Power line collision risk in bustards 445


TABLE 1 Description and summary statistics for the predictor variables used to assess the drivers of power line collision risk for the great bustard Otis tarda and little bustard Tetrax tetrax. Mean ± SD and range are provided for continuous variables; frequency per class is presented for categorical variables (n = 144 sampled transects).


Variable


Open_Habitat D_Habitat


Configuration


Marking Effort


Description


Proportion of open farmland habitat in a 5 km buffer around the power line section (DGT, 2007)


Dominant habitat in a 1 km buffer around the power line section (DGT, 2007): open farmland (land-cover categories according to DGT, 2007: 2.1.1, 2.1.2, 2.3.1), forest (3.1.1, 3.1.2, 3.1.3, 3.2.4) & agro-forestry (2.4.4)


Power line configuration (Supplementary Fig. 1): small, hori- zontal 150 kV; medium, horizontal 400 kV; large, vertical 150 or 400 kV


Wire marker devices (presence/absence)


Total accumulated surveyed distance (km) in all sampling visits (length of power line section × minimum number of samples)


a 2 km wide buffer strip around each power line section (1 km either side of the power line) as a surrogate of habitat availability in the vicinity of the line. Additionally, we mea- sured the proportion of open farmland in a 10 km buffer surrounding each power line section (5 km either side) as an indicator of the availability of potentially suitable habi- tat in the area. We categorized power line sections accord- ing to three main configurations: (1) small configuration (low pylon height) and (2)medium configuration (medium pylon heights), both with conductor wires displaced hori- zontally and wires at two levels, and (3) large configuration, with conductor wires displaced vertically and wires at four levels, higher pylons, and larger distance between top and bottom wires (Supplementary Fig. 1). All three power line configurations have two earth wires above the conductors. We used a presence/absence variable for anti-collision de- vices (of any type) in each 2 km section. Finally, we included an indicator of survey effort (accumulated surveyed distance in all sampling visits; Table 1), to account for potential sur- vey bias amongst power line sections. We used Spearman’s correlation coefficient and variance


inflation factors to check for collinearity between explana- tory variables (Zuur et al., 2009). Variance inflation factor values (all,1.6) and pairwise correlation between explana- tory variables (all |r| ,0.60) were low and therefore we used all variables in the analysis. We assessed the overall effect and relative importance


of each explanatory variable with boosted regression trees (De’Ath, 2007), using the dismo package (Hijmans et al., 2016)in R 3.3.1 (R Core Team, 2016), following the recom- mendations of Elith et al. (2008). Boosted regression trees is a non-parametric machine-learning method that fits a large number of simple classification or regression trees (models that relate a response to their predictors by recursive binary splits), with predictions combined to give robust response estimates (De’ath & Fabricius, 2000; Elith et al., 2008).


Mean ± SD/Frequency 0.43 ± 0.19


Open: 81 Forest: 36 Agroforestry: 27


Small: 77 Medium: 36 Large: 29


Absent: 90 Present: 54 48.3 ± 50.5


4.6–265.5 Range 0.14–0.87


The advantages of boosted regression trees include accom- modation of missing values, ability to use both continuous and categorical predictor variables, immunity to the effects of extreme outliers, and the facility to fit interactions be- tween predictors (Leathwick et al., 2006). We generated one boosted regression trees model for


each species, using presence/absence of mortality on power line sections as the response variable. As input parame- ters we used the Bernoulli family, a tree complexity of three (i.e. the complexity of variable interactions that may be fitted), a 0.0005 learning rate (the weight applied to indi- vidual trees) and a bag fraction of 0.8 (at each iteration, 80% of the data were drawn at random). Each model was built with a default 10-fold cross-validation (using the function gbm.step). When fitting initial models (Supplementary Fig. 2), some of the fitted functions generated had varying shapes without underlying ecological meaning. For exam- ple, the fitted function for the effects of the proportion of open habitat on the great bustard was variable for a large range of values. Additionally, it seemed that collision risk for this species was higher in the presence of wire marking. Such patterns often result from combinations or interactions of variables in specific geographical contexts that, in spite of having no ecological meaning, contribute to increased model fit (Leathwick et al., 2006; Elith et al., 2008). For example, the higher likelihood of great bustard collisions in the presence of wire marking is probably a re- sult of the fact that most power lines that cross important areas for this species during the breeding season have been marked. Therefore, following Leathwick et al. (2006), we refitted the models by imposing monotonically increas- ing (for the proportion of open habitat and survey effort) or decreasing (for wire marking) constrained functions for some variables. Imposing monotonic trends reduces the total amount of deviance explained by the models but also decreases the likelihood of overfitting (Leathwick et al.,


Oryx, 2021, 55(3), 442–451 © The Author(s), 2020. Published by Cambridge University Press on behalf of Fauna & Flora International doi:10.1017/S0030605319000292


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48  |  Page 49  |  Page 50  |  Page 51  |  Page 52  |  Page 53  |  Page 54  |  Page 55  |  Page 56  |  Page 57  |  Page 58  |  Page 59  |  Page 60  |  Page 61  |  Page 62  |  Page 63  |  Page 64  |  Page 65  |  Page 66  |  Page 67  |  Page 68  |  Page 69  |  Page 70  |  Page 71  |  Page 72  |  Page 73  |  Page 74  |  Page 75  |  Page 76  |  Page 77  |  Page 78  |  Page 79  |  Page 80  |  Page 81  |  Page 82  |  Page 83  |  Page 84  |  Page 85  |  Page 86  |  Page 87  |  Page 88  |  Page 89  |  Page 90  |  Page 91  |  Page 92  |  Page 93  |  Page 94  |  Page 95  |  Page 96  |  Page 97  |  Page 98  |  Page 99  |  Page 100  |  Page 101  |  Page 102  |  Page 103  |  Page 104  |  Page 105  |  Page 106  |  Page 107  |  Page 108  |  Page 109  |  Page 110  |  Page 111  |  Page 112  |  Page 113  |  Page 114  |  Page 115  |  Page 116  |  Page 117  |  Page 118  |  Page 119  |  Page 120  |  Page 121  |  Page 122  |  Page 123  |  Page 124  |  Page 125  |  Page 126  |  Page 127  |  Page 128  |  Page 129  |  Page 130  |  Page 131  |  Page 132  |  Page 133  |  Page 134  |  Page 135  |  Page 136  |  Page 137  |  Page 138  |  Page 139  |  Page 140  |  Page 141  |  Page 142  |  Page 143  |  Page 144  |  Page 145  |  Page 146  |  Page 147  |  Page 148  |  Page 149  |  Page 150  |  Page 151  |  Page 152  |  Page 153  |  Page 154  |  Page 155  |  Page 156  |  Page 157  |  Page 158  |  Page 159  |  Page 160  |  Page 161  |  Page 162  |  Page 163  |  Page 164