松嫩平原常见鹤类足迹特征及足迹信息学分析

Due to fluctuations in environmental quality,crane populations in China show different fluctuating trends.As one of the three major plains in northeast China,the Songnen Plain has been of great interest as a major area of wetland distribution and an important stopover for crane migration in China.The crane footprint is one of the elements that have received attention and is directly related to biological information such as crane species identification,population size,and habitat use.With the advent of the post-genomic era,the use of machine learning to mine biological data has become a hot spot in biological research.In this study,we hypothesized that crane footprints could be used as a basis for interspecific discrimination of cranes,and on this premise,we combined traditional morphometry and support vector machine(SVM)to explore the feasibility and accuracy of this discrimination.Five species of cranes distributed in the Songnen Plain were used as the study objects(Red-crowned Crane,White Crane,White-naped Crane,Hooded Crane and Common Crane),using the line-intercept method,sited-fixed observation method,synchronous observation method,and field measurement method were used to analyze the footprint characteristics,interspecific differences,adult and juvenile differences,seasonal differences,and substrate effects of the five common crane species,and the results are as follows. (1)The interspecific differences in crane tracks were significant.White-naped Crane,adult Red-crowned Crane and adult White Crane were more similar in terms of footprint length characteristics;White Crane juveniles,Red-crowned Crane juveniles,and mixed populations of Hooded Crane and Common Crane were less similar in terms of footprint length characteristics.The angle between the toes of adult Red-crowned Crane and adult White Crane was large,and the angle between the toes of White-naped Crane was small.The support vector machine algorithm for crane footprint discrimination based on principal component analysis verified the feasibility of footprint as a discriminator of crane species. (2)There were significant seasonal differences in crane footprints,and the length and angle eigenvalues of footprints in the autumn migration period were significantly lower than those in the spring migration period. (3)There are obvious differences between adult and juvenile crane tracks.The footprint lengths of adult Red-crowned Crane and White Crane were significantly larger than those of juveniles,and the footprint lengths fluctuated greatly.The difference between adults and juveniles in the angle between the toes of White Crane was not obvious,while the adults in the angle between the toes of Red-crowned Crane was significantly larger than the juveniles. (4)The footprint characteristics in different substrates differed significantly.The differences in footprint characteristics among meadow,saline and cornfield were highly significant,with the footprint length,footprint width,II toe length,III toe length and IV toe length being saline>cornfield>meadow,and the inter-toe angle being meadow>saline>cornfield.The trail feature values in the dirt road and rice field only differed significantly from some of the bearing traces. Further analysis showed that the crane footprint discrimination algorithm based on support vector machine with principal component analysis could be used for crane species discrimination;footprints in soft or wet areas would show I-toe,footprint angles in maize fields were mostly asymmetrical,and footprints in maize field rhizomes,meadow water sources,and next to meadow plants had small and deeper sunken single and compound step lengths;footprints could be used for preliminary prediction of population size,spatial distribution patterns of mixed populations of Hooded Crane and Common Crane.The tracks can be used to predict the population size,spatial distribution pattern and population dynamics of the mixed flock of White-headed Crane and Gray Crane.In this study,we investigated the multivariate differences of crane tracks and the influence of substrate on tracks,and established a support vector machine discriminant algorithm based on principal component analysis for crane species discrimination,which provides a new solution for crane population monitoring.