Can Thailand Protect 30% of Its Land Area for Biodiversity, and Will This Be Enough?

Abstract

The draft post-2020 Global Biodiversity Framework asks CBD parties to conserve at least 30% of the planet by 2030 ‘through a well-connected and effective system of protected areas … with the focus on areas particularly important for biodiversity’. We use Thailand as a case study for the ability of a densely populated, hyper diverse, tropical, middle-income country to meet this target at a national level. Existing protected areas (PAs) total 24.3% of Thailand’s land area. Adding forest on government land adjacent to existing PAs, plus unprotected areas of Ramsar sites, raises this to 29.5%. To assess the importance for biodiversity, we used modeled distributions of birds and mammals plus, as proxies for other biodiversity components, elevation, bioclimate, forest type, and WWF ecoregion. All modeled species occur in the current PA system but <30% meet representation targets. Expansion of the system increases the proportion of mammals and birds adequately protected and increases the protection for underrepresented bioclimatic zones and forest types. The expanded system remains fragmented and underrepresents key habitats, but opportunities for increasing protection of these are limited. It is also still vulnerable to climate change, although projected impacts are reduced. Additional protection is needed for wetland and coastal habitats, and limestone karsts.

1. Introduction

Despite widespread recognition of the key role that protected areas have played in biodiversity conservation [1], numerical targets for the total area under protection continue to be controversial. The two major concerns are, first, that area per se is a poor proxy for biodiversity, which is the real target, and second, that the ‘protection’ given has, in practice, often been insufficient to prevent a continued decline in biodiversity because of insufficient resources, ineffective management, and the lack of good data for assessing effectiveness [2,3]. Aichi Target 11 in the CBD’s Strategic Plan for Biodiversity 2011–2020 endeavored to address these issues by combining a numerical target—‘at least 17% of terrestrial and inland water’—with additional qualitative targets, requiring protected areas to be ‘effectively and equitably managed, ecologically representative and well connected, … and integrated into the wider landscapes…’ (www.cbd.int/sp/targets/) (accessed on 2 March 2022). However, while the numerical target was met on a global scale [4] and by many individual countries, few countries have achieved the qualitative targets, and many are nowhere near doing so [2,5].

Despite the justifiable criticisms of area-based targets for conservation, they continue to be popular with governments, and they are much easier to communicate and assess than the outcome-based alternatives that have been suggested. Visconti et al. [2], for example, suggested as an alternative protected area target ‘all sites of global significance for biodiversity’, but this has not been widely adopted, in part because these sites are the least easy to identify in the countries with the most biodiversity. Instead, the first draft of the CBD’s post-2020 Global Biodiversity Framework again combines quantitative and qualitative targets, aiming to protect and conserve ‘at least 30 per cent of the planet’… ‘through a well-connected and effective system of protected areas and other effective area-based conservation measures’ … ‘with the focus on areas particularly important for biodiversity’ by 2030 [6]. Clearly, there is still a danger that the qualitative targets will be ignored or given less weight, and the 30% target will encourage the addition of remote areas of low biodiversity and low current threat. However, management capacity has increased in many countries in the last decade, as has the availability of new technologies and other resources to assist in conservation planning, so this outcome is not inevitable. The key is likely to be if and how these targets are translated into national plans and policies, and then how well these are funded [7].

How practical is a 30% area target for terrestrial areas by 2030? The 30% target is for the planet, but the presumption is that national targets will be aligned with the post-2020 framework as far as possible. For many low- and middle-income countries with high population densities, excluding 30% of the land area from economic activities is either simply not practical or will result in biodiversity being assigned to only the colder, steeper, less fertile, and more remote areas of the country that are unsuitable for agriculture or settlements. In some cases, the concept of ‘other effective area-based conservation measures (OECMs)’, which are areas that are effective in conserving biodiversity even though this is not their primary management objective, may permit a compromise between economic activity and conservation. OECMs were included in the Aichi 17% area target, although undefined and widely ignored, but there is now an official CBD definition, and the IUCN has published guidelines for recognizing and reporting them [8]. Unlike protected areas, which are defined largely on the basis of their legal status, OECMs are defined by their ability to deliver effective conservation, regardless of their legal objectives.

In this paper, we use Thailand as a case study for the ability of a densely populated, hyper diverse, tropical, middle-income country to meet the draft 2030 area target at the national scale in the face of climate change. Although less wealthy and more diverse biologically, Thailand is similar to the European Union in mean population density (137 in Thailand vs. 112 per km² in the EU) and total fertility (1.5 vs. 1.6 births per woman) (https://population.un.org/wpp/) (accessed on 2 March 2022), and in forest cover (31% vs. 37%) and the proportion covered by protected areas (24% vs. 26%) [9,10]. The reported problems with the existing protected area system are also very similar in Thailand and the EU: they are not fully ecologically representative, some sites are under-managed, and most protected areas are small and isolated [9,10]. It is not obvious, however, if the many similarities outweigh the large differences in wealth and biodiversity. We, therefore, asked the following questions:

• How can Thailand meet the proposed terrestrial area target of 30% by 2030?
• Will this 30% be ‘well-connected’ and ‘important for biodiversity’?
• How vulnerable will this 30% be to climate change?

A recent study [11] looked at the ability of Thailand’s existing protected area system to protect the five WWF Global 200 Ecoregions of ecological significance found in the country, as well as richness hotspots for vertebrates. However, by the end of 2021, protected area coverage had increased substantially, from 19% to more than 24%, and the availability of a high-resolution forest map [12] now makes it possible to not only identify under-represented ecoregions and hotspots but also assess the potential for additional protection in these areas. Moreover, new, spatially explicit, climate-change projections for Thailand make it possible to assess potential vulnerability to climate change [10].

2. Materials and Methods

2.1. Study Area

Thailand is a large (517,624 km²) country with a 70-million population in the center of Southeast Asia, bordering Laos, Cambodia, Myanmar, and Malaysia, as well as the Andaman Sea and Gulf of Thailand (Figure 1). It has an upper middle income in the World Bank classification and, although agriculture still accounts for almost a third of employment, it is only 8% of GDP in an economy now dominated by industry and services, including tourism [13]. Most of the country is low-lying (64% is < 250 m a.s.l.) and relatively flat, but there are mountain ranges with peaks up to 2564 m in the west and north. There are distinct wet and dry seasons associated with the Asian summer and winter monsoons, respectively. Temperature declines consistently with altitude, but rainfall has a more complex pattern, with a drier center and east of the country, and wetter south, west, and north [10]. Forest still covers at least 31.7% of the country’s total area [12]. Half the forest area is mixed deciduous forest, with dry evergreen forest, moist evergreen forest, deciduous dipterocarp forest, and hill evergreen forest making up most of the rest, along with smaller areas of pine forest, mangrove forest, and freshwater swamp forest. Dry evergreen forest, mixed deciduous forest, and deciduous dipterocarp forest form—or formed until recently—a mosaic over much of lowland Thailand, controlled by non-climatic factors.

2.2. Data Sources

The study area covers the total land area of Thailand. Thailand also has marine protected areas, but the data needed to evaluate these are not available. We obtained shapefiles for existing and currently planned terrestrial protected areas from the Department of National Parks, Wildlife and Plant Conservation (DNP); for natural and semi-natural vegetation, from the Royal Forest Department; for water catchments, from the Royal Irrigation Department; for Ramsar sites, from the UNEP-WCMC ‘protected planet’ website [14]; and for bioclimate zones, for the present and 2070, from [10]. Elevation was downloaded from the CGIAR-Consortium for Spatial Information [15].

We initially obtained location data for wild species from several sources: standardized vertebrate surveys by trained staff of the Department of National Parks, Wildlife and Plant Conservation (DNP) from 2017 to 2018; data downloaded from the Global Biodiversity Information Facility for 1960–2019 for vertebrates, mostly birds [16] and mammals [17], with more than 95% of the bird records identified as originating in eBird [18], which is popular among birders in Thailand; and plant data from the DNP’s forest resource inventory project from 221 plots, the DNP’s Forest Herbarium, and the Botanical Information and Ecology Network (BIEN, https://bien.nceas.ucsb.edu/bien/) (accessed on 5 December 2019). No invertebrate taxon had adequate coverage for Thailand. We removed suspect records, duplicates from the same locality, and species with <10 localities. In the cleaned dataset, fewer than 20% of Thai amphibian and reptile species were represented and most records were non-forest species of no conservation concern, so both groups were excluded from further analyses. The plant dataset was dominated by forest species, but only c. 6% of Thailand’s flora was included so we also excluded plants. The bird dataset, by contrast, included 702 species, c. 70% of Thailand’s avifauna, and although we had enough data for only 80 mammal species, this included most large and medium-sized species, many of conservation concern. We, therefore, used only the bird and mammal records (325,008 and 243,423, respectively) in subsequent analyses. Note that around a third of the bird species covered were from coastal and inland wetlands, while almost all the mammals were from forests.

2.3. Achieving 30%

We started with the existing terrestrial protected areas covering 24.3% of the land area in total. We define ‘protected areas’ as areas in IUCN categories I–V. [19] which are managed by the DNP, and excluded conservation mangroves, watersheds, and forest reserves, which are only partly protected. To this, we then added additional forest areas on government land outside the protected area system. A priori, large contiguous areas are of greater conservation value than the same area in fragments, particularly in view of the high level of fragmentation of the existing protected areas, so we prioritized patches >10 km² adjacent to existing PAs. Contiguity with existing PAs also minimizes the additional requirements for demarcation, patrolling, and management. Smaller and/or more isolated patches may also be of high conservation value (e.g., support narrow-range endemic plant or invertebrate species), but we did not have the species location data needed to identify these.

We first removed small gaps between forests (i.e., those caused by a road or single crop field) and merged classes of forest, so that distance calculations in relation to protected areas would be consistent. The forest layer was rasterized using the conversion tools in ArcGIS (ESRI) with a resolution of 30 m and no distinction between different types of natural forest. A distance buffer of 250 m was then added by using the path distance function, then multiplied by zero to form a single forest layer, with small gaps removed. This was then converted back into a shapefile, which provided a unique identifier for each forest patch. The calculate geometry tool was then used to calculate the area of each forest patch, and patches of under 10 km² were removed. We then used the path distance function to calculate the distance to protected areas, using a cell size of 250 m to be consistent with our area removal. Once the distance to protected areas had been calculated for the whole of Thailand, we then used the zonal statistics to calculate the minimum distance to a protected area. Those with a minimum of “0” overlaid the protected areas, with the maximum distance to a protected area from areas outside the protected area also given. The percentage of each patch protected was also calculated using zonal geometry to assess what percentage of any given forest patch currently falls within a protected area.

2.4. Assessing Fragmentation

The needs and opportunities for wildlife corridors to connect adjacent protected areas were the subject of a recent 316-page report by the Department of National Parks, Wildlife and Plant Conservation [20], so we did not investigate this aspect in detail, although it is clearly important in the context of climate change. Instead, we assessed the number, mean size and size distribution, shape index (edge to area ratio), and distance to the nearest large (>100 km²) patch for protected patches before and after the proposed expansion from 24.3% to 30% coverage, using Patch Analyst version 5.2.0.16 (developer Rob Rempel; downloaded from https://patch-analyst.software.informer.com/, accessed on 5 December 2021) and the Near tool in the Proximity toolset in ArcGIS. In Thailand, the smaller and/or more isolated PAs rarely support large forest-dependent vertebrates and are more expensive to manage.

2.5. Assessing Importance for Birds and Mammals

For the 80 mammal and 702 bird species for which we had at least 10 independent locations, we used Maxent to predict current and future ranges [21]. We used 10 bioclimatic variables from WorldClim version 1.4 based on averages of 1970–1990 [22]; altitudes downloaded from the CGIAR-Consortium for Spatial Information, CGIAR-CSI version 4.1 [15]; slope and aspect generated using surface tools in ArcGIS; soil pH from ISRIC-World Soil Information version 2.0 [23]; mean tree density per km² from Crowther et al. version 2 [24]; and percentage coverage of forest per km² from the European Space Agency (ESA) GlobCover Version 2.3 [25]. Three replicates were run for each species and the average was used for further analysis. We converted the output into binary presence–absence maps using the 10% cumulative logistic threshold [26] and evaluated the predictive performance of the models using the continuous Boyce Index [27]. Values of this index exceeded 0.5 for all models (median 0.79 for mammals and 0.69 for birds), suggesting an acceptable performance, so all results were used.

Following [28], we scaled representation targets for the percentage of each species’ modeled range in Thailand from 100% for species with total ranges <1000 km² to 10% for species with ranges >250,000 km². We used a linear scale (rather than the log-linear scale used in [28]) for interpolation between these extremes. Thailand only has c. 160,000 km² of forest remaining, so this procedure sets near-impossible targets for a few widespread forest-dependent species, but in the absence of data on population sizes and on non-area-related threats, this conservative approach seems to be justified. We also mapped the density of species with inadequate (i.e., less than target) representation to identify areas that should be investigated for additional protection.

2.6. Assessing Importance for Other Elements of Biodiversity

No one surrogate can comprehensively represent spatial patterns of biodiversity [29]. Therefore, as proxies for other elements of biodiversity—other vertebrates, invertebrates, plants, and ecosystems—we used elevation in four ranges (0–250 m, 250–750 m, 750–1500 m, and >1500 m a.s.l.) following [11], the WWF ecoregions and the WWF Global 200 subset of these, present and projected 2070 bioclimates from [10], and a 1:50,000 map of forest types based on Sentinel-2 and Landsat 8 images from 2018 and extensive ground checking [12]. We used representativeness as a measure of the importance and assessed this in several ways: the total area of the altitudinal range, or ecoregion, or bioclimatic zone, or forest type within the current and future 30% protected area system; the percentage of that category which is or will be protected; the percentage of the total protected area system which is or will be in that category; and the percentage of the remaining natural vegetation within that category which is or will be protected. The first of these indicates how much of that category is protected in total, the second suggests whether it is likely to be an adequate sample of the area of that category in Thailand, and the third shows how much of the whole protected area system it constitutes. The fourth shows if there is an opportunity to improve the representation of that category by increasing the area protected.

2.7. Assessing Vulnerability to Climate Change

For birds and mammals, we ran the same Maxent models, changing only climate, using the same bioclimatic variables projected for 2070 by three CMIP5 Earth System Models, CNRM-CM5, GFDL-CM3, and HadGEM2-ES, with two Representative Concentration Pathways, RCP2.6 and RCP8.5. These RCPs represent low and high greenhouse-gas concentration scenarios, respectively, and thus the potential range of radiative forcing by 2070 [30]. RCP2.6 is consistent with meeting the Paris Agreement’s 2 °C global warming target. As a proxy for impacts on other species and ecosystems, we used changes in the bioclimatic zonation of Thailand between the present and 2070, and the representation of these zones in the protected area system, although these impacts will not be fully realized by 2070.

3. Results

It was possible to reach 29.5% coverage (152,705 km²) by adding only large (>10 km²) forest patches (totaling 26,825 km²) that are adjacent to the existing (115,712 km²) and planned (9968 km²) protected areas and not currently included in the PA system, plus the small area (200 km²) of existing Ramsar sites which is not in the current protected area system (Figure 2). Other forest patches that could potentially be added were too small (<10 km²) or too isolated to meet our selection criteria. They might be of conservation value, but this should be evaluated on the ground. The expanded PA system consists of fewer (608 vs. 670), larger (mean area 251 km² vs. 188 km²), protected patches, and less isolated from each other than in the present system (Table S1 in Supplementary Materials).

All the modeled mammal and bird species are expected to be represented in the present (24.3% coverage) PA system, but only 28% and 26%, respectively, met the representation targets for adequate protection, with these values increasing to 60% and 38% in the expanded (29.5%) system (Table S2 in Supplementary Materials). The highest modeled densities of inadequately protected mammal species were generally in areas adjacent to PAs in the expanded PA system (Figure 3). In contrast, around half of the birds that missed the target were coastal and/or wetland species (SM), reflecting the low representation of these areas in the expanded PA system, even with Ramsar sites included. The highest densities of inadequately protected species are along the coast of the Inner Gulf of Thailand, followed by peri-urban Bangkok and the Central Plains. This region includes a variety of natural, semi-natural, and artificial wetlands that are used by birds, particularly in winter.

For the biodiversity proxies, the expansion from 24.3% to 29.5% substantially increased the area of protected forest in the currently underrepresented Extremely Hot and Moist and Extremely Hot and Xeric bioclimate zones by 23% and 54%, respectively (

Table 1. Representation of bioclimatic zones in protected areas in Thailand currently (with 24.3% of the land area protected) and after expansion (with 29.5% protected). Data from [10].

Bioclimatic Zone	Total Area of Zone (km2)	Area Protected (km2)		% of Zone Protected		% of Total Area Protected		% of Natural Vegetation Protected
		24.3%	29.5%	24.3%	29.5%	24.3%	29.5%	24.3%	29.5%
Extremely Hot and Moist	262,852	74,998	92,101	28.5	35.0	63.1	63.1	83.1	86.3
Extremely Hot and Xeric	200,660	7825	12,037	3.9	6.0	6.6	8.2	69.9	80.2
Hot and Mesic	43,791	31,600	36,529	72.2	83.4	26.6	25.0	92.4	93.7
Hot and Dry	5818	4448	5165	76.5	88.8	3.7	3.5	93.2	93.6
Warm Temperate and Mesic	83	76	81	91.6	97.6	0.1	0.1	100.0	98.8

). These bioclimates are still underrepresented in terms of the percentage of the zone protected, now 35% and 6%, respectively, but there is little more forest to protect in the Extremely Hot and Moist zone and only isolated small patches in the Extremely Hot and Xeric zone. The expansion also substantially increased the protection for two forest types that are underrepresented in the current protected area system; the percentage of deciduous dipterocarp forest protected increased from 42% to 62% (an increase of 47%), with increases in all regions except the south, where it is absent, and the percentage of mixed deciduous forest from 61% to 81% (34% increase), particularly in the north (

Table 2. Representation of forest types in protected areas in Thailand by region, currently (with 24.3% of the land area protected) and with expansion (29.5% protected). Vegetation from the Royal Forest Department [12].

Forest Type	Land Area Protected	% of Total Area of the Habitat Type
		Central	East	North	North-East	South	West	All Regions
Hill Evergreen Forest	24.3%	98	81	61	98	0	97	73
	29.5%	100	75	91	100	0	99	94
Pine Forest	24.3%	81	0	49	61	0	54	54
	29.5%	98	0	53	87	0	67	60
Deciduous Dipterocarp F.	24.3%	43	7	48	24	0	50	42
	29.5%	55	10	72	33	0	74	62
Mixed Deciduous Forest	24.3%	61	57	58	53	94	70	61
	29.5%	77	60	86	64	99	85	81
Dry Evergreen Forest	24.3%	88	84	68	80	0	79	80
	29.5%	97	90	82	85	0	89	88
Moist Evergreen Forest	24.3%	0	75	0	0	78	93	80
	29.5%	0	88	0	0	89	100	91
Savanna	24.3%	96	100	92	88	0	84	91
	29.5%	96	100	100	89	0	85	92
Other forest	24.3%	0	0	0	0	40	21	31
	29.5%	1	2	0	1	43	24	34

). The increased protection for these two bioclimatic zones and two forest types reflects the expansion of protection in the lowlands (

Table 3. Representation of elevational ranges in protected areas in Thailand currently (with 24.3% protected) and after expansion (with 29.5%).

Elevation Range (m)	Total Area (km2)	Area Protected (km2)		% of Range Protected		% of Total Area Protected		% of Natural Vegetation Protected
		24.3%	29.5%	24.3%	29.5%	24.3%	29.5%	24.3%	29.5%
0–250	395,814	25,181	30,151	6	8	17	17	58	65
250–750	169,217	80,106	100,356	47	59	55	57	90	92
750–1500	55,788	38,246	45,773	69	82	26	26	92	94
>1500	1482	1175	1330	79	90	1	1	95	97

). The 0–250 m elevation range is still greatly underrepresented (only 8% protected) after the expansion of the PA system, but most forests within this range will be protected (65%) and the remaining unprotected forest is in small, isolated fragments.

Representation of the 15 ecoregions present in Thailand (Figure S1 in Supplementary Materials) in the expanded PA system varies from 1% for the Chao Phraya freshwater swamp forests to 90% for the Peninsular Malaysian montane rain forests (

Table 4. Representation of the 15 WWF ecoregions present in Thailand in protected areas, currently (24.3%) and after expansion (29.5%).

Ecoregion	Total Area(km2)	Area Protected(km2)		% of Ecoregion Protected		% of Total AreaProtected		Natural Vegetation(km2)	% VegetationProtected
		24.3%	29.5%	24.3%	29.5%	24.3%	29.5%		24.3%	29.5%
Indochina mangroves	9614	691	707	7.2	7.4	0.5	0.4	873.0	36.0	36.8
Central Indochina dry forests	257,114	17,708	24,121	6.9	9.4	12.3	13.7	34,386.0	78.0	84.1
Luang Prabang montane rain forests	23432	12,645	14,272	54.0	60.9	8.8	8.1	14,325.0	87.1	88.6
Myanmar Coast mangroves	4032	759	808	18.8	20.0	0.5	0.5	1829.0	70.2	71.7
Tenasserim-South Thailand semi-evergreen rainforests	75,814	24,421	26,631	32.2	35.1	17.0	15.1	25,739.0	84.0	85.3
Chao Phraya lowland moist deciduous forests	24,282	4080	4560	16.8	18.8	2.8	2.6	4750.0	71.1	74.4
Chao Phraya freshwater swamp forests	46,965	318	389	0.7	0.8	0.2	0.2	603.0	46.5	55.0
Southeastern Indochina dry evergreen forests	18,036	11,189	11,802	62.0	65.4	7.8	6.7	11,050.0	86.0	87.0
Northern Indochina subtropical forests	5312	2297	3153	43.2	59.4	1.6	1.8	3234.0	81.6	87.0
Kayah–Karen montane rain forests	78,552	48,540	60,558	61.8	77.1	33.8	34.3	62,256.0	90.9	92.8
Northern Thailand–Laos moist deciduous forests	37,371	13,880	20,258	37.1	54.2	9.7	11.5	23,633.0	89.9	93.0
Northern Khorat Plateau moist deciduous forests	13,703	1420	1769	10.4	12.9	1.0	1.0	1779.0	66.8	71.6
Cardamom Mountains rainforests	14,348	3551	3998	24.7	27.9	2.5	2.3	4422.0	87.4	88.6
Peninsular Malaysian montane rainforests	824	509	739	61.8	89.7	0.4	0.4	738.0	97.8	98.0
Peninsular Malaysian rainforests	11,579	1716	2571	14.8	22.2	1.2	1.5	2557.0	75.8	83.1

). However, most ecoregions already have >70% of their remaining natural vegetation protected so opportunities for further expansion are limited, with freshwater swamp forests and mangroves the only significant exceptions. For the five Global 200 ecoregions present in Thailand, all have 80% or more of the remaining natural vegetation protected, except for the Indochina dry forests, at 67% (

Table 5. Representation of the five WWF Global 200 ecoregions present in Thailand in protected areas currently (24.3%) and after expansion (29.5%).

WWF Global 200 Ecoregion	Total Area (km2)	Area Protected (km2)		% of Ecoregion Protected		% of Total Area Protected		Natural Vegetation(km2)	% Vegetation Protected
		24.3%	29.5%	24.3%	29.5%	24.3%	29.5%		24.3%	29.5%
Kayah–Karen/Tenasserim moist forests	154,366	72,674	87,161	47.1	56.5	66.3	65.3	87,995	73.5	89.7
Indochina dry forests	275,150	28,861	35,919	10.5	13.1	26.3	26.9	45,436	51.6	67.2
North Indochina subtropical moist forests	5312	2275	3153	42.8	59.4	2.1	2.4	3234	57.9	85.2
Cardamom Mountains moist forests	14,348	3527	3997	24.6	27.9	3.2	3.0	4422	70.1	79.8
Peninsular Malaysia lowland and montane forests	12,403	2223	3304	17.9	26.6	2.0	2.5	3295	54.7	87.4

).

The model projections for mammal and bird distributions under climate change suggest significant declines in protection adequacy, although the expanded PA system is more resilient, particularly for birds (Table S2 in Supplementary Materials). In addition, note the large spread between the projections based on three different earth system models, particularly under RCP8.5. The expansion does little to reduce the projected impacts of climate change on bioclimatic representation since the additional forest is mostly in the warmest zones and there is very little unprotected forest to add in the coolest zones. By 2070, even under RCP2.6, more than 90% of the protected area system will consist of Extremely Hot bioclimates, with the areas of the three cooler zones greatly reduced (Table S3 in Supplementary Materials). Under RCP8.5, the Warm Temperate and Mesic zone will be virtually lost from the protected area system (and from Thailand as a whole), and more than 98% of the protected area will be Extremely Hot (Table S4 in Supplementary Data). Vegetation is also expected to change in response to climate change, but probably with a multidecadal time lag, at least for forests. Moreover, other factors, including soils and fire regimes, are important in determining vegetation types in Thailand so we have not attempted to model these changes.

4. Discussion

Of the three questions posed in the Introduction, the answer to the first—How can Thailand meet the proposed area target of 30% by 2030?—is surprisingly easy, given Thailand’s high population density and intensive use of most of its land area. The total of 29.5%—2600 km² shy of 30%—can be achieved by adding only large (>10 km²) patches of forest adjacent to the existing protected area system, plus a small area of Ramsar sites outside the current system. This could, in theory, be done by administrative changes since these areas are already controlled by government departments. However, local people have made use of these areas for a long time, so any future expansion of the protected area system will involve difficult negotiations. The additional 0.5% needed to bring the total to 30% could be found by relaxing our criteria for size or isolation, but it makes more sense to use this additional area to plug gaps in the forest-dominated protected area system, as discussed below.

The second question—will the PA system be ‘well-connected’ and ‘important for biodiversity’?—has a more complex answer. The proposed expansion will increase both patch sizes and connectivity, particularly in the north and west, but the PA system will still be highly fragmented. Further increases in connectivity will mostly require extensive ecological restoration and, in some cases, the acquisition of private land. We assessed the importance for biodiversity in several ways. All the mammal and bird species we modeled were present in the system even before expansion, but only 60% of mammals and 38% of birds met our, admittedly ambitious, targets for adequate protection of their ranges. Moreover, despite improvements, the expanded protected area system will still underrepresent the lowland area <250 m a.s.l., the Extremely Hot and Xeric bioclimate zone, deciduous dipterocarp forest, and several ecoregions, including the Indochina dry forests Global 200 ecoregion, which is considered a global priority for conservation. Yet the percentage of natural vegetation protected is high (mostly >70%) in most of these categories, so opportunities for increasing protection without substantial ecological restoration area are limited. Moreover, where the percentage of natural vegetation protected is <70%, this is usually because the unprotected area is in numerous small and isolated patches.

Note that the focus in this study on the CBD’s quantitative, area-based targets, and the use of easily mappable proxies for biodiversity inevitably understate the importance of the quality components of protected areas and their management. We could not map poaching of birds and mammals, harvesting of firewood, mushrooms, and other forest products, and the fires and other disturbances associated with illegal use. Protected area management in Thailand is good by regional standards and improving, but managers in developing countries must deal with a wide range of human pressures.

Both the analyses and the literature suggest several situations where opportunities for expanding protection may exist but could not be mapped from the data sources we used. First, we ignored small (<10 km²) forest patches, but such patches may support viable populations of narrow-range plant, invertebrate, and small-vertebrate species, and act as stepping-stones between larger patches, and thus deserve protection. Experience elsewhere suggests that the presence of narrow-range endemics is most likely for communities on extreme substrates, such as limestone karsts and ultramafic rocks [31]. In support of this, many of the new species recently described from Thailand come from limestone karsts [32,33,34]. More surveys of such areas in Thailand will probably identify additional sites in need of protection. Second, the ‘other forests’ category in Table 2 and the Chao Phraya freshwater swamp forest and Indochina mangrove ecoregions in Table 4 will be underrepresented in the expanded system and still, apparently, have opportunities for expansion. Again, these are mostly small areas that need to be identified from ground surveys. In the highly underrepresented Chao Phraya freshwater swamp forest ecoregion, for example, a recent survey [35] found small, unprotected patches of floodplain forest, freshwater swamp forest, and mangrove in the rapidly urbanizing landscape along the Chao Phraya River.

The biggest opportunities for additional conservation, however, are likely to be found in non-forest systems along rivers, floodplains, and the coastline (Figure 3a). The hotspot of inadequately protected bird species coincides with the sprawling urban conglomeration of Bangkok, the increasingly developed margins of the Inner Gulf of Thailand to the south, and Thailand’s most productive agricultural land to the north. In this area, different bird species, and other wildlife, depend on intertidal mudflats, salt pans, aquaculture ponds, ‘pond and fill’ land developments, and the remnants of the once extensive open marshlands of the central plains [36,37,38,39].

The vulnerability of species dependent on Thailand’s most valuable urban and agricultural land is highlighted by the global extinction from open marshlands in this region of Schomburgk’s deer (Rucervus schomburgki) in 1932 [40] and the more recent probably extinction of the white-eyed river-martin (Eurochelidon sirintarae), last seen in 1978 [41]. There are still opportunities for establishing new protected areas in this region, as well as along both coastlines to the south, but in other places conservation will require agreements with landowners and developers, and the management of artificial habitats meeting the needs of particular bird species and groups of species [39]. However, although these less-formal methods of wetland conservation may be the only practical approach in many areas, the lack of long-term security probably rules out their recognition as ‘other effective area-based conservation measures’ (OECMs) [8].

The final question we asked was: How vulnerable will this 30% be to climate change? The answer to this question is mixed. The major projected impact of climate change by 2070 is the upwards shift of all bioclimatic zones and, with a time lag, vegetation types and their associated biota [10,42]. Most of the area added to achieve 30% is <750 m a.s.l. so it will not directly help with this issue. There are no opportunities to expand the protected area system to higher altitudes. However, both the increases in the area of many existing protected areas and the increases in connectivity for some, are expected to help reduce the adverse impacts of climate change by facilitating movements of species within and between areas [42].

How typical is Thailand of the biodiversity-rich tropical countries? Although the most diverse include some low-income countries (Democratic Republic of the Congo, Madagascar), for whom biodiversity has a low priority, most are now lower-middle-income (Philippines, Papua New Guinea, Sri Lanka) or upper-middle-income (Brazil, Colombia, Ecuador, Indonesia, Malaysia, Mexico, Peru, Venezuela) economies (https://data.worldbank.org/country) (accessed on 2 March 2022). Most of these middle-income countries are now increasingly urbanized and have a large middle class. Rising wages and alternative employment reduce the pressure from agriculture on marginal land, while economic security, education, and urban living increase support for conservation. However, currently most lag well behind Thailand in protected area coverage.

A comparison between Thailand and the EU is also instructive. As in Thailand, expanding the existing protected area coverage in the EU to 30% by 2030 is achievable by protecting currently unprotected natural areas, but this strategy will leave some ecoregions underrepresented [9]. However, in the EU, only 3 of 41 ecoregions do not have enough unprotected natural or semi-natural habitat to achieve 30% coverage, while, in Thailand, 8 out 15 ecoregions do not have enough habitat left (Table 4). In the EU study, the authors propose the restoration of semi-natural habitats on arable land in these ecoregions [9]. Restoration on a much larger scale that would be required to fully meet 30% ecoregional targets in Thailand will probably be impractical for the foreseeable future but targeting restoration of areas that extend and connect existing fragments of underrepresented ecoregions would be worthwhile and should be considered.

5. Conclusions

In conclusion, Thailand is well-positioned for a ‘30% by 2030’ terrestrial protected area target in terms of the total area under formal protection. The qualitative targets are more difficult to meet, but, overall, the 2030 protected area system proposed in this study appears to do a good job of protecting the remaining forest biodiversity in Thailand. Improvements in forest protection are undoubtedly possible, but these will require intensive surveys on the ground. The situation with non-forest biodiversity is much less positive, but this reflects the reality in East Asia, that coasts, estuaries, and marshy floodplains were the best sites for intensive rice cultivation and, more recently, for urbanization and industrialization [40]. In Thailand, the recently announced National Strategy 2018–2037 includes, in a section on sustainable growth, targets of 35% coverage of natural forests by 2033–2037, as well as 15% economic forests, and 5% urban and rural green areas [43]. We hope that this study can contribute to these goals as well as the CBD’s global targets.