Assessing Land Use and Land Cover Change in Coastal Urban Wetlands of Ghana (PDF)

Wetlands Ecol Manage https://doi.org/10.1007/s11273-020-09712-5 ( 01234567 (0123456789().,-volV) 89().,-volV) ORIGINAL PAPER Assessing land use and land cover change in coastal urban wetlands of international importance in G...

Wetlands Ecol Manage https://doi.org/10.1007/s11273-020-09712-5 ( 01234567 (0123456789().,-volV) 89().,-volV) ORIGINAL PAPER Assessing land use and land cover change in coastal urban wetlands of international importance in Ghana using Intensity Analysis Bernard Ekumah. Frederick Ato Armah. Ernest K. A. Afrifa. Denis Worlanyo Aheto. Justice Odoiquaye Odoi. Abdul-Rahaman Afitiri Received: 18 September 2019 / Accepted: 13 February 2020 Ó Springer Nature B.V. 2020 Abstract In the era of global environmental change, driving the wetland landscape transformation. Wet- land transformation is acknowledged as a critical lands considered for this study were the Densu Delta, subject that needs to be addressed. Even though some Sakumo II and Muni-Pomadze Ramsar Sites. The studies have been carried out in Ghana on land use and observed overall annual change in the first time land cover (LULC) change of wetlands, the conven- interval; 2.42% (Densu Delta), 1.47% (Sakumo II) tional methods used were unable to reveal the and 2.65% (Muni-Pomadze) was smaller compared to underlying processes associated with the land trans- that of the second time interval 2.60% (Densu Delta), formation. This study employed Intensity Analysis to 3.55% (Sakumo II) and 2.91% (Muni-Pomadze). The assess LULC change pattern (1985–2017) in three human-induced LULC categories continuously grew coastal urban wetlands of international importance in at the expense of natural LULC categories. Large Ghana in order to identify the fundamental processes transitions from natural LULC categories to built-up were observed in all the three wetlands and in addition, large transition of dense forest to cultivated land was B. Ekumah (&)
F. A. Armah
E. K. A. Afrifa
recorded at the Muni-Pomadze Ramsar Site. The main A.-R. Afitiri underlying process associated with land transforma- Department of Environmental Science, School of Biological Sciences, College of Agriculture and Natural tion in the wetlands was urbanization. Besides, Sciences, University of Cape Coast, Cape Coast, Ghana agricultural activities also contributed substantially e-mail: [email protected] to LULC changes at the Muni-Pomadze wetland. F. A. Armah e-mail: [email protected] Keywords Land transformation
Ramsar site
E. K. A. Afrifa Wetland landscape
Urbanization e-mail: [email protected] A.-R. Afitiri e-mail: [email protected] Introduction D. W. Aheto Centre for Coastal Management, University of Cape Coast, Cape Coast, Ghana Wetlands are fragile ecosystems characterized by e-mail: [email protected] complexity, dynamism, and diversity (Keddy 2010). As interphase between aquatic and terrestrial ecosys- J. O. Odoi Nature Today, P. O. Box OS 1455, Osu-Accra, Ghana tems (Kadlec and Wallace 2009), wetlands provide e-mail: [email protected] important ecosystem services to humankind (Bacani 123  Wetlands Ecol Manage et. al. 2016). Nonetheless, due to rapid urbanization, processes that drive LULC changes in coastal urban industrialization, agricultural activities (Zedler and wetlands in Ghana remain unclear. The understanding Kercher 2005) and some natural processes such as of patterns and processes associated with LULC climate change, these critical wetlands are continu- change of wetlands will inform decision makers and ously being degraded, generating several ecological researchers on appropriate wetland management and social problems (Bacani et. al. 2016). strategies. Aldwaik and Pontius Jr (2012, 2013), Globally, more than 50% of wetlands have been developed a method known as Intensity Analysis that lost (Davidson 2014; Li et al. 2018). Wetlands in urban assesses the gains and losses of LULC categories and landscapes in developing countries are most affected also reveals the patterns and underlying processes due to rapid urban growth (Kometa et al. 2017; Turner causing the transformations. The common application 1991). In the coastal cities, the pressure on wetlands is of Intensity Analysis is assessment of LULC change far worse because of the exponential growth in human through time. It is useful in assessing the evidence of population (Nicholls 2004). In developing countries, particular hypothesized process of change and can also coastal urban areas are the hotspots for most of the be used to develop new hypotheses regarding pro- economic activities and these increase rural urban cesses of change (Pontius Jr et al. 2013). Braimoh migration (Hinrichsen 1999). Despite the fact that (2006), employed Intensity Analysis to study random wetland ecosystems are subjected to unbridled and systematic land cover transitions in northern destruction by anthropogenic activities, studies eval- Ghana and Alo and Pontius Jr (2008), also used it to uating their state are patchy and limited, especially in identify systematic land-cover transitions of forests Africa, South America and Oceania (Davidson 2014). inside and outside protected areas in south western Assessing and monitoring LULC change are necessary Ghana. Hitherto, no study has employed Intensity for evaluating impact of change on ecosystems at Analysis to analyse LULC change of wetlands in various spatial scales and time. LULC change has Ghana to reveal patterns and processes associated with been identified as a primary cause of loss of ecosystem wetland landscape transformation. and fragmentation of wetland landscapes (Torbick The study applies Intensity Analysis to assess et al. 2006). Remotely sensed data coupled with LULC change of coastal urban wetlands in Ghana to advancement in geospatial techniques have improved quantify the patterns and to identify the underlying the assessment of wetland landscapes (Davidson et al. processes across two time intervals (1985–2002 and 2018). 2002–2017). Wetland ecosystems in Ghana constitute about 10% of the country’s total land surface (Ministry of Lands and Forestry 1999). Studies have revealed that in Materials and methods Ghana, urbanization, high population growth, fuel wood gathering, salt production and sand winning are Study area the main factors threatening wetland ecosystems along the coast (Attuquayefio and Wuver 2003; Gbogbo and Three coastal urban wetlands were selected for this Attuquayefio 2010; Nartey et al. 2011). In Ghana, study; Densu Delta, Sakumo II and Muni-Pomadze, several studies have been carried out to analyse the Ramsar Sites. Among the six wetlands designated as spatial and temporal patterns of LULC change (Abass wetlands of international importance in Ghana, the et al. 2018; Ashiagbor et al. 2019; Basommi et al. three wetlands considered in this study are the ones 2015). However, the mere monitoring of LULC located in coastal urban areas. Based on the Ramsar change does not reveal the underlying processes that Convention criteria, the three wetlands are classified drive the land transformation (Alo and Pontius Jr under marine/coastal wetlands (Ministry of Lands and 2008). The conventional methods of analysing change Forestry 1999). matrix do not provide sufficient quantitative and The Densu Delta Ramsar Site is found in the south- intensive signals of LULC pattern (Huang et al. western part of Accra, about 11 km from Ghana’s 2012). Researchers usually want to associate land capital city. The Ramsar Site constitutes the lower change patterns with processes (De Alban et al. 2019). reaches of the Densu River water course and its The fundamental understanding of patterns and confluence with the Atlantic Ocean and occupies an 123 Wetlands Ecol Manage area of about 46.2 km2 (Gbogbo and Attuquayefio Baidu and Gordon (1991); the freshwater marsh, the 2010). It is located between latitude 5° 300 000 –5° 360 000 open lagoon, the coastal savanna grassland and the N and longitude 0° 180 000 –0° 240 000 W (Fig. 1). There surrounding floodplain. The main livelihoods of the are five major habitats found in the Ramsar Site; surrounding communities are fishing, farming and freshwater marsh, brackish lagoon, sand dunes, industrial activities. coastal savanna grassland and thicket (Ntiamoa-Baidu The Muni-Pomadze Ramsar Site is situated to the and Gordon 1991). The main livelihood activities of west of Winneba in the Central Region of Ghana, the people in the surrounding communities are fishing, about 55 km from Accra and occupies an area of large scale commercial salt extraction and peasant approximately 95 km2 (Gordon et al. 2000). The Muni farming (Ntiamoa-Baidu and Gordon 1991). Lagoon covers an area of 3 km2. It is located between Sakumo II Ramsar Site is located in the Tema latitude 5° 180 000 –5° 180 000 N and longitude 0° 360 000 – Metropolitan Assembly, about 15 km East of Accra. It 0° 480 000 W (Fig. 1). This Ramsar Site is bounded by is situated between latitude 5° 360 000 –5° 420 000 N and the Atlantic Ocean at the south, to the west by the longitude 0° 00 000 –0° 60 000 W (Fig. 1). The Ramsar Yenku Forest Reserve and to the East by Winneba Site covers a total area of 13.4 km2 (Gbogbo et al. Township. It has four main habitat types; floodplain 2012). There are four main types of habitats in the grassland, open water, forest and scrubland (Ntiamoa- Sakumo II Ramsar Site as identified by Ntiamoa- Baidu and Gordon 1991). The primary livelihood Fig. 1 A Map Showing the Locations of the Densu Delta, Sakumo II and Muni-Pomadze Ramsar Sites (*RS Ramsar Site) 123  Wetlands Ecol Manage activities of the fringe communities are farming and LULC maps was carried out using samples from UAV fishing. images, Google Earth and other available topographic maps. Post-classification comparison was carried out Landsat satellite image acquisition to estimate changes in the LULC categories. Change and classification detection statistics tool in ENVI 5.3 was used to quantify changes that occurred during the two time Landsat satellite images for the years 1985, 2002, and intervals in the wetlands. 2017 were downloaded from the United States Geo- logical Survey website (USGS EarthExplorer 2019) Intensity analysis for this study. The details of the Landsat satellite images are provided in Table 1. Intensity Analysis is a mathematical framework to The satellite images were classified using Super- express differences within a set of categories that exist vised Classification with Maximum Likelihood Clas- at multiple time points (Quan et al. 2019). Intensity sifier in ENVI 5.3. Radiometric and atmospheric Analysis was carried out to assess the sizes and corrections were carried out at the preprocessing stage. intensities of temporal changes among LULC cate- The individual bands of the satellite images were gories. Intensity Analysis was conducted at three stacked and projected into the Universal Transverse levels: interval, category, and transition. The interval Mercator (UTM) projection system (Zone: 30 N, level examines the overall change in each time Datum: WGS84). The boundary of the wetlands were interval. It determines how the size and rate of change digitized from Ntiamoa-Baidu and Gordon (1991). varies across time intervals. The estimated annual area Existing topographic maps, Google Earth images and of overall change for each time interval is compared to GPS coordinates from ground information were used the uniform intensity, which is the annual change if all as reference data to classify the satellite images. Densu the changes were distributed uniformly across the Delta and Sakumo II Ramsar Sites were classified into entire temporal extent. Interval Intensity Analysis four LULC classes; marsh, thicket, water and built-up/ determines the time interval in which the overall bareland. The Muni-Pomadze Ramsar Site had a annual rate of change was relatively slow or fast. An different classification scheme due to the prevailing annual change of a time interval is considered as fast conditions. It was classified into six categories; Dense when it is larger than the uniform change whereas a forest, shrub/grassland, water, built/bareland, burnt slow annual change occurs when the annual change is land and cultivated land. Wildfires are common in the smaller than the uniform change. Muni-Pomadze Ramsar Site especially during the dry The category Intensity Analysis examines the season (Attuquayefio and Wuver 2003). In essence, variations in terms of intensity of change among the this study included burnt land as a LULC category in categories of LULC. At this level, the intensity of order to assess how it had changed over the time annual gross gains and losses for each category are intervals. Cultivated land represents a mosaic of farm computed and compared to the uniform intensity that lands and fallow lands. would have occurred if all categories gain or lose with the same intensity in each time interval. An annual Land use and land cover (LULC) change change of a LULC category is said to be dormant when it is smaller than the uniform intensity of change in a After the image classification, LULC maps of all the 3 time interval. An active change occurs when the years (1985, 2002 and 2017) of the study areas were annual change of a LULC category is larger than the developed. The accuracy assessment of the generated uniform intensity of change in a time interval. Table 1 Landsat satellite Satellite Sensor Path/row Spatial resolution (m) Acquisition date Source images used in the study Landsat 5 TM 193/56 30 06/03/1985 USGS 2018 Landsat 7 ETM? 193/56 30 26/12/2002 USGS 2018 Landsat 8 OLI/TIRS 193/56 30 11/12/2017 USGS 2018 123 Wetlands Ecol Manage The transition level shows the variation in intensity Tables 3 and 4 provide land transition matrices for with which the gain of a particular LULC category the wetlands during two time intervals. Land transition transitions from other LULC categories within each matrix provides detailed information on the area of time interval (Aldwaik and Pontius Jr 2012). It gross loss, gross gain and net change for each LULC compares how each category gains from other cate- category as well as the overall change for the wetlands gories during each time interval (Quan et al. 2019). in the two time intervals. The numbers on the diagonal The transition level analyzes whether a category’s (bold face) indicate persistence. The Densu Delta gain intensively targets or avoids the other categories Ramsar Site had marsh as the largest LULC type in at the start of each time interval. The uniform 1985 (17.5 km2, 37.1%) and 2002 (17.2 km2, 36.4%) transition intensity is compared to the observed but, built-up/bareland became the largest in 2017 (23.9 intensities recorded in each time interval. If the km2, 50.6%). Marsh was the dominant LULC type for observed transition intensity of a particular LULC the Sakumo II Ramsar Site in 1985 (9.1 km2, 63.6%) category is greater than the uniform transition inten- and 2002 (8.2 km2, 56.9%). As observed in the Densu sity, then it implies that the gain of that category Delta, built-up/bareland became the dominant LULC targets the other category. Contrarily, if the observed type in 2017 (4.8 km2, 33.8%). The Muni-Pomadze transition intensity of a particular LULC category is Ramsar Site had shrub/grassland as the largest LULC less than the uniform transition intensity, then the gain category in 1985 (41.3 km2, 43.0%) and 2002 (30.1 of that category avoids the other category. In this km2, 31.3%) however, cultivated land became the study, transition analysis was done for the largest dominant LULC type in 2017 (30.8 km2, 32.1%). gaining categories for each time interval. The gross gains for the first time interval for all The Intensity Analysis was carried out using an categories with the exception of built-up/bareland open source Microsoft Excel programme from Inten- were greater than the gross gains for the second time sity Analysis website (https://sites.google.com/site/ interval. Built-up/bareland decreased in gross loss in intensityanalysis/). The programme was developed by the second time interval at the Densu Delta Ramsar Safaa Zakaria Aldwaik and Robert Gilmore Pontius Jr. Site. Persistence of all the LULC categories decreased The mathematical equations for the three levels of in the second time interval except built-up/bareland Intensity Analysis used in developing the Microsoft which increased. At the Sakumo II Ramsar Site, the Excel programme and adopted in this study are found gross gains in the second time interval for all in Aldwaik and Pontius Jr (2012). categories with the exception of marsh were greater than the gross gains for the first time interval. Similar to the Densu Delta, persistence of all the LULC Results categories reduced in the second time interval except built-up/bareland which increased. At the Muni- Changes in LULC in the wetlands Pomadze Ramsar Site, gross gain for the second time interval for cultivated land, built-up/bareland, The study revealed noticeable changes in LULC in the shrub/grassland and water were greater than gross wetlands across the three time points. Figure 2 gain recorded in the first time interval. Dense forest presents the LULC maps for the wetlands. and burnt land increased in gross loss in the second Table 2 shows the accuracy assessment report for time interval. Persistence of shrub/grassland, dense the LULC maps generated. Both the overall and class forest and burnt land decreased in the second time accuracy were provided. The user’s and producer’s interval while that of built-up/bareland, cultivated accuracy were reported for all the LULC classes. The land and water increased. accuracy of the maps for the three time points for all the wetlands was generally high. Interval level Intensity Analysis The overall accuracy for the LULC maps were all above 80 percent with the exception of Sakumo II Figure 3 provides the interval level Intensity Analysis (2017) and Muni-Pomadze (2002) maps which were for the three wetlands for 1985–2002 and 2002–2017 79.0 and 77.5 percent repectively. time intervals. In all the wetlands, annual change in the second time interval was relatively fast compared to 123  Wetlands Ecol Manage Fig. 2 LULC maps for 1985, 2002 and 2017. A Densu Delta Ramsar Site. B Sakumo II Ramsar Site. C Muni-Pomadze Ramsar Site the first time interval. At the Densu Delta Ramsar Site, Category level Intensity Analysis the annual change (2.42%) in the first time interval was slow compared to that of the second time interval Figure 4 presents the results for category level Inten- (2.60%). Similar to the Densu Delta, at the Sakumo II sity Analysis for the three wetlands for 1985–2002 and Ramsar Site, the annual change in the first time 2002–2017. Thicket experienced active losses but interval (1.47%) was relatively slow compared to the dormant gains in the first time interval in the Densu annual change that was recorded in the second time Delta Ramsar Site. Marsh and water recorded dormant interval (3.55%). The Muni-Pomadze Ramsar Site gains and losses. The gains and losses for built-up/ also experienced a slower annual change in first time bareland were active. In the second time interval, interval (2.65%) compared with that of the second thicket recorded active gains and losses. Marsh was time interval (2.91%). dormant in terms of gains but experienced active losses. Built-up/bareland recorded active gains and 123 Wetlands Ecol Manage Table 2 Accuracy assessment report for the LULC maps of the wetlands 1985 2002 2017 Producer’s User’s Producer’s User’s Producer’s User’s accuracy (%) accuracy (%) accuracy (%) accuracy (%) accuracy (%) accuracy (%) Densu Delta Ramsar Site Thicket 86.5 92.6 82.2 89.0 66.2 71.0 Marsh 88.1 70.8 88.2 72.0 86.1 84.8 Built 75.1 79.4 79.3 80.8 95.2 89.5 Water 68.4 94.3 77.1 97.7 87.3 97.4 Overall 81.5 82.3 88.4 accuracy Sakumo II Ramsar Site Thicket 78.5 83.9 85.1 79.3 90.2 68.8 Marsh 90.7 84.8 92.1 89.5 81.6 92.1 Built 73.2 58.7 74.8 75.4 93.7 73.7 Water 63.3 89.6 72.3 90.2 45.7 81.9 Overall 82.2 85.5 79.0 accuracy Muni-Pomadze Ramsar Site Degraded 75.6 86.4 74.8 81.3 93.3 83.7 Dense 84.2 72.6 80.2 72.4 84.7 92.2 Water 33.8 99.7 48.4 97.8 79.9 99.7 Built 84.7 84.8 83.0 79.5 86.2 94.0 Scrub 89.3 81.0 79.7 85.7 86.8 86.4 Burnt 88.9 99.7 73.2 47.7 64.7 50.4 Overall 81.4 77.5 87.0 accuracy dormant losses. Water experienced dormant gains and losses. Cultivated land, built-up/bareland and water losses in the second time interval. recorded active gains and dormant losses. Dense forest At the Sakumo II Ramsar Site, in the first time and shrub/grassland recorded dormant gains and interval, thicket and built-up/bareland recorded active active losses. Contrary to the first time interval, water gains and losses whereas marsh experienced dormant experienced active gains and losses in the second time gains and losses. Water was dormant in terms of gains interval. and active in losses. In the second time interval, thicket experienced dormant gains and losses. Marsh Transition level Intensity Analysis recorded dormant gains but experienced active losses. Built-up/bareland recorded active gains and dormant The study considered the large gross gains in each of losses. Water experienced active gains and losses. the wetlands to quantify patterns of LULC change. At the Muni-Pomadze Ramsar Site, cultivated land, Figure 5 shows results from Intensity Analysis’ tran- dense forest and burnt land recorded active gains and sition level for the Densu Delta and Sakumo II Ramsar losses in the first time interval. Built-up/Bareland, Sites. shrub/grassland and water experienced dormant gains The two sites had large gains of built-up/bareland and losses. In the second time interval, burnt land was during both time intervals. The gain of built-up/ the only category that experienced active gains and bareland targeted marsh and thicket at the Densu Delta 123  Wetlands Ecol Manage Table 3 Land transition matrices (km2) for the Densu Delta and Sakumo II Ramsar Sites for 1985–2002 and 2002–2017 Thicket Marsh Built-up/Bareland Water Initial total Gross loss Densu Delta Ramsar Site 2002 1985 Thicket 4.3 4.3 2.3 0.0 10.9 6.6 Marsh 2.2 10.7 3.9 0.7 17.5 6.8 Built-up/Bareland 0.5 1.8 6.7 2.8 11.8 5.1 Water 0.1 0.4 0.4 6.1 7.0 0.9 Final Total 7.1 17.2 13.2 9.7 47.2 19.4 Gross Gain 2.8 6.5 6.5 3.6 19.4 2017 2002 Thicket 2.0 1.4 3.7 0.0 7.1 5.1 Marsh 1.2 7.3 8.0 0.8 17.2 9.9 Built-up/Bareland 0.1 0.5 11.5 1.1 13.2 1.7 Water 0.2 0.7 0.7 8.0 9.7 1.7 Final Total 3.5 9.8 23.9 10.0 47.2 18.4 Gross Gain 1.5 2.6 12.4 1.9 18.4 Sakumo II Ramsar Site 2002 1985 Thicket 1.8 0.4 0.3 0.1 2.6 0.8 Marsh 0.8 7.4 0.9 0.0 9.1 1.8 Built-up/Bareland 0.2 0.3 0.9 0.0 1.4 0.5 Water 0.3 0.1 0.2 0.7 1.2 0.5 Final total 3.1 8.2 2.3 0.8 14.4 3.6 Gross gain 1.2 0.8 1.4 0.2 3.7 2017 2002 Thicket 1.7 0.2 0.4 0.6 3.1 1.3 Marsh 0.7 3.7 3.3 0.4 8.2 4.4 Built-up/Bareland 0.4 0.4 1.1 0.5 2.3 1.2 Water 0.7 0.0 0.0 0.1 0.8 0.7 Final total 3.5 4.3 4.8 1.6 14.4 7.6 Gross gain 1.8 0.6 3.7 1.5 7.6 Ramsar Site. At the Sakumo II Ramsar Site, the gain of Discussion built-up/bareland targeted only water during the first time interval and targeted only thicket during the This study employed Intensity Analysis to assess second time interval. LULC changes in three wetlands of international Figure 6 presents results from Intensity Analysis’ importance located within urban settlements in the transition level observed at the Muni-Pomadze Ram- coastal zone of Ghana. The study covers thirty-two sar Site. The gain of cultivated land targeted dense (32) years, comprising two time intervals; 1985–2002 forest most intensively during both time intervals. The and 2002–2017. The interval level results show that gain of built-up/bareland targeted shrub/grassland LULC change in the wetlands was more intensive in during the second time interval. the second time interval. This implies that the underlying processes that drive the changes progres- sively increased over the study years. These 123 Wetlands Ecol Manage Table 4 Land transition matrix (km2) for the Muni-Pomadze Ramsar Site for 1985–2002 and 2002–2017 Cultivated Dense Burnt Built-up/ Shrub/ Water Initial Gross Land forest Land Bareland Grassland total loss 2002 Cultivated 6.8 8.7 1.1 0.1 0.4 0.0 17.1 10.3 Land Dense forest 9.6 9.1 1.7 0.2 1.0 0.0 21.5 12.4 Burnt Land 0.3 0.7 0.6 0.01 2.1 0.0 3.7 3.1 Built-up/ 0.4 0.0 0.0 9.6 0.9 0.1 11.0 1.4 Bareland Shrub/ 4.0 1.2 4.0 6.4 25.6 0.1 41.3 15.7 Grassland Water 0.0 0.0 0.0 0.3 0.01 1.0 1.4 0.4 Final total 21.1 19.6 7.4 16.6 30.1 1.2 96.0 43.3 Gross gain 14.2 10.6 6.8 7.0 4.5 0.2 43.3 2017 2002 Cultivated 15.6 2.5 0.3 1.7 0.9 0.0 21.1 5.5 Dense Forest 10.6 7.6 0.3 0.5 0.6 0.0 19.6 12.0 Burnt 2.9 0.1 0.4 0.8 3.3 0.0 7.4 7.0 Built-up/ 0.3 0.0 0.1 12.5 0.8 2.9 16.6 4.1 Bareland Shrub/ 1.5 0.0 1.9 9.8 17.0 0.2 30.1 13.1 Grassland Water 0.0 0.0 0.0 0.1 0.0 1.1 1.2 0.1 Final Total 30.8 10.2 3.0 25.3 22.6 4.3 96.0 41.9 Gross Gain 15.2 2.6 2.6 12.7 5.6 3.2 41.9 underlying processes could be linked to anthropogenic Division 2019). The implication is that these urban activities because, in 1985 and 2002, the largest LULC wetlands will eventually become a built environment categories for all the wetlands were natural LULC depriving urban dwellers of their important ecosystem types but in 2017, human-induced LULC categories services but supplying humans with housing. became the largest. Despite their ecological and social The findings from the category and transition importance, these wetlands are rapidly being con- Intensity Analysis of the wetlands provide information verted into built areas and this calls for the immediate on the gains, losses and the large transitions of the attention of relevant stakeholders. various LULC categories. At the Densu Delta Ramsar The increase in anthropogenic activities in the Site, built-up/bareland expanded in both time intervals wetlands could be attributed to urbanization which is while the natural LULC types decreased. Built-Up/ mainly as a result of increased human population. The Bareland gains were intensively derived from thicket human population of Accra and Winneba doubled and marsh. It implies that marsh and thicket were between 1984 and 2010 according to the Ghana cleared to give way for physical developments. Statistical Service (2013). The population rise could Similar pattern was observed in the second time drive the rapid land transformation observed in the interval, built-up/bareland continued to expand at the second time interval. It is predicted that the shift from expense of thicket and marsh. The land transformation rural to a predominantly urban population will pattern at the Densu Delta Ramsar Site clearly shows continue, with close to about 90% of this increase that urbanization is the main process driving the occurring in Africa and Asia (United Nations, Depart- changes. ment of Economic and Social Affairs, Population 123  Wetlands Ecol Manage expansion was derived from thicket. The wetland was dominated by human activities such as veg- etable farming, livestock rearing, and industrial activ- ities in the early 2000s (Nartey et al. 2011). In recent times the main threat to the wetland is built-up expansion. The results of the transition level Intensity Analysis show that LULC changes at the Sakumo II Ramsar Site in both time intervals were driven intensively by urbanization. The implication is that, progressively the wetland is being converted to impervious surfaces, which has multiple adverse effects on biodiversity. LULC changes in the Muni-Poamdze Ramsar Site were unlike what was observed in the Densu Delta and Sakumo II Ramsar Sites which were largely driven by built-up expansion. In the first time interval, the active losing categories were cultivated land, dense forest and burnt land and they were the same categories that had large gains. These changes were primarily driven by agricultural activities. The only large transition observed was transition from dense forest to cultivated land. The Ramsar Site is surrounded by eleven communities whose main occupation is farming (Gordon et al. 2000). This explains why dense forest was intensely converted into cultivated lands. In the second time interval, two large transitions were observed; dense forest to cultivated land and shrub/grassland to built-up/bareland. The land trans- formation at the wetlands could be attributed to agriculture and urbanization. The massive reduction in dense forest and expansion of cultivated land was as a result of increased farming activities by the fringe communities. Built-up/bareland also grew substan- tially at the expense of shrub/grassland which implies that shrub/grassland areas of the wetland were the Fig. 3 Interval level Intensity Analysis results for 1985–2002 hotspots for development of residential infrastructure. and 2002–2017 time intervals Built-up expansion occurred more at the south eastern part of the wetland and close to the Winneba The LULC changes at Sakumo II wetland had a Township. This reveals that expansion of Winneba similar trend like that of the Densu Delta wetland. Town mainly occur at the shrub/grassland areas of the Built-up/bareland gained substantially from the natu- wetland. ral LULC types. The gains of built-up/bareland from Coastal urban wetlands in Ghana need urgent water in the first time interval could largely be attention from the relevant stakeholders in order to attributed to reclamation of wetlands for physical protect them from rampant destruction as a result of development such as residential facilities. However, increasing demand of coastal urban lands for physical data error from image classification might account for infrastructural development for residential and com- some of the apparent transition from water to built-up/ mercial purposes. Interventions that seek to protect bareland. In the second time interval, the land coastal urban wetlands should include innovative transformation pattern changed. Built-up/bareland ways of maximising the use of limited urban space 123 Wetlands Ecol Manage Fig. 4 Category level Intensity Analysis showing active losing and gaining categories in the wetlands for 1985–2002 and 2002–2017 time intervals to accommodate the rising demand. There should be Study limitations deliberate and concerted effort by the relevant agen- cies to coordinate for effective implementation and Some important LULC categories such as built-up and enforcement of laws and regulations that protect bareland were classified together as one because, the wetlands. Although massive efforts have been put in resolution of the satellites images (30 m) was not high towards wetland conservation in recent years, Ghana enough to enable the separation of these two LULC must prevent more wetlands from being wiped out by categories. High resolution satellite images of the urbanization in order to save the country from the study areas were not available for all the 3 years associated ecological and social implications. considered in this study. The available high resolution 123  Wetlands Ecol Manage Fig. 5 Transition to built-up/bareland at the Densu Delta and Sakumo II Ramsar Sites for 1985–2002 and 2002–2017 time intervals images were captured recently. A similar study also of the wetlands were natural categories but, human- merged the two LULC categories into one (Ashiagbor induced LULC categories such as built-up (Densu et al. 2019). In addition, the two time intervals Delta and Sakumo II) and cultivated land (Muni- considered for this study were unequal. Even though Pomadze) became the largest in 2017. Vegetated Landsat satellite images have been available since surfaces were rapidly replaced by impervious sur- 1972, most of the images did not meet the criteria set faces, indicating the enormous influence of anthro- for this study. However, it is noteworthy that unequal pogenic activities. This situation has potential time interval does not affect Intensity Analysis. implications on the frequency and magnitude of flooding in coastal urban areas, which in turn requires costly flood control infrastructure. The process that Conclusion could be associated with the LULC change pattern observed in the Densu Delta and Sakumo II Ramsar Intensity Analysis was used to assess LULC changes Sites was urbanization whereas agriculture and urban- in three wetlands of international importance in ization were the main drivers of land transformation at coastal urban areas in Ghana. The findings of this the Muni-Pomadze Ramsar Site. study show that LULC changes in the wetlands Given the many watershed services wetlands progressively increased with time and this implies provide, wetland conservation and restoration should that the underlying processes that drive the changes be an integral part of a comprehensive local watershed became more intense with time. These underlying management strategy. Further research is required to processes could be linked to anthropogenic activities comprehensively delineate the indirect impacts of because, in 1985 and 2002, the largest LULC category LULC on urban wetlands, and several priority 123 Wetlands Ecol Manage Fig. 6 Transition to cultivated land and built-up/bareland at the Muni-Pomadze Ramsar Site for 1985–2002 and 2002–2017 time intervals research strategies are warranted at the national and Acknowledgements The authors are indebted to USAID/ sub-national levels. From the big picture perspective, UCC Fisheries and Coastal Management Capacity Building Support Project for providing financial support for this study. however, the current science on wetland impacts from anthropogenic activities presents a compelling case to Funding None. support greater local regulation and management of wetlands and their contributing drainage areas. The corpus of literature on indirect impacts to wetlands in References Ghana and beyond has been generated by a constel- lation of different academic disciplines that rarely Abass K, Afriyie K, Gyasi RM (2018) From green to grey: the interact with each other. Urban wetland research has dynamics of land use/land cover change in urban Ghana. Landsc Res. https://doi.org/10.1080/01426397.2018. been published by geographers, hydrologists, her- 1552251 petologists, landscape ecologists, botanists, wildlife Aldwaik SZ, Pontius RG Jr (2012) Planning intensity analysis to managers, conservation biologist, toxicologists, unify measurements of size and stationarity of land chan- stormwater engineers and wetland scientists. It is ges by interval, category, and transition. Landsc Urban Plan 106(1):103–114. https://doi.org/10.1016/j. recommended that a national network be launched to landurbplan.2012.02.010 improve communication among the diverse research Aldwaik SZ, Pontius RG Jr (2013) Map errors that could community currently working on the topic of direct account for deviations from a uniform intensity of land and indirect impacts of LULC to urban wetlands. change. Int J Geogr Inf Sci 27(9):1717–1739 Alo CA, Pontius RG Jr (2008) Identifying systematic land-cover transitions using remote sensing and GIS: the fate of forests 123  Wetlands Ecol Manage inside and outside protected areas of Southwestern Ghana. Kadlec RH, Wallace S (2009) Treatment Wetlands, 2nd edn. Environ Plan B 35(2):280–295 CRC Press, Boca Raton, p 1020 Ashiagbor G, Amoako C, Asabere SB, Quaye-ballard JA (2019) Keddy PA (2010) Wetland ecology: principles and conservation Landscape Transformations in Rapidly Developing Peri- (second). Cambridge University Press, Cambridge. https:// urban Areas of Accra, Ghana : Results of 30 years. Open doi.org/10.1075/tsl.98.09bur Geoscience 11:172–182. https://doi.org/10.1515/geo- Kometa S, Kimengsi J, Petiangma D (2017) Urban development 2019-0014 and its implications on wetland ecosystem services in Attuquayefio DK, Wuver A (2003) A study of bushfires in a Ndop. Environ Manag Sustain Dev, Cameroon. https://doi. Ghanaian coastal wetland. I. Impact on small mammals. org/10.5296/emsd.v7i1.12141 West Afr J Appl Ecol 4(1):1–11 Li X, Bellerby R, Craft C, Widney SE (2018) Challenges for Bacani VM, Sakamoto AY, Quénol H, Vannier C, Corgne S restoration. Anthropocene Coasts 15:1–15 (2016) Markov chains–cellular automata modeling and Ministry of Lands and Forestry (1999) Managing Ghana’s multicriteria analysis of land cover change in the Lower wetlands: a national wetlands conservation strategy, vol Nhecolândia subregion of the Brazilian Pantanal wetland. 1999. Accra J Appl Remote Sens 10(1):016004 Nartey VK, Edor KA, Doamekpor LK, Bobobee LH (2011) Basommi PL, Guan Q, Cheng D (2015) Exploring land use and Nutrient load of the Sakumo Lagoon at the Sakumo Ramsar land cover change in the mining areas of Wa East District, Site in Tema, Ghana. West Afr J Appl Ecol 19:93–105 Ghana using satellite imagery. Open Geosci 7(1):618–626. Nicholls RJ (2004) Coastal flooding and wetland loss in the 21st https://doi.org/10.1515/geo-2015-0058 century: changes under the SRES climate and socio-eco- Braimoh AK (2006) Random and systematic land-cover tran- nomic scenarios. Glob Environ Change 14:69–86. https:// sitions in northern Ghana. Agriculture. Ecosyst Environ doi.org/10.1016/j.gloenvcha.2003.10.007 101(113):254–263. https://doi.org/10.1016/j.agee.2005. Ntiamoa-Baidu Y, Gordon C (1991) Coastal wetlands man- 10.019 agement plans: Ghana. Accra. https://oai.dtic.mil/oai/ Davidson NC (2014) How much wetland has the world lost? oai?verb=getRecord&metadataPrefix=html&identifier= Long-term and recent trends in global wetland area. Mar AD0702727 Freshw Res 65(10):934–941. https://doi.org/10.1071/ Pontius RG Jr, Gao Y, Giner N, Kohyama T, Osaki M, Hirose K MF14173 (2013) Design and interpretation of intensity analysis Davidson NC, Fluet-Chouinard E, Finlayson CM (2018) Global illustrated by land change in Central Kalimantan, Indone- extent and distribution of wetlands: trends and issues. Mar sia. Land 2(3):351–369 Freshw Res 69(4):620–627. https://doi.org/10.1071/ Quan B, Pontius RG Jr, Song H (2019) Intensity Analysis to MF17019 communicate land change during three time intervals in De Alban JDT, Prescott GW, Woods KM, Jamaludin J, Latt KT, two regions of Quanzhou City, China. GIScie Remote Sens Lim CL et al (2019) Integrating analytical frameworks to 20:1–16 investigate land-cover regime shifts in dynamic land- Torbick NM, Qi J, Roloff GJ, Stevenson RJ, Observation E, scapes. Sustainability 11(4):1139 Lansing E et al (2006) Investigating impacts of land-use Gbogbo F, Attuquayefio DK (2010) Issues arising from changes land cover change on wetlands in the Muskegon River in waterbird population estimates in coastal Ghana 1. Bird Watershed, Michigan, USA. Wetlands 26(4):1103–1113 Popul 10:79–87 Turner K (1991) Economics and wetland management. Ambio Gbogbo F, Langpuur R, Billah MK (2012) Forage potential, 20:59–63 micro-spatial and temporal distribution of ground arthro- United Nations Department of Economic and Social Affairs pods in the flood. Afr J Sci Technol 12(1):80–88 Population Division (2019) World urbanization prospects. Ghana Statistical Serice (2013) 2010 Population and Housing United Nations Department of Economic and Social Census National Analytical Report. Affairs Population Division, New York Gordon C, Ntiamoa-baidu YAA, Ryan JM (2000) The Muni- USGS EarthExplorer (2019) Landsat satellite images. https:// Pomadze Ramsar site. Biodivers Conserv 9:447–464 earthexplorer.usgs.gov/. Accessed 2 Apr 2019 Hinrichsen D (1999) Coastal waters of the world: trends, threats, Zedler JB, Kercher S (2005) Wetland resources: status, trends, and strategies. Island Press, Washington ecosystem services, and restorability. Annu Rev Environ Huang J, Pontius RG Jr, Li Q, Zhang Y (2012) Use of intensity Resour 30:39–74 analysis to link patterns with processes of land change from 1986 to 2007 in a coastal watershed of southeast China. Publisher’s Note Springer Nature remains neutral with Appl Geogr 34:371–384. https://doi.org/10.1016/j.apgeog. regard to jurisdictional claims in published maps and 2012.01.001 institutional affiliations. Intensity Analysis (n.d.) https://sites.google.com/site/ intensityanalysis/free-computer-programs. Accessed 23 July 2019 123