Results and Discussion

Housing and industrial layouts were digitized and considered as the future growth poles to simulate the possible land use transitions. The net neighbourhood influences were determined by the 5 × 5 contiguity filter which explains past land use changes and used to simulate future changes. Land use predictions of 2015 were made using CA coupled with Markov chains based on the transitional probability generated for 1999–2009. The validation of predicted land use was done by comparing it with reference land use of 2015 (actual) through the computation of kappa index (for location and quantity). This helped in the visualization of likely land uses in 2027, considering 6-year time intervals.Table 1. Land Use Classes Considered for the Analysis

Status and Transition of Land Use Land Cover from 1973 to 2015

LULC dynamics is assessed using temporal RS data. The study region (BNP with a 5-km buffer) has witnessed large- scale land cover transformation from 1973 to 2015, due to unplanned developmental activities with large-scale anthropogenic activities. The fast economic and social transformation in the Bangalore metropolitan region has large-scale impacts on BNP and its environs. Temporal variations in NDVI values helped to quantify spectrally distinct vegetation and non-vegetation regions, which highlighted the loss of vegetation cover from 85.78 per cent (1973) to 66.37 per cent (2015) in the study region (Table 2, Figure 4). Land use analyses from 1973 to 2015 reflect the impact of urban expansion and the status of forests in conservation, supports wildlife migration and acts as a refugee and a grazing ground for livestock. The western and eastern part of the study region (BNP with the buffer) consist of human habitations with agricultural activities. Figure 5 and Table 3 reflect changes in the forest cover of the BNP buffer region from 1973 to 2015. The large tracts of deciduous cover in the Kanakapura Taluk and Anekal Taluk have disappeared due to the unauthorized expansion of horticulture and agricultural activities in recent years. The region has lost moist deciduous cover, from 26.1 per cent to 13.8 per cent (1973–2015), and horticulture has increased, from 8.5 to 11 per cent. Large-scale land use changes with an increase in built-up areas from 0.4 per cent to 4.5 per cent in the periphery are due to illegal activities such as stone quarrying, granite and sand mining and medium-scale industries. The overall accuracy of land use maps for 1973, 1989, 1999, 2009 and 2015 was 87.86 percent, 88.4 per cent, 88.8 per cent, 87.85 per cent and 91.66 per cent, respectively. Accordingly, the kappa coefficients were 0.84, 0.86, 0.86, 0.85 and 0.89, respectively.

Bangalore’s unrealistic metropolitan growth has intrinsic relations with vegetation loss in the peri-urban landscape of BNP. The increase in the population of Kanakapura Taluk and Maralwadi Town resulted in more transition of forests to other land uses. Encroachments in Kanakapura forests, revenue lands and gomala regions (grazing lands) have resulted in deforestation, and the formation of housing layouts has enhanced the instances of human-animal conflicts and loss of human life and crop. The region has 9,254.21 ha of degraded forest patches, which have to be reforested with native plant species to improve the food and fodder availability for wild fauna.

Table 2. Land Cover Changes from 1973 to 2015

Year	1973	1989		1999		2009		2015
Category	Ha	%	Ha	%	Ha	%	Ha	%	Ha	%
Vegetation	107087.8	87.98	104039.5	85.5	99992.2	82.2	94693.3	77.8	89419.0	73.46
Non-vegetation	14633.13	12.02	17682.5	14.5	21728.7	17.8	27027.6	22.2	32301.9	26.54
Total area					121720.9

Note: Highlighted values indicate significant changes.

Figure 4. Land Cover from 1973 to 2015 in Bannerghatta National Park with 5-km Buffer

Figure 5. Land Use in BNP with 5-km Buffer (1973–2015)

Table 3. Land Use Changes in Bannerghatta National Park with 5-km Buffer from 1973 to 2015

Year Category

	Ha	%	Ha	%	Ha	%	Ha	%	Ha	%
Dry deciduous forests	32415.6	26.6	31609.3	26.0	26433.6	21.7	24042.3	19.8	22729.9	18.7
Moist deciduous forests	31725.9	26.1	29694.3	24.4	25941.5	21.3	16916.1	13.9	16822.2	13.8
Grass/scrub forests	4876.8	4.0	4500.4	3.7	4914.4	4.0	8967.2	7.4	9254.2	7.6
Water	377.8	0.3	922.5	0.8	1170.8	1.0	1251.6	1.0	1834.0	1.5
Horticulture	10361.3	8.5	11768.4	9.7	12707.1	10.4	13470.5	11.1	13363.2	11.0
Agriculture	36027.9	29.6	37222.3	30.6	40053.5	32.9	44975.1	36.9	45613.8	37.5
Urban	490.6	0.4	581.6	0.5	1934.9	1.6	3216.0	2.6	5462.1	4.5
Barren land	4461.8	3.7	4251.7	3.5	6833.9	5.6	6114.5	5.0	3536.6	2.9
Forest plantations	921.4	0.8	1060.5	0.9	1011.0	0.8	1533.1	1.3	1566.3	1.3
Mining area	61.8	0.1	109.9	0.1	720.3	0.6	1234.5	1.0	1538.7	1.3

Modelling and Visualization of Forest Transitions

The prediction of likely land uses for 2021 and 2017 was done through CA-Markov, incorporating various land use decisions in transition rules (of 1999–2009 and 2009– 2015). Land use of 2015 is predicted (Figure 6) considering land use of 1999 (Table 4, with transition details of 1999– 2009), which was compared with actual land uses of 2015. This helped in validating the CA-Markov technique for predication, and the results are given in Table 5. A higher kappa value of 0.86 indicates a significant correlation and agreement between the simulated and actual land uses. Kno indicates the overall accuracy of the simulation as compared to the reference map. Klocation evaluates the accuracy of the simulation in specifying a particular location. Kstandard shows the location error with quantification as per reference map. The prediction of land uses for 2021 was done considering base land use data from 2009 to 2015. Similarly, the land use of 2027 was predicted based on land uses of 2015 and 2021 (simulated). The projected land use of 2027 shows an alarming picture of the loss of forest cover, from 41.38 per cent to 35.59 per cent, with an increase in urban area (4.49%–9.62%) due to irresponsible land use changes with housing projects in an ecologically sensitive region (Figure 7). These anthropogenic activities would pose serious threats to the forest ecosystem in BNP, which has been aiding as a carbon sink (in lieu of higher greenhouse gas (GHG) emissions in Bangalore (Ramachandra, Bharath et al., 2015). The forests in southern parts project minimal disturbances (connected to the Tali Reserve Forests and Cauvery Wildlife Sanctuary), whereas northern portions show very disturbing trends of higher rates of transition. The uncontrolled and unplanned growth of Greater Bangalore would certainly spell doom to the survival of fauna and the sustenance of forest cover in BNP. The major growth poles are built-up expansions in the Anekal industrial area, Kalkere, Basavanapura and Weavers Colony, Uttarahalli Manavartha Kaval.

Figure 6. Simulate and Projected Land Use of BNP 2015–2027

Table 4. Transition Probability Matrix Based on Land Use from 1999 to 2009

			Probability of Changing to
Given	Forest	Agriculture	Barren Land	Water	Built-up
Forest	0.783	0.1649	0.0252	0.0022	0.0246
Agriculture	0.025	0.9	0.025	0.025	0.025
Barren land	0.1573	0.259	0.4493	0.0026	0.1318
Water	0.1055	0.0803	0.0165	0.7011	0.0967
Built-up	0.025	0.025	0.025	0.025	0.9
Source: The authors.

Table 5. Simulated, Projected Land Use of Bannerghatta National Park from 2015 to 2027 and Accuracy of the Analysis

Year Simulated (2015) Projected (2021) Projected (2027)

Categories	Ha	%	Ha	%	Ha	%
Forest	49444.88	40.62	45895.57	37.67	43315.05	35.59
Agriculture	58207.18	48.64	60410.02	49.67	60206.03	49.46
Barren land	6668.57	4.66	5030.71	4.13	4460.76	3.66
Water	2011.68	1.65	2033.13	1.67	2031.11	1.67
Urban	5387.75	4.43	8350.63	6.85	11707.15	9.62
Total area			121720.9
Index	Validation of Simulated LU 2015
Kno	0.93
Klocation	0.86
Kstandard	0.87
Source: The authors.

Figure 7. The Regions Likely to Experience Higher Land Use Changes

The global population growth is projected to increase from 7.2 billion to 9.6 billion from 2013 to 2050 (Wimberly, Sohl, Liu, & Lamsal, 2015), which would escalate anthropogenic pressure on forest resources. Global forest cover destruction has resulted in augmented GHG emissions from 12 per cent to 17 per cent (Van der Werf et al., 2009) due to shunting of the carbon sequestration process. Deforestation in Southeast Asia alone has resulted in ∼1.4 ± 0.5 PgC yr−1 net emissions of Carbon (C) between 1990 and 2010 (Houghton et al., 2012). Deforestation and subsequent land use changes will have irretrievable impacts on ecosystem goods and services apart from the decline of carbon sequestration capabilities and an increase in atmospheric GHG emissions. This would also alter the climate and hydrologic regimes that would in turn threaten water and food security. Large-scale LULC changes and loss of forest cover in the study region (BNP with a 5-km buffer) are due to urban sprawl with irresponsible and unplanned urbanization processes in Bangalore. The city witnessed urbanization with the unrealistic increase of paved surfaces by 1,028 per cent from 1973 to 2018 with loss of vegetation (88%) and 79 per cent of water bodies (Halmy & Gessler, 2015; Ramachandra & Bharath, 2016; Ramachandra, Setturu et al., 2012; Ramachandra, Bharath et al., 2012). Uncontrolled and uncoordinated urban growth was noticed subsequent to globalization and the push towards industrialization in recent decades. Drivers of urbanization include political and economic in various proportions. Urban sprawl already has telling impacts evident from resource scarcity and enhanced instances of human-animal conflicts, which necessitate the restoration of the integrity of ecologically sensitive buffer regions with the afforestation of degraded forest patches with native species of flora and immediate eviction of unauthorized occupations of forests and common lands (grazing lands, etc.). Various spatially explicit models have been developed in recent years to simulate transitions in LULC, thereby enabling the advanced visualization of dynamic phenomena such as urban growth. The CA-Markov model adopted for the visualization of LULC changes highlights the likely ecological and environmental implications in the absence of appropriate policy interventions for the conservation of BNP.

Visualization through CA-Markov (Briassoulis, 2000; Keshtkar & Voigt, 2016) was comparable to actual growth (2015). Predicted LU changes for 2021 and 2027 have provided valuable insights into landscape dynamics and likely implications. Non-agent-based (AGB) models however have drawbacks of the non-inclusion of agents and in particular human decision-making (Briassoulis, 2000; Keshtkar & Voigt, 2016; Truong et al., 2015). Compared to this, AGB dynamics are well conceived for conditional decision-making of nonlinear behaviour by

rules that distinguish them from mathematically continuous models such as CA-Markov (Brown, Riolo, Robinson, North, & Rand, 2005; Li, Oyana, & Mukwaya, 2016). AGB involves the identification of agents which have the ability to satisfy internal goals through actions based on a set of rules of a temporal framework within which those agents perform actions but computationally intensive ones (Daniel, Frid, Sleeter, & Fortin, 2016; Mozumder, Tripathi, & Losiri, 2016; Nicholls, Amelung, & Student, 2016; Rand et al., 2003). The multi-agent models with human perceptions, which encapsulate hierarchically the behaviours of biophysical drivers, will increase the precision of prediction.

BNP, a part of Western Ghats and a repository of unique flora and fauna with biological, social, hydrological and ecological significance, needs immediate measures with prudent biodiversity conservation policies. Due to water and food security in peninsular India with perennial water resources, the Western Ghats is aptly known as the water tower for peninsular India. Any imbalances in the ecologically sensitive regions would not only affect the local population but also threaten global climate. The community-based conservation (CBC) path involving local stakeholders is crucial to conserving biological diversity and sustaining natural resources. The involvement of local stakeholders in decision-making and enhancing the livelihood prospects of the dependent population would help in the protection of forest ecosystems. The local community’s knowledge and the experience of wildlife and their habitats would help in strengthening conservation endeavours. Ecologically hazardous activities such as mining, the expansion of agriculture and horticulture in the core area must be restricted immediately to protect wildlife and flora. Incentives to support organic farming, setting up agro-processing industries, establishing cottage industries to support local livelihoods and setting up fodder farms to support local livestock population would help in minimizing forest degradation. Steps have to be taken to enrich forests impoverished of wild animal fodder plants. Appropriate cropping has to be encouraged with strict regulations to minimize instances of human-animal conflicts (Radha Devi, 2003). Other conservation and awareness initiatives include involving education institutions to document biodiversity in the neighbourhood (at the village level); eco clubs at all schools; students to take part in environment monitoring (part of the curriculum), the development of forest nurseries of local species through the active participation of women and incentives to villagers for conservation and so on. Adopting the integrated clustering development of villages for inclusive growth is suggested to promote eco-friendly, local resources, local skills and manpower-based thematic developmental programmes through a stronger foundation for sustainable growth (Ramachandra, Hegde, Subash Chandran, Tejaswini, & Vishnumayananda, 2015).