0017.8017.1052.301.7516.435.734.6286.0615.950.180.000.0010876.78T47278.80126.7016.5914.9045.750.6014.834.327.6287.7113.750.120.001.277675.31T48259.90108.0017.6216.2546.810.9014.237.631.3687.6414.550.160.002.709547.39T49276.90112.1015.6518.3549.700.8016.638.333.3586.9614.500.160.000.009942.36T50302.80135.1017.4118.8546.331.3514.839.532.1882.0814.250.144.059.468105.73T51268.10126.1019.5218.1554.360.6518.838.530.0287.7515.150.190.004.0511001.51T52283.30117.0018.5219.0547.621.0016.840.423.5285.8113.650.130.007.696633.15
续表2
组合X1/cmX2/cmX3/mmX4/cmX5/mmX6/cmX7X8X9/gX10/%X11/%X12/kgX13/%X14/%X0/(kg·hm-2)T53251.9086.6017.1618.2548.330.6516.837.830.6886.8715.250.180.000.0010359.32T54303.50113.0018.7917.3047.431.7015.434.532.7585.5414.900.140.006.588411.72T55246.0094.1018.0518.2046.630.8516.238.431.6990.0612.500.185.410.0010331.02ck1281.68142.6219.8616.9447.170.3414.140.032.5485.3914.150.153.734.938909.69ck2257.88113.8518.9418.3646.740.5715.539.529.2987.0913.590.141.586.088299.01平均值275.36117.2718.7217.6148.740.8816.1738.0131.0187.0114.070.155.504.319014.97标准差21.5213.721.321.002.490.411.352.572.791.551.040.0211.114.051306.44变异系数7.8211.707.055.685.1147.188.386.778.991.787.4214.27202.0593.9114.49
2.2 玉米不同新组合产量与农艺性状的关联度分析
不同玉米组合各农艺性状与产量的灰色关联系数见表3。不同组合之间同一性状的关联系数表现各不相同,其中,秃尖长的关联系数范围在0.333 4~0.995 9之间,变化幅度最大;单株粒重的关联系数范围在0.863 9~0.999 3之间,变化幅度最小。此外,同一组合不同性状间的关联系数变化明显。其中,恩932不同性状的关联系数变化范围在0.842 0~0.982 0之间,青青733的关联系数变化范围在0.333 4~0.988 0之间。
表3 产量与主要农艺性状的关联系数
组合X1X2X3X4X5X6X7X8X9X10X11X12X13X14T10.75680.65520.85020.78760.94660.96630.91780.82910.94440.88330.98950.96620.86860.9292T20.74680.68020.95580.87960.99580.68530.92560.91170.94530.93800.91380.99840.95040.9701T30.60210.54680.55350.62250.65250.42050.64780.68370.63880.63150.69840.99900.74850.7154T40.96730.95150.89960.93600.97520.79100.98760.96170.94110.96370.93540.97810.96140.9643T50.97220.93540.90390.96180.99710.64880.87990.96690.86510.98880.89660.88030.98030.9773T60.84440.89280.87950.92940.87860.48260.87630.89880.77310.87560.94650.88900.88910.8196T70.87580.89100.92910.81670.99740.55760.99520.86360.99350.89650.97260.96560.91480.8519T80.80870.69230.83560.75420.93280.64440.89570.72040.82810.81280.72940.97860.80230.8141T90.94840.98480.86310.99110.94280.52270.93700.99340.79000.96140.94080.98220.92280.9633T100.97990.96030.94200.94410.91820.86900.99870.88570.78830.95910.90560.99820.90100.9347T110.75460.97850.88160.82460.89150.81950.89460.95060.78260.86330.81890.96740.79860.8901T120.94010.88791.00000.95490.94000.80150.90170.91330.96470.88730.74000.98320.84280.8451T130.80930.74810.82510.91430.86710.82470.87680.82900.88810.81830.91410.99860.78600.8156T140.75920.77020.82880.86310.80270.99590.81730.90980.92340.84240.90340.98070.82000.8371T150.92910.88690.90050.98040.92480.75940.78600.97930.97440.98600.92070.88110.92210.9443T160.76100.83490.83110.83670.86830.72390.81420.79390.88200.83850.81400.99860.80410.7791T170.85530.92350.76920.71800.81690.46360.96130.79490.77220.75340.81340.99870.73590.7261T180.88380.95280.99630.88860.84330.91730.72870.91560.94030.82950.87980.91030.79350.8638T190.84510.71990.84020.84290.85050.64730.78120.80670.96310.85120.86560.97970.90260.8669T200.81990.88030.91650.99090.83750.82690.84700.96870.83570.87670.99300.99810.93880.8504T210.84770.81800.88180.86630.83970.63460.88710.79740.99570.89520.84440.98040.95260.9497T220.85080.81020.83360.80760.81770.98730.79990.70560.94320.79850.75250.97900.75680.8353T230.87740.90990.91560.82610.96690.50670.89110.80040.93690.85390.94660.99810.90780.8249T240.74470.75200.77900.72060.72020.43200.68520.80500.80960.71020.68390.97210.69870.7423T250.70130.62430.78190.69230.73410.79330.86920.70000.69030.71160.73180.93130.73680.7241T260.69820.82220.76310.76320.79600.45950.76800.73440.86980.79850.84980.97860.82950.8272T270.77460.68030.60630.75610.75080.46820.73960.70730.67940.69670.76600.98700.58630.6753T280.79760.84250.84620.87520.77860.87640.73900.85310.78040.78000.81620.93520.75730.7439T290.96820.98120.73150.85600.87410.51950.83470.74190.78360.81140.96630.94660.78610.9999T300.91050.95640.88660.99090.97720.56530.98400.84880.94410.93670.95330.96410.89880.9014T310.88410.88730.80870.80670.98830.76750.89170.85640.89820.84600.96950.93360.88210.9350T320.86960.86200.95260.93050.97630.67720.82520.85370.86570.90720.77290.96050.68550.9045T330.89900.89080.79820.79200.82490.46100.86480.76160.90790.83850.81680.94570.84600.8818T340.94200.96050.90100.93460.93960.84200.95120.94260.91840.97640.97350.98200.91660.9377T350.89470.70710.69080.83790.77250.70700.84920.88190.67200.78660.81400.99870.72200.7716T360.87940.77330.92380.98790.98650.52510.92010.93800.94010.96550.94180.89570.98750.8682
续表3
组合X1X2X3X4X5X6X7X8X9X10X11X12X13X14T370.89990.97470.79910.68250.73110.51100.80050.64890.94840.79420.88310.93120.75370.8588T380.86810.82510.96790.80540.88590.41080.89960.91660.85220.91580.91560.95980.93290.9123T390.84440.80410.95750.89500.97750.70790.94600.90920.99190.89400.99180.98320.83930.8416T400.76170.67200.75600.79970.79920.40860.92490.72370.74510.78570.90340.97810.80000.7846T410.91030.98350.89650.93900.84110.33340.84380.98800.91150.96030.97480.95690.96600.9690T420.88530.80380.92470.93380.87170.76720.84830.95470.85680.91710.81350.97630.95750.8984T430.88060.84100.84140.88830.97650.71780.96740.91150.92090.98190.96500.98150.95820.9811T440.80440.90380.91750.98140.93900.56050.96450.91700.99710.94250.92360.94650.91120.9885T450.98930.98270.78110.89070.90630.53240.85300.80320.82560.93200.88000.96150.98290.9998T460.78580.70840.71680.73270.82760.44170.77190.70830.87600.74650.89860.91510.79620.7941T470.80360.74030.95210.98860.88460.80800.91150.92940.94560.81060.83900.92490.77620.7906T480.84830.82520.84480.82460.86620.97830.78250.90140.92910.92090.96220.99820.97090.9292T490.86850.81610.70830.91120.88420.78910.89470.87080.95650.85830.89940.93990.91150.9088T500.76650.72070.95770.79490.92990.49420.97660.82500.82690.94200.85240.99850.86920.9421T510.72440.81800.78390.77220.85850.58340.91790.75730.71970.75120.81890.99790.78300.7436T520.69120.71360.72190.65540.73190.60560.68320.66720.96880.72600.73630.86390.68240.7463T530.73460.61370.73560.84960.80320.62310.85490.80680.80150.80810.90820.99800.85620.8538T540.79580.95550.90510.93330.94520.38550.97210.96050.84350.93440.83900.96530.85980.9477T550.71980.65500.78010.84990.77310.80360.81850.82620.83820.84980.71650.99800.80050.8573ck10.95180.74130.90250.95780.96650.52290.84590.91230.91710.98340.97510.99930.98140.9968ck20.97830.92860.87980.84550.94740.71400.94410.84800.96750.89560.93630.97710.86430.9240
玉米不同组合各农艺状与产量的灰色关联度结果见表4。按照灰色关联度大小依次为单株粒重>穗粗>籽粒含水量>百粒重>空杆率>穗行数>出籽率>穗长>倒折率>行粒数>茎粗>株高>穗位高>秃尖长。关联度大于0.90有1个性状,即单株粒重,其关联度为0.964 4,表明它与玉米产量关系最为密切,对产量的影响最大;关联度小于0.7也仅有1个性状,即秃尖长,其关联度为0.654 2,表明它与产量关系最远,对产量的影响最小。而其余性状的关联度则在0.825 9~0.877 67之间,表明这些性状均对产量有较大影响。因此,在西南区玉米育种中,应将单株粒重放在首位,统筹考虑其他性状,而秃尖长可适当调整。
表4 不同农艺性状与产量的关联度及排序
性状关联度排序株高0.840612穗位高0.825913茎粗0.846211穗长0.85648穗粗0.87672秃尖长0.654214穗行数0.86656性状关联度排序行粒数0.847210百粒重0.87214出籽率0.86187籽粒含水量0.87413单株粒重0.96441倒折率0.85069空杆率0.86935
为了进一步分析不同组合的优劣,本研究选取产量最高的组合1703为理想品种,对玉米不同新组合与理想品种的关联度进行了比较分析。依据灰色理论的分析原理,关联度越大则与理想品种越相似。从表5可以看出,不同组合与理想品种的关联度在0.795 1~0.922 0之间。其中,关联度大于0.9且排在前5位的组合,按照关联度从大到小依次为成单388、绵1308、川单1751、荣玉1756和渝单701,前4个组合的产量均排在前10位以内,而渝单701的关联度与产量位次差别明显,主要是由于渝单701的出籽率和空杆率与理想品种相差较大所致。而新组合川单1752虽然产量排名第1位,但关联度排名第17位,主要是由于穗行数与理想品种相差较大所致。新组合青青711的产量排第3位,关联度则排在第20位,主要是由于秃尖长与理想品种差异较大所致。关联度排在2个对照前面的新组合共有28个,与产量排序基本一致,有个别组合排序不同。与采用产量单一性状相比,关联度分析可较为全面地比较不同组合的优劣。
2.3 不同玉米新组合主要农艺性状的主成分分析
利用SPSS软件对玉米新组合产量相关的14个农艺性状变量进行主成分分析,特征值、贡献率和累积贡献率计算结果如表6所示。从特征值大小来看,前6个主成分的特征值均大于1,累积贡献率达76.18%。从单个主成分的贡献率来看,前4个主成分的贡献率均大于10%,其余主成分的贡献率依次降低。为了降低信息的损失量,本研究选取前9个主成分进行下一步分析,其累积贡献率达到了91.11%,足以反映数据的变化规律。
表5 不同组合与产量最高品种的关联度
组合品种名称关联度关联度排序产量排序T51成单3880.922012T40绵13080.913324T8川单17510.909937T28荣玉17560.908845T15渝单7010.9024530T48HZ17210.8987621T49恩9120.8942714T53青青7220.891189T38黔17020.8904915T19黔17050.89031011T22黔17010.8899118T34恩9320.88951231T55苏科玉16040.88601310T6渝单8360.87941412T36GD16760.87931523T20渝单8410.87851616T24川单17520.8771171T1217010.87621839T26黔17040.8754196T46青青7110.8739203T4恩11240.87182129T21苏科16140.87112217T9黔17030.87082333T23恩10100.87042413T43渝单7030.86642526T1817020.86562644T10荣玉17550.86442724T39靖2017-30.86082840组合品种名称关联度关联度排序产量排序ck1桂单162 0.86072928T31桂A11500.86043019T44HZ17200.86033132T41青青7330.85973225T30CS17-20.85473334T14南T6050.85463445T42S1830.85353518T13靖2017-10.85253649T5云瑞12070.85013727T45成单6580.85003822ck2渝单8号0.84833938T11恩10770.84604043T52渝单8960.84574154T33南T9310.84484246T16绵TX17010.83754347T54HZ17220.83654436T29云瑞5060.83574550T2绵732 0.82874635T17成17020.82734752T35德32390.82524853T32川单17530.82424920T7CS17-10.82225037T37南T9280.82195151T47苏科16180.81835248T1荣玉17540.81155341T50德100.81035442T27云瑞0070.79675555T3靖2017-20.79515656
表6 各主成分特征根与主成分贡献率
因子特征值贡献率/%累积贡献率/%12.77119.7919.7922.09314.9534.7431.92113.7248.4741.46310.4558.9251.369.7168.6361.0577.5576.1870.8185.8482.0280.6844.8986.9190.5894.2091.11
主成分特征向量如表7所示,第一主成分中贡献率为19.79%,其特征向量中影响力较大的因子包括单株粒重(0.821)、穗粗(0.713)和穗行数(0.665),这些性状可以作为代表性评价指标。第二主成分贡献率为14.95%,影响力最大的因子是百粒重(0.637),其可作为代表性评价指标。第三主成分贡献率为13.72%,株高(0.602)、穗位高(0.578)和茎粗(0.561)的影响力较大,可作为代表性评价指标。综上所述,从产量角度综合考虑,要想获得较高的产量,在新品种选育过程中应优先考虑单株粒重、穗粗、穗行数、百粒重和株高等性状。
表7 选取主成分的特征向量
性状Y1Y2Y3Y4Y5Y6Y7Y8Y9株高-0.0280.4230.602-0.4710.076-0.1560.072-0.0510.342穗位高-0.3020.4610.578-0.4260.174-0.0360.014-0.047-0.179茎粗-0.255-0.0430.5610.4840.375-0.006-0.1130.341-0.071穗长0.102-0.5830.472-0.037-0.2550.2510.194-0.3350.275穗粗0.7130.2350.2940.321-0.075-0.186-0.157-0.176-0.144秃尖长-0.0220.2350.1240.473-0.642-0.0860.4540.0980.035穗行数0.665-0.0450.1460.1310.021-0.645-0.044-0.158-0.047行粒数0.032-0.7140.450-0.056-0.0360.260-0.005-0.067-0.346百粒重0.1730.6370.0360.4330.0740.518-0.143-0.0770.181出籽率0.405-0.480-0.0130.0950.505-0.1070.2270.2980.327籽粒含水量0.5780.2030.204-0.322-0.2500.1670.2820.440-0.179单株粒重0.8210.0520.1940.0400.2230.336-0.076-0.075-0.019倒折率-0.5180.1360.1910.3630.421-0.0670.452-0.230-0.143空杆率-0.428-0.1990.4950.210-0.385-0.129-0.4090.1850.148
表8 不同玉米新组合综合评价结果
品种名称F值分组川单17511.3726Ⅰ青青7111.3288Ⅰ恩10101.2978Ⅰ成单3881.2353Ⅰ绵13081.2304Ⅰ青青7331.1726Ⅰ17031.0509Ⅰ川单17530.9057Ⅱ川单17520.8983ⅡS1830.7805Ⅱ桂A11500.7046Ⅱ荣玉17560.6462Ⅱ黔17010.6102Ⅱ黔17040.5967Ⅱ黔17030.5807Ⅱ黔17050.4089Ⅲ渝单7010.4043Ⅲ黔17020.3840Ⅲ渝单8360.3047Ⅲ品种名称F值分组恩9120.2501Ⅲ绵7320.2358ⅢHZ17200.1877Ⅲ青青7220.1666ⅢHZ17220.1619ⅢGD16760.0927Ⅲ恩9320.0344Ⅲ苏科1614-0.0390Ⅳ苏科玉1604-0.0414Ⅳ渝单841-0.1001Ⅳ恩1124-0.1425Ⅳ1701-0.1434Ⅳ南T605-0.1500Ⅳ桂单162-0.1512Ⅳ云瑞1207-0.1984ⅣHZ1721-0.2127Ⅳ成单658-0.2832Ⅳ荣玉1755-0.3123Ⅳ渝单703-0.3665Ⅳ品种名称F值分组恩1077-0.3933Ⅳ南T931-0.4376Ⅳ荣玉1754-0.4818Ⅳ云瑞007-0.5273ⅤCS17-2-0.5472Ⅴ德10-0.5787Ⅴ渝单8号-0.7125Ⅴ靖2017-3-0.7361Ⅴ德3239-0.7600Ⅴ1702-0.7974Ⅴ靖2017-1-0.8281Ⅴ靖2017-2-0.8793Ⅴ南T928-0.8795Ⅴ云瑞506-0.9124Ⅴ苏科1618-0.9749Ⅴ绵TX1701-0.9961Ⅴ渝单896-1.0775ⅥCS17-1-1.1825Ⅵ成1702-1.1990Ⅵ
为了更好的评价不同玉米新组合的优劣,根据表7主成分特征向量值,构建主成分与农艺性状之间的线性模型,即:
Y1=-0.028Z1-0.302Z2-0.255Z3+0.102Z4+0.713Z5-0.022Z6+0.665Z7+0.032Z8+0.173Z9+0.405Z10+0.578Z11+0.821Z12-0.518Z13-0.428Z14;
Y2=0.423Z1+0.461Z2-0.043Z3-0.583Z4+0.235Z5+0.235Z6-0.045Z7-0.714Z8+0.637Z9-0.480Z10+0.203Z11+0.052Z12+0.136Z13-0.199Z14;
……
Y9=0.342Z1-0.179Z2-0.071Z3+0.275Z4-0.144Z5+0.035Z6-0.047Z7-0.346Z8+0.181Z9+0.327Z10-0.179Z11-0.019Z12-0.143Z13+0.148Z14。
根据表6所提供的9个主成分与其对应的贡献率构建玉米新组合综合评价模型F,F为所选9个主成分的线性组成,即:
F=0.197 9Y1+0.149 5Y2+0.137 2Y3+0.104 5Y4+0.097 1Y5+0.075 5Y6+0.058 4Y7+0.048 9Y8+0.042 0Y9。根据该模型对55个参试组合与2个对照品种进行综合评价,结果如表8所示。根据综合评价结果,依据F值大小将所有参试的57份材料分为6组,Ⅰ组为F值≥1.000 0,包括川单1751、青青711、恩1010、成单388绵1308、青青733和1703等8个材料;Ⅱ组为1.000 0>F值≥0.500 0,包括川单1753、川单1752、S 183、桂A 1150和荣玉1756等8个材料;Ⅲ组为0.500 0>F值≥0.000 0,包括黔1705、渝单701、黔1702、渝单836等11个材料;Ⅳ组为0.000 0>F值≥-0.500 0,包括苏科1614、苏科玉1604、渝单841、恩1124和对照品种桂单162等15个材料;Ⅴ组为-0.500 0>F值≥-1.000 0,包括云瑞007、CS 17-2、德10、靖2017-3和对照品种渝单8号等13个材料;Ⅵ组为F值<-1.000 0,包括渝单896、CS 17-1和成1702共3个材料。
图1 参试材料系统聚类图
2.4 不同玉米新组合主要农艺性状的聚类分析
应用DPS软件对所有57个参试材料的14个农艺性状进行标准化处理后聚类分析,结果如图1所示。当距离为6.50时,可以将57份材料分为5类,其中,Ⅰ类包括绵1308、青青733、青青711、成单388、川单1751、渝单701和1703共7份材料;Ⅱ类包括桂A 1150、恩932、德3239、德10、HZ 1722、恩1077、黔1703和恩1010共7份材料;Ⅲ类包括靖2017-1、CS 17-2、CS 17-1、成1702、渝单703、苏科玉1604和渝单8号等共22份材料;Ⅳ类包括靖2017-2、云瑞007和云瑞506共3份材料;Ⅴ类包括荣玉1754、绵732、川单1752、黔1702、S 183、HZ 1720和桂单162(ck 1)等共17份材料。聚类分析法的分类结果与主成分分析法的分类结果基本一致。
3 小结与讨论
玉米的产量与植株多个农艺性状密切相关。前人针对不同类型玉米已开展了较多研究,结果各有不同。陈荣丽等研究发现,与甜玉米鲜穗产量密切相关的性状依次是穗粗、穗长、行粒数,而出苗、采收对产量的影响最小[3];李清超等研究认为,与玉米产量关联度较大的性状依次是单穗粒重、株高和穗位高,关联度最小的是穗长[7];马全姿等研究发现,与糯玉米鲜穗产量密切关联的性状是穗粗、穗行数和株高,秃尖长的影响最小[25];安治良认为,穗粒重、百粒重、穗行数是影响夏玉米产量的较大因素[26];陈灿等认为,穗粗、百粒重和穗行数对普通玉米的产量影响较大[27]。不同的玉米类型,从农艺性状的关联度大小来看,表现各有差异。本研究中发现,与玉米产量关联度最大的性状是单株粒重,其次为穗粗、籽粒含水量、百粒重和空杆率,关联度最低的是秃尖长。这与韩学坤等[9]的研究有一定的一致性,均认为单株粒重与产量的关联度最大,除病害性状外,秃尖长与产量的关联度最小。
在灰色关联度分析中,参考品种的选择对于参试材料的准确评价至关重要。在本研究中,所有参试材料按照产量高低排序,对照品种桂单162和渝单8号分别排在29位和39位。若用2个对照品种为参考品种,不能较好的评价其它组合的优劣,故选择了所有参试组合中产量最高的组合1703为参考品种。
目前,灰色关联度分析法、主成分分析法和聚类分析法已在农作物综合评价中广泛运用,取得了较好效果。但除这些方法外,综合评价的常见方法还有层次分析法[28]、隶属函数法[29]、同异分析法[30]等。由于不同的方法其算法与原理均有差别,所以针对同一份数据,分析获得的结果也有差异。此外,不论采用何种方法,评价结果均建立在专家打分的基础上,客观性均受到一定程度的影响。因此,为了获得更加客观公正的分析结果,可以同时采用多种方法对评价材料进行综合分析。
综合应用灰色关联度分析法、主成分分析法和聚类分析法,对55个玉米新组合进行了评价。灰色关联度分析法表明,以产量最高组合1703为参考序列,与其关联度最高的组合依次是成单388、绵1308、川单1751、荣玉1756和渝单701。依据主成分综合评价模型,得分值大于1的组合依次是川单1751、青青711、恩1010、成单388、绵1308、青青733和1703。聚类分析则表明绵1308、青青733、青青711、成单388、川单1751、渝单701和1703优先聚为一类。综上可知,玉米新组合川单1751、成单388、绵1308和1703表现最优。