**5. Discussion**

The present study was conducted in order to quantitatively analyze the landscape change dynamics in the ICT from 1976 to 2016 using medium spatial resolution Landsat images. In this study, we observed an increase in settlements over the last forty years. The dramatic land cover change in settlements is exerting severe pressure on other land cover classes, particularly tree cover and soil. Already existing urban areas can be seen expanding through rapid construction of residential blocks in the form of housing societies, industrial blocks and road expansions, leading to horizontal and vertical developments in the city and development of lavish farm houses in the vicinity areas (Figure 7). This rapid increase in urbanization is linked also with migrations. Urban growth may have positive or negative impacts on the environment but unplanned growth of urban areas has negative e ffects. For the economic development of the country, necessary planning is required to make urbanization helpful, as social, health, and environmental issues often accompany the process of urbanization.

In the time span of forty years, an overall decrease has been observed in the area of tree cover classes (i.e., greater than and less than 40% tree canopy) in the ICT, and most of the tree loss has occurred after the year 2000 with a corresponding increase in built-up areas. Apart from tree cutting, another important reason behind rapid tree loss or degradation is forest fires. In the Margalla Hills, forest fires usually take place during dry hot climate conditions when there is no rain for months and temperature goes up to 45 ◦C. According to Khalid and Ahmad [78], a total of 320 forest fires were recorded from 2002 to 2012 and approximately 8 km<sup>2</sup> area go<sup>t</sup> burnt as a result. In the ICT, due to uncontrolled urbanization and lack of awareness, huge tree loss has been observed in the last sixteen years, i.e., 2000–2016. Another factor for forest degradation is uncontrolled grazing of livestock [78]. As such, there are no adequate plans or manageable methods to stop grazing activities. For ecotourism and public awareness drives, a number of jogging and hiking trails have been formed in Margalla Hills National Park, and a large number of visitors has severely a ffected the Margalla Hills National Park by dumping waste. These illegal and unmonitored activities in the forested area cause threats to the forest ecosystem [50,79].

Massive migrations have occurred over the past few years from rural to urban areas, mostly due to low cultivated land output, landlessness, sub-division of land, poor economy, and better educational and health opportunities in urban areas. The rapid increase in population has contributed towards natural resource depletion and rapid deforestation close to settlements [80].

The LCRPGR parameter is an indicator of urban sustainable development, whether urban expansion is in balance with population growth or not. According to literature review, limited scientific peer reviewed studies have been reported on the monitoring and mapping of LCRPGR. Under each SDG, a number of targets and indicators have been defined, which countries have to quantify, but most of the developing countries do not have comprehensive databases through which they can compute, quantify, and report the SDGs indicators. To the best of the authors' knowledge, only Nicolau et al. [77] and Wang et al. [81] have computed and reported scientific results of SDG proposed LCRPGR over urban areas. Nicolau et al. [77] based their study over the mainland of Portugal, while Wang et al. [81] carried out their study over mainland China, using earth observation and population census data, and reported the increase of LCRPGR value from 1.69 in 1990–2000 to 1.78 in 2000–2010. The LCRPGR related research findings from these studies show that in most cities, both horizontal and vertical urban expansions are carried out in an unplanned manner which has already e ffected the equilibrium of land consumption versus population increase to attain e ffective development goals by 2030 [77,81]. In this study over the ICT, the LCRPGR ratio was 0.62 from 1976 to 2000, which increased to 1.36 from 2000 to 2016. Based on studies conducted on the global scale, in the most of the cases LCR is higher than or equal to PGR due to high demand of luxurious occupancies in the urban areas [26,30].

**Figure 7.** Temporal Landsat satellite images of 1976, 2000, and 2016 in standard False Color Composite (FCC), illustrating an overall increase in built-up areas over the years in the ICT. The area of the ICT shown here represents the bounded box inset covering Zone I and Zone II partially in Figure 1.

The Shenzhen SEZ and ICT cities were both developed in late 1970s. In terms of urban forest cover change, Shenzhen SEZ had been restored to ~85% (1973–2005) [29] while in this study, we detected urban forest loss of ~27% (1976–2016) in the ICT. In both cities (SEZ and ICT), urban forest fragmentation results revealed losses in forest patches. In South Asia, the landscape of the ICT resembles strongly with the cities of Kathmandu (the capital of Nepal) and Thimphu (the capital of Bhutan), as they are all cities topographically surrounded by the pine trees. As compared to ICT, Kathmandu and Thimphu cities are older but highly migrant resident populated, vastly encroached, and over grazed. Although many studies have investigated the land cover dynamics of Kathmandu

and Thimphu cities using temporal satellite data [82–85], detailed analysis using parameters such as landscape metrics, forest fragmentation, and LCRPGR has not been performed. In this aspect, this study serves as a methodological framework for application and analysis over other similar cities in developing countries.

Due to rapid tree loss and urbanization, the ICT has observed rapid spells of dust storms and soil erosion [86]. Soil erosion, dry temperatures, and consequent dust storms have a negative impact in the form of air pollution [87] and land degradation. The soil in ICT and the surrounding areas is shallow and has a clay composition [63]. The alluvial lands and terraces in the area tend to have low agricultural productivity and in the southern and western parts of the Potohar plateau, the soil is thin and infertile [88]. Streams and ravines cut the loose plain and cause erosion and steep slopes. This land is generally unsuitable for cultivation. However, large patches of deep, fertile soil are found in the depressions and sheltered parts of the plateau and these support small forests and agriculture [89]. Butt et al. [45] described the most significant land degradation issues in the Rawal watershed as soil erosion and loss of soil nutrients. They further showed that, between 1992 to 2012, the majority of land that was previously vegetation, bare soil, or water bodies was converted to agriculture and settlements, suggesting increased pressure on natural resources in the Rawal watershed [45].

The results of landscape analysis in this study reveal that the ICT urban landscape has become more heterogeneous, disproportional and diverse, and tree patches have declined. Alarmingly, core forests of >500 acres have declined almost 15% in forty years. Although at the individual level, the residents of ICT, civil society, and the local governmen<sup>t</sup> are trying to recover tree cover loss by planting trees, but these initiatives should be continuous and ongoing on a regular basis to monitor the growth of trees without damaging the existing mature standing trees. The temporal forest fragmentation analysis shows that due to tree cover loss, the three categories of core forest fragmentations (i.e., <250 acres, 250–500 acres, and >500 acres) have decreased in forty years (1976–2016). The loss in forest fragmentation negatively influences the habitat and terrestrial biodiversity of ICT [79,90,91]. Based on landscape metrics analysis over the Dhaka metropolitan, a similar Asian developing city, Dewan et al. [31] revealed that cultivated areas and vegetated lands became highly fragmented with increasing anthropogenic disturbances and urban built up category became aggregated and convoluted.

While our analysis of ICT land cover change dynamics in this study is important and unique, there are a few caveats which are worth mentioning. First of all, for the 1976 land cover map, we have relied on approximately 57 m spatial resolution Landsat 3 Multispectral Scanner System (MSS) sensor data which is relatively coarser spatially as compared to Landsat 5 and 8 (i.e., 30 m), which may affect the spatial heterogeneity and accuracy of developed land cover maps. Second, as Landsat is a pioneer Earth Observation (EO) satellite program initiated in 1972, so acquiring remote sensing imagery of the years before that is not possible, and thus we cannot derive land cover maps before the 1960s, when the Islamic Republic of Pakistan's capital shifted from Karachi to Islamabad. Third, from Landsat 30 m medium spatial resolution satellite data, we cannot detect and delineate the boundaries of built-up areas as well as can be done from sub-meter VHRS images available from the 2000s onward. However, the VHRS datasets come with a high cost, especially when acquisitions at multiples times have to be acquired, and may not be therefore feasible for study sites in developing countries. Of course, research in this domain is ongoing with regards to utilization of VHRS data for study of urban areas, using methods like object-based image analysis, machine learning, etc. Fourth, there may be some level of uncertainty for the field measurement data, as there is always a potential for human error especially when ground truth is collected over larger areas. Fifth, in this study, we utilized only temporal optical satellite data, which may cause optical signal saturation in closed canopy forests, and atmospheric effects (i.e., cloud coverage, haze, and smog).
