The average landholding of sample HH was 14.12 kattha (SD = 13.28, 79.09 kattha = 1 ha). The sample HH in MB, MBBZ and TBBZ had an average landholding of 16.18 kattha, 14.55 kattha and 9.36 kattha respectively. The differences of average landholding of the sample HH between 3 clusters, as shown in Fig. 3, was not significant (p = 0.19). The average landholding of sample HH within buffer zone and outside buffer zone, 12.94 kattha and 16.18 kattha respectively, were also not significantly different (p = 0.24).
The annual crop production was calculated from the average of last three years’ crop harvest. The sum of the crop production gave a calculated aggregated crop yield of 4.85 ton per ha. The calculated aggregated crop yield within study area was significantly higher than the annual aggregated crop yield in Bardiya district in 2019/20 (4.15 ton per ha) (MOALD 2020). The average annual crop production per HH in MB, MBBZ, and TBBZ clusters were 897.91 kg, 1011.00 kg, and 531.06 kg with an average annual yield of 4.39 ton/ha, 5.50 ton/ha and 4.49 ton/ha respectively. The average annual crop production per household within buffer zone was 862.052 kg and outside buffer zone was 897.91 kg with an average annual yield of 5.26 ton/ha and 4.39 ton/ha respectively. There was no significant difference in average annual yield per HH between HHs within three clusters (as shown in Fig. 4). Likewise, difference in the average landholding and the average annual yield in HHs inside and outside the buffer zone also wasn’t significant.
The major crops were paddy, maize, wheat, potato, and vegetables and paddy was the major crop for 96% of sample HHs. Paddy also was the most depredated crop from the elephant raid. Thirty percent of sample HHs experienced the crop damage on the field only, more than one third of sample HHs experienced the stored crop damage only, 6% experienced no damage whereas, and more than a quarter experienced both types of damages.
In MB, MBBZ, and TBBZ clusters, the total annual crop damage was 45.01%, 18.67% and 24.33% of the average annual production per HH. The highest total annual crop damage of 404.15 kg per HH was suffered by HHs in MB cluster, followed by MBBZ (188.76 kg/HH), and TBBZ (129.18 kg/HH) clusters. Out of the total annual crop damage in MB cluster, 80.45% was on the field which was significantly higher than MBBZ (12.17%) and TBBZ (72.05%).
The highest stored crop damage, as shown in Fig. 5, was suffered by HHs in MBBZ cluster (165.79 kg/HH) followed by MB (80.093kg/HH) and TBBZ (38.24 kg/HH) clusters (Fig. 6). Similarly, the average annual crop damage on the field was highest in MB (325.14 kg/HH), as shown in Fig. 7, followed by TBBZ (93.07 kg/HH), and MBBZ (22.97 kg/HH) clusters (Fig. 8). The annual crop damage on the field in MB cluster was 1.59 ton/ha that was significantly higher than the stored crop damage (p < 0.05).
The differences observed in average crop damages between different clusters and locations with respect to buffer zones were statistically significant. HHs outside buffer zone suffered significantly higher crop damage on field (325.144 kg) than HHs inside buffer zone (p < 0.05). Even though the stored crop damage in MBBZ cluster was significantly higher than in TBBZ cluster, the difference in the stored crop damage between HHs inside buffer zone and outside buffer zone was not significant (p = 0.16).
The major spatial characteristics of sample clusters were defined by the location with respect to the buffer zone and the distance from spatial features i.e., the BNP boundary, the KWS boundary, the forest frontage, and the river edge. All the sample clusters were spatially located between the BNP boundary and the KWS boundary as shown in Fig. 2. Conflict sites, agricultural fields or houses that were attacked by elephants, were either adjacent to or within 50m from the respective sample HH. Conflict sites within the buffer zone were significantly closer to the BNP boundary with an average distance of 1411.20m and significantly farther from the KWS boundary with an average distance of 6556.76m than conflict sites outside the buffer zone (4437.00m from the BNP boundary and 2814.89m from the KWS boundary). Conflict sites outside the buffer zone were significantly closer to the forest frontage (145.19m average nearest distance) and the river edge (494.48m average nearest distance) than those inside the buffer zone.
MB cluster:
Sample HHs and conflict sites were located in between Geruwa river and Ganesh Sisinuya Community Forest (CF). The average closest distance from the conflict site to the river edge of Geruwa river was 494.04m and to the forest frontage of Ganesh Sisinuya CF was 145.19m. The average closest distance from the conflict site to the KWS boundary (India) was 2814.89m and was the closest cluster to the KWS boundary (India), whereas it was 4437.00m to the BNP boundary.
MBBZ cluster:
Lying in between MB and TBBZ clusters and in the proximity of Durga CF, two prominent river systems; Geruwa river and Aurahi river flows from either side of the MBBZ cluster. The average closest distance from the conflict site to the river edge of Geruwa river was 1073.55m, to Aurahi river was 657.48m, and to the forest frontage of Durga CF was 196.23m. Likewise, the average closest distance from the conflict site to the KWS boundary (India) was 5736.36m and to the BNP boundary was 1914.35m.
TBBZ cluster:
Sample HHs and conflict sites in TBBZ cluster were the closest among three clusters to the river edge. The average closest distance from the conflict site to the river edge of Geruwa river was 443.97m and to the nearest forest frontage was 505.18m. Likewise, the average distance from the conflict site to the KWS boundary (India) was 8379.90m, whereas to the BNP boundary was 293.19m.
Conflict sites in TBBZ cluster were significantly farthest among 3 clusters from the forest frontage whereas conflict sites in MBBZ cluster were significantly the farthest among 3 from the river edge.
The contrasting effect was shown by the distance to the spatial features on the quantified damage suffered by the respective HH. The moderate positive correlation (p < 0.01, cor = 0.49) was shown by the distance to the BNP boundary (Fig. 9) whereas, the moderate negative correlation (p < 0.01, cor = -0.45) was shown by the distance to the KWS boundary (India) with the field damage suffered by the respective HH (Fig. 10).
The negligible negative correlation was shown by the distance to the river edge with the crop damage on the field (Fig. 11) and the negligible positive correlation with the stored crop damage (Fig. 12). Similarly, negligible significant correlation was identified in between the distance to the forest frontage with both types of crop damages suffered by the respective HH as shown in Figs. 13 and 14 (n = 100). However, within the MB cluster, the stored crop damage was moderately correlated to the distance to the forest frontage (cor = 0.40, n = 42).
The number of elephants per raid ranged from 1 to 8 individuals (mean = 3, SD = 1.53). The number of elephants per raid significantly varied between clusters. MB cluster generally was raided by the herd of 3 to 8 elephants (average of 4 individuals) whereas the MBBZ and TBBZ clusters were raided by 1–3 elephants (average of 2 individuals). The average number of elephants per raid in MB cluster was significantly higher than the average number of elephants per raid in MBBZ and TBBZ clusters. The number of elephants per raid was significantly correlated to the distance to the BNP boundary and to the KWS boundary (India). A strong positive correlation in between the number of elephants per raid with the distance to the BNP boundary (p < 0.001, cor = 0.74) was seen. Contrastingly, a strong negative correlation with the distance to the KWS boundary (India) (p < 0.01, cor = -0.72) was seen. The total annual crop damage was positively correlated to the number of elephants per raid (p < 0.01, cor = 0.47) (Fig. 15). The number of elephants per raid and the total annual crop damage was on a decreasing trend along the Khata Corridor from the Nepal – India border towards the BNP boundary as shown in Fig. 16.
A moderately strong correlation was noted in between the number of elephants per raid and the severity of damage on the field. However, no significant correlation was exhibited by the number of elephants per raid with the severity of stored crop damage.