Socioeconomic drivers on annual mangrove cover changes
Annual mangrove cover change is a vector indicator for each country or territory, i.e., if negatively correlating with socioeconomic indicators means mangrove cover loss with the socioeconomic indicator increase and vice versa. As an important socioeconomical indicator, GDP per capita showed significant and positive correlations with annual mangrove cover change regardless of continents and the globe. The multi-databases dataset with 83 countries and territories in 1990-2016 (p < 0.01) had a greater significance level of the corrections than the CGMFC dataset with 78 countries and territories in 2001-2012 (p < 0.05), except that the Oceania (p < 0.001) and South America (not significant) had negative correlations in 2001-2012 (Table 1). The positive corrections were in agreement with the findings of Barbier and Cox (2003) but the negative ones suggest that the mangrove cover was deforesting with GDP per capita increases in the countries and territories of the Oceania and South America in 2001-2012 (p < 0.01, Table 1). The 6 included Oceanian countries and territories (Solomon Islands, Vanuatu, Fiji, Palau, New Zealand, and Australia in an increasing order) had a wider range of GDP per capita from 1,913 to 68,012 current USD than the 7 involved countries and territories in the South America (Guyana, Ecuador, Peru, Colombia, Suriname, Brazil, and Venezuela in an increasing order) with a range of 3,788 - 12,985 current USD by 2012, calculated from the World Bank data (https://data.worldbank.org).
Conversions to urban land uses were considerable in some countries (Branoff 2007; Richards and Friess 2016). It was true that the Oceania and South America had significant and negative correlations between annual mangrove cover change and urban population percentile, a common indicator of urbanization, in 2001-2012 (p < 0.05, Table 1). However, most mangrove covers are located in rural coasts except some coastal cities in the world. Therefore, it might be true in some regions that the more people moved into urban areas, the less mangrove covers were impacted and lost. In Asia, it was confirmed that urban population percentile was significantly and positively correlated to annual mangrove cover change for both periodic datasets, i.e., 1990-2016 (p < 0.05) and 2001-2012 (p < 0.001) (Table 1) although Vietnam had a large loss to urban lands in 2000-2012 (Richards and Friess 2016). Goldberg et al. (2020) pointed out that the urban land conversion did not impact mangrove covers as great as aquaculture and agricultural plantations.
Aquaculture had resulted in tremendous deforestation of the global mangrove covers in the past, especially in certain regions (Bryan-Brown et al. 2020; Goldberg et al. 2020). The aquacultural indicators, production per capita and sales value per capita, were significantly and negatively correlated with annual mangrove cover change for the globe and the continents (except Africa) in 2001-2012 while the aquacultural contribution to the national GDP had significant and negative correlations only in Asia, North and Central America, and the globe (p < 0.01, Table 1). The multi-databases dataset in 1990-2016 had negative correlations between the aquacultural contribution to the national GDP and annual mangrove cover change only in Asia (p < 0.05, Table 1). Among the 7 countries with the aquacultural contribution to the national GDP above 1% by 2016, Asia had 6 countries, including Vietnam (4.45%), Myanmar (2.97%), Bangladesh (2.54%), Cambodia (1.59%), China (1.28%), and Indonesia (1.18%) and only Ecuador (2.34%) from the South America. By 2012, only Indonesia was excluded from the abovementioned list with two additions of Belize (1.35%) and Honduras (1.27%) from the Central America. No African mangrove-holding countries had the aquacultural contribution above 1% of the national GDP across the period 1980-2016 except Seychelles with 1.4% only by 2003. The top Asian countries and Ecuador had increased aquacultural contributions to the national GDP from 2012 to 2016. On the other hand, aquacultural product price was positively and significantly correlated with annual mangrove cover change in Asia and Oceania for both periodic datasets and Africa and the globe only in 1990-2016, but negatively in Africa in 2001-2012 (p < 0.05, Table 1). The negative correlations suggest that aquaculture development drove mangrove deforestation in the period 2001-2012 and Asia was the hotspot, in agreement to the land conversion studies (Bryan-Brown et al. 2020; Goldberg et al. 2020). However, the aquacultural product price was not simply a motivative factor for farmers to deforest mangroves as hypothesized but also a consequence of mangrove cover protections, i.e., the less mangrove cover lost the higher price was in 2001-2012, except in Africa with low contributions to national GDP per capita (Table 1). It might be partially coincided to the findings of declining aquaculture impacts on mangrove covers in Southeast Asia in 2000-2012 (Richards and Friess 2016).
Overall patterns of socioeconomic-aquacultural indicators and annual mangrove cover change were relatively similar among continents between the two datasets, i.e., 1990-2016 and 2001-2012, visualized by a multidimensional preference analysis (Fig. 2). Only Africa was obviously separated from other continents mainly by the aquacultural indicators, including production per capita, sales value per capita, and contribution to the national GDP. However, annual mangrove cover change and aquacultural product price were highly overlapped with close relations to GDP per capita and urban population percentile in 1990-2016 but evidently separated in 2001-2012 (Fig. 2). GDP per capita and urban population percentile were consistently overlapped for both periods (Fig. 2). These patterns were highly coincident with the results of correlation analysis (Table 1) and suggested that annual mangrove cover change was mainly determined by the aquaculture but influenced by the national economy (GDP per capita) and urbanization (urban population percentile). It is evident that the influences of the national economy and urbanization were stronger on annual mangrove cover change in 1990-2016 than those in 2001-2012. It further confirmed the declining aquacultural impacts (Richards and Friess 2016). Among the continents, annual mangrove cover change was mainly determined by aquaculture in Asia and North and Central America while the others were driven by the national economy and urbanization, especially GDP per capita, in 2001-2012 (Fig. 4). However, in the period 1990-2016 the continental patterns were interactively impacted by the three drivers, i.e., the national economy, urbanization, and aquaculture (Fig. 3). On the other hand, aquacultural product price was consistently determined by GDP per capita and urban population percentile for all the continents in the two periodic datasets except the South America in 1990-2016 (Fig. 3 and 4), in coincidence with the correlations (Table 1).
Integrating findings from the correlation analyses and multidimensional preference analyses, annual mangrove cover change was interactively driven by the national economy, urbanization, and aquaculture, among which the aquacultural impacts were declining and the influences of national economy and urbanization were increasing. It was in agreement with findings in southeast Asia (Richards and Friess 2016). Meanwhile, aquacultural product price might be feedbacked by mangrove conservation, especially in Asia and Oceania within both datasets, as well as a motivative factor for mangrove deforestation in Africa within the CGMFC dataset as hypothesized. Removing temporal effects using an ANOVA-Repeated Measures (Proc Mixed), annual mangrove cover change was not significantly influenced by any selected socioeconomic-aquacultural indicators in 1990-2016 (the multi-databases dataset) and only by GDP per capita, and aquacultural production per capita and product price in 2001-2012 (the CGMFC dataset) although both models were statistically significant (p < 0.05, Table 2). On the contrary, the aquacultural product price was consistently and significantly determined by GDP per capita, urban population percentile, and per capita aquacultural production and sales value in both periodical datasets excluding the temporal effects (p < 0.05, Table 2). In 2001-2012, the aquacultural product price was also significantly associated with annual mangrove cover change (p < 0.001, Table 2). It further confirms that the socioeconomic patterns of annual mangrove cover change periodically responded to the national socioeconomic development and regulatory governance (Barbier and Cox 2003; Slobodian et al. 2018; Turschwell et al. 2020). The aquaculture farmer’s livelihood had to be insured during tradeoffs of mangrove restoration and conservation (Ha et al. 2012).
Impacts of global mangrove cover changes over the periods
Given socioeconomic development with varying periodic patterns in mangrove-holding countries and territories, it led to periodic changes of global mangrove covers.
Global mangrove covers had been dramatically declined at 1776 km2 a year in 1980-2005 but at 379 km2 a year in 1990-2020 upon the national survey by FAO (Food and Agriculture Organization 2003, 2007, 2020) (Fig. 1). However, the significant shift of annual declining rate might be contributed by the estimation technology. The FAO remarkably lowered the global cover estimate in 1990 and afterwards by 2020 in comparison with the previous estimates (Food and Agriculture Organization 2003, 2007, 2020) probably due to the remote sensing technology available to the most world, which were relatively comparable to the cover estimates and annual cover change (175 km2 a year) in 1996-2016 via satellite image interpretation consisting of the NASA snapshot in 2000 (Giri et al. 2002) and GMW databases in 1996-2016 (Bunting et al. 2018) (Fig. 1). It had concluded that the fast decline of global mangrove covers in the late 20th century was slowed in the early 21st century (Friess et al. 2019), but overestimates might be possible for the national survey prior to the remote sensing application, with the global coverage around 198,000 km2 in 1980 (Groombridge 1992; Fisher and Spalding 1993; Food and Agricultural Organization 2003). The earliest interpretation of integrated aerial and satellite images for global mangrove distribution mapping was the World Mangrove Atlas (Spalding et al. 1997) with an estimate of 181,077 km2 in a wide timing span since 1980s, followed by the World Atlas of Mangroves at 152,361 km2 mainly within 1999-2003 (Spalding et al. 2010) (Fig. 1). The cover estimates with mangrove biome and canopy (MFW) in the CGMFC dataset by Hamilton and Casey (2016) were quite different from the estimates of national surveys (Food and Agricultural Organization 2003, 2007, 2020) and satellite image interpretations (NASA and GMW) due to different approaches. The global mangrove biome cover of the CGMFC dataset (Hamilton and Casey 2016) was used individually in this study to compare with the FAO and GMW multi-databases for identifying socioeconomic drivers of annual mangrove cover changes globally. Given the remarkable shifts of mangrove cover estimates by various approaches over time, annual mangrove cover change might alleviate the systematic differences of the estimation approaches to explore socioeconomic patterns of mangrove cover changes instead of those direct estimates as the assumption. Meanwhile, the two different databases did present periodic socioeconomic patterns of annual mangrove changes (Table 1 and 2, and Fig. 2, 3, and 4).
At patch-level of the GMW dataset, approximately 17.1% of mangrove patches (over 118 km2) were fully lost while over 65% of mangrove patches were under shrinkage with an area loss of 8,384 km2 (5.6% of mangrove cover area in 1996) during the 20 years in the world (Table 3). Importantly, the patch size significantly increased from 18.6 ha to 23.3 ha of the national average and from 14.2 ha to 17.1 ha of the national median (Table 3). It suggests that the small patches dominated mangrove cover losses, especially the full patch loss. Small patches of mangroves are important to its value as an ecosystem for blue carbon stock but also to surrounding habitats as nodes for interconnectedness (Barbier 2017; Curnick et al. 2019). In most developing regions, family-run aquacultural ponds were dominated, such as in India (Sarkar et al. 2015) where 83% ponds had a size less than 0.3 ha, relatively similar to the fully-lost mangrove patches, 0.105 ha of average and 0.069 ha of median according to the GMW databases (Table 3). On the other hand, the abandoned aquacultural ponds were highly recommended as an effective measure to restore mangrove ecosystems recently (Stevenson et al. 1999; van Bijsterveldt et al. 2020). However, it was very limited up to 2016, approximately 3,749 patches and 2.79 km2 in area with a median patch size at 0.076 ha, similar to the fully-lost patches as above-mentioned (Table 3). Alternatively, the natural proliferation and expansion was remarkable in 1996-2016, approximately 3,498 km2 in 80,248 patches globally (Table 3), maybe substantially contributed by the reserves or protected areas (Turschwell et al. 2020). Therefore, both deforestation and restoration/conservation activities or patterns might influence socioeconomic patterns of mangrove cover changes. More details required substantial studies in the future.
Table 3
Changes of mangrove covers in patch and area in the period 1996-2016¶
Item
|
Year
|
Existing patch
|
Full patch loss
|
Partial patch loss
|
Expanding patch
|
Rehabilitation patch
|
Increase of patch
|
Difference between existing and losses
|
Number of countries and territories
|
|
105
|
89
|
97
|
87
|
9
|
|
|
Patch number
|
1996
|
701,016
|
|
|
|
|
|
|
2016
|
584,632
|
120,128
|
456,449
|
80,248
|
3,749
|
3,749
|
|
Loss
|
116,384
|
120,128
|
|
|
|
|
-3,744
|
%
|
|
17.1%
|
65.1%
|
11.4%
|
|
|
|
Area (km2)
|
1996
|
149,481.52
|
|
|
|
|
|
|
2016
|
144,283.49
|
118.17
|
8,384.48
|
3,498.13
|
2.79
|
3,500.93
|
|
Loss
|
5,198.03
|
118.17
|
8,384.48
|
|
|
|
-3,304.63
|
%
|
|
0.08%
|
5.61%
|
2.34%
|
|
|
|
Patch size (ha)
|
Country Mean
|
1996
|
18.6
|
|
|
|
|
|
|
2016
|
23.3
|
0.105
|
3.17
|
24.8
|
0.108
|
|
|
Country Median
|
1996
|
14.2
|
|
|
|
|
|
|
2016
|
17.1
|
0.069
|
1.33
|
1.86
|
0.076
|
|
|
¶Data were extracted from the Global Mangrove Watch (Bunting et al., 2018). |