Identifying the protection status of vulnerable and endangered plant species in the Tropical Dry Forest of Colombia

(1)

Identifying the protection status of vulnerable and endangered plant species

in the Tropical Dry Forest of Colombia

Bachelor thesis Reineke van Tol Future Planet Studies Major Biology University of Amsterdam Supervisor: Dr. Suzette G.A. Flantua

Examiner: Dr. Daniel Kissling Project leader: Dr. Mauricio Diazgranados

(2)

2

Abstract

Colombia comprises many of the world’s biodiversity hotspots. Many of these areas are highly threatened and rapidly losing valuable species. The tropical dry forest (TDF) is the most threatened and neglected ecosystem, while harbouring unique and very vulnerable plant species. The conservation status of these species is to a great extend unknown. In this study, a new method is applied to identify Important Plant Areas (IPAs) for conservation that contain a significant amount of Prioritized Plant Species (PPS). Through a spatial analysis in ArcGIS, the protection status of the PPS and current threats are analysed. The results of this study show that the currently protected areas are not sufficiently protecting PPS in the Colombian dry forest. The study also shows that the newly proposed method can be applied in several ways. Climate change, coal mining and land-use seem to be the most important threats for TDF-PPS in Colombia. Results of this study contribute to the conservation of Colombia’s rich and unique biodiversity as well as to the establishment of a new IPA analysis method.

Keywords: Colombia, Conservation, GIS, Important Plant Areas (IPAs), Prioritized Plant Species (PPS), Threats, Tropical dry

(3)

3

1. Introduction

Colombia has been identified as one of the 17 megadiverse countries in world, harbouring many biodiversity hotpots (Mittermeier et al., 1997). This means that the country holds a great variety of endemic species, while suffering from severe losses in species and habitat in the meantime (Meyers et al., 2000). These hotspots span over a wide geographical range and comprise a high variety of biomes. In general, all of them have been prioritized for conservation, but proper management is still lacking in many places (Brooks et al., 2002). Moreover, Knight et al. (2008) reported that although there is a lot of knowledge on nature conservation, this knowledge is often not translated into action.

Within the range of tropical ecosystems, the tropical dry forest (TDF) is the most threatened (Wilson, 1988). Miles et al. (2006) identified major threats for dry forests on a global scale, namely climate change, fire, deforestation conversion for agriculture and human population growth. Armenteras et al. (2003) reported that, although a major part of the fragmentation is natural, tropical dry forests are highly degraded and fragmented and over 98% is unprotected. Less than 2% is sufficiently intact to deserve attention from traditional conservationists (Wilson, 1988). Overall dry forests have been poorly studied and have received substantially less attention for conservation than other neotropical biomes. Also, spatial representation of biome maps, mainly of extremely fragmented areas, is highly insufficient to represent the TDF (Särkinen et al., 2011).

The TDF forms a unique habitat due to the fact that it experiences a prolonged dry season of 4 to 7 months. Where the forest is uniformly green during the wet season, during the dry season it changes into a highly heterogenic system. Differential drying rates of vegetation and soil types give rise to a high variety of different niches that are only suitable for very specialized species that can handle the harsh circumstances (Wilson, 1988). Banda et al. (2016) confirmed the fact that TDF species are highly specialized and found that dry forest floristic groups experience high species turnover at relatively small spatial scales, resulting in extremely high species endemism. Only a few species are widespread or shared across dry forest fragments, meaning that losing fragments would result in major loss of biodiversity. Also, dry forests provide important ecosystem services, including soil stabilization, prevention of erosion and desertification, water regulation and contribution to productivity of agriculture and livestock farming (Pizano & García, 2014).

Colombia comprises a significant proportion of these poorly protected but highly biodiverse dry forest fragments (Portillo-Quintero & Sánchez-Azofeifa, 2010). Over 90% of the Colombian TDF has been lost due to deforestation for agriculture. Of the remaining forest, less than 5% is formally protected (Pizano & García, 2014). However, although fragmented and highly unprotected, the Colombian dry forest harbours a great variety of endangered and vulnerable plant species (Pizano & García, 2014). Little is known about the effective conservation status of these species, making proper conservation management impossible.

One way to assess the protection status of prioritized plant species (PPS) is to identify important plant areas (IPAs) for conservation based on the spatial distribution of PPS. This was done mainly for IPAs in Europe, but not yet for the tropics of South America. The Humboldt institute for biological research (IAvH, 2017) proposed a new method to identify IPAs in Colombia, based on the method of Anderson (2002) that was applied mainly in Europe. According to this method, an area is considered an IPA if it holds a significant percentage of the total plant population at local, regional or global scale. If IPAs lie significant outside protected areas, this is a major indication for poor conservation status of PPS in this study. I will use this approach to identify the conservation status of dry forest PPS in Colombia. Hereby contributing to the recently initiated Colombia Bio programme that aims to conserve Colombia’s rich biodiversity.

The aim of this research is to investigate whether PPS in dry forest systems in Colombia currently fall under protection of officially protected areas and whether the newly addressed method is

(5)

5 suitable to address this question. Additionally, this study addresses sources of threats for PPSs falling both in and outside protected areas. Knowledge on threats and the conservation status of PPS is important for effective protection strategies of species in the tropical dry forest. Results of this study contribute to the conservation of one of the most neglected and rapidly disappearing ecosystems in Colombia. Moreover, this study contributes to the establishment of a new method to identify important plant areas for conservation.

One main question and two additional questions are dressed in this research. The main question is: Are current protected dry forest areas in Colombia preserving the most important populations of PPS for conservation? Two additional questions are addressed, being (1) If PPS are not sufficiently protected, in what type of habitat do these unprotected PPS occur? And (2) What might be threats to populations of PPS both in and outside protected areas? Based on the literature discussed above, I expect the conservation status of TDF-PPS to be poor, meaning that a significant proportion of PPS falls outside protected areas. Also I expect that current land use practices and related deforestation for economic incentives form a major threat to the TDF-PPS.

2. Methods

2.1 Main question: Are current protected dry forest areas in Colombia preserving the most

important populations of PPS for conservation?

2.1.1 Data collection

PPSs were identified using dry forest species lists from Pizano & García (2014) and the Colombian biological catalogue CEIBA (Catalogador de Información Biológica). According to Anderson (2002), species indicated with CR (critically endangered), EN (endangered) and VU (vulnerable) are considered as PPS. Two categories of PPS were identified, based on the proposed method of IAvH (2017). Globally threatened (A1) species (16 species) were based on IUCN red list registration (‘lista roja’), whereas locally threatened species (A3) (21 species) were based on Colombian red lists (‘libro rojo’). The second category (A2) includes regionally threatened species, but unfortunately no regional red lists of tropical dry forest plants exist. Category A2 is therefore left out of this study.

Distribution data of the identified PPS was obtained from the Global Biodiversity Information Facility (GBIF) and the Botanical Information and Ecological Network (BIEN). All data was processed in Excel (2016) and checked for irrelevant or incorrect records. Data without spatial reference was deleted. In total 11508 occurrences were left for analysis.

Most GIS layers were obtained from Sistema de Información Ambiental de Colombia (SIAC) and some additional sources (Appendix 1 Table 1).

2.1.2 IPA criteria

Criteria used for IPA analysis are as proposed by IAvH (2017):

• For globally threatened species (A1) an area is considered an IPA if it holds ≥ 1% of the global population of one or more PPS.

• For locally threatened species (A3), an area is considered an IPA if it holds ≥ 10% of the national population of one or more PPS.

(6)

6 “If for A1 there isn't any UA with ≥1% of specimens for species "x", or when there are less than 10 specimens in total, we select the 5 UA with most specimens. The same for A2 and A3.”

For the data in this study, this rule seems to be redundant. If species are very rare, their threshold value will be very low as well. In real practice, IPA thresholds often turned out to be 1 and therefore an area would already be considered an IPA if it contained 1 record of a specific species. For species that did appear in the IPA list, often less than 5 AUs were an IPA for a particular species. Selecting 5 AUs for species that do not meet the threshold values would therefore give a distorted representation of the situation. In many cases there are not even 5 AUs where a species is present at all. This rule was therefore left out of the analysis.

As proposed by IAvH (2017), each separate fragment of the TDF (each AU) was considered as an area to which the IPA rules were applied. Occurrences (in analysis A-C) or total area (analysis D) were used as surrogate population sizes, since actual population sizes are unknown.

2.1.3 Analysis methods

Applying the proposed IPA method to the TDF data turned out to be possible in several ways, that could each be relevant. I used 4 different methods to see whether the results differ for different applications of the proposed method. In this section a short description and relevance of each method is given. The exact steps in each method follow in the next section.

A. Whole dataset

The first analysis was done on the whole dataset (11508 records). In the data many records exist with the same geographic coordinates. However, it is not clear whether these records represent populations measured at the same geographic location or that they are duplicates that should be eradicated. For the analysis this might not be a problem since also the calculations of relative values (number of occurrences of a species in a TDF analysis unit relative to its global or national extend) was done on the data containing duplicates. In a relative value this should not matter. However, I repeated the analysis for the data without duplicates to see whether there is a difference.

B. No duplicates

Since it is not sure whether duplicate records are actual errors, a second analysis was done in which records with exactly the same coordinates were removed. Now, from the original 11508 records, 5915 remained. Further treatment was the same as in A.

C. Buffers

In analysis method A and B, only records falling exactly inside the TDF layer were included in the analysis. However, many occurrence records appear to fall just outside the TDF layer or in the little gaps between the forest fragments (Figure 1). These records are in the above outlined steps excluded for IPA analysis. However, due to potential georeferencing inaccuracies these records might actually belong to the TDF. Also, in the occurrence data it is not clear if single occurrences represent unique observations or local populations of an individual. Moreover, the data only provides presence records, no absence data. Therefore in the next steps a buffer is used to correct for these uncertainties. Schulman et al. (2007) suggested buffering as the most valid correction method to handle these uncertainties. How large a buffer radius must be depends on the dispersal ability of the species (Schulman et al., 2007). However, since the data contains many different species, one standard radius of 1,5 km has been chosen, based on previous studies that used radii of 0,75 – 1,75 km (Del Valle et al., 2004) and because the remaining TDF is so fragmented that often little

(7)

7 interspace in the range of 0,1 – 2 km exists between fragments. As a result of the buffer, records falling just in between forest fragments are now incorporated (Figure 2).

D. Area calculation

Usually, an analysis based on buffered occurrences uses surface area calculation instead of counts (Flantua, S.G.A., personal communication, May 2017). However, for this analysis this might be less accurate since TDF analysis units are so small that thresholds values based one dispersal area will in many cases never be met (Appendix 2). Nonetheless, to give a complete list of options I conclude with an analysis using area instead of occurrence count.

2.1.4 Practical steps

All spatial processing steps were performed in ESRI ArcGIS 10.4.1. For analyses A and B, all layers were projected to GCS_WGS_1984 to make them compatible for analysis (using the PROJECT tool). For analyses C and D layers were projected into Bogota_UTM_Zone_18N in order to make calculations on the buffered records.

Layers of biosphere reserves, national and regional parks and areas protected by civil society were merged into one layer of protected areas (using the MERGE tool). From the world borders layer, Colombia was selected and made into a separate layer.

Figure 1. Occurrences falling just in between forest fragments

(8)

8 Occurrence records were then displayed in Arc Map (using the option 'display x,y data') and made into a point shapefile. For analysis B,C and D duplicate records were removed (using the ‘delete identical’ tool, fields: x, y). For analysis C and D, buffers of 1,5 km were created around each occurrence, using the BUFFER tool. There was corrected for overlap between buffered records using the ‘dissolve LIST’ option in the buffering tool, dissolving on species, TDF ID and occurrence ID. Occurrence data was intersected with the Colombia borders layer to be able to distinguish between global and national occurrences (using the INTERSECT tool).Then either buffered or non-buffered occurrence layers were intersected with the TDF layer to see which occurrences actually fall inside in the current extend of the TDF (using the INTERSECT tool) (Figure 3).

Figure 3. Overview of the total distribution of PPS, the current extend of the TDF, the protected areas

(9)

9 IPA thresholds were calculated per species according to the criteria described in section 2.1.2 (Appendix 2). Total occurrences were used as surrogate population sizes, since actual population sizes are unknown. For analysis D, the area of each species in total and in the TDF fragments was calculated (using the ‘calculate geometry’ option in the attribute table) in order to calculate the IPA thresholds. Areas meeting the IPA criteria were selected and converted to a separate shapefile.

IPAs were then intersected with the merged protected areas layer. With the result, two separate shapefiles were made for respectively protected and unprotected IPAs. However, often only parts of IPAs intersecting with protected areas were actually protected (Figure 4), so for each IPA there was identified if the particular species that makes the area an IPA was protected. If so, the area was identified as an protected IPA. In the case of the figure the area was considered as unprotected.

Figure 4. IPA (purple) intersecting with protected areas (blue), but PPS (green) are not protected.

2.1.5 Analysis

Output data was exported to Excel (2016) and analysed in pivot tables. A Chi-square goodness of fit test was performed in R studio to test whether the results are significant (α = 0.05). For analysis A and B, the assumptions for a Chi-square test were not met (more than 20% of the expected values is less than 5), so a binomial test was used as non-parametrical alternative. The hypotheses being tested are: H0: Conservation status is neither poor nor good (IPAsin = IPAsout)

(10)

10 HA: IPAsin ≠ IPAsout

- If IPAsin > IPAsout, the conservation status is positive - If IPAsin < IPAsout ,the conservation status is negative

2.2 Sub-questions 2 and 3: Land use and Threats

For sub-questions 2 about land use in and around IPAs, maps of ecosystems and projected land use were overlaid with the IPA results. For sub-question 3 about threats for IPAs, first literature research was done to identify threats, after which GIS data on specific threats was overlaid with the IPA results. Used layers can be found in Appendix 1 Table 2.

3. Results

3.1 IPAs

In total 3,15% (37/1174) of all dry forest species deserves the PPS status (Figure 5), thus being classified as Critically endangered, Endangered or Vulnerable (based on the CEIBA dry forest plant list of in total 1174 species) (Appendix 3). However, not all of these species occur within the remaining tropical dry forest of Colombia, or at least they were not registered in the databases. For Acoelorraphe wrightii and Cattleya quadricolor (A3), no occurrences were found within Colombia. For Libidibia ebano, no occurrences were found at all. Only 21 out of the 37 PPS were found in the current TDF (Figure 5). Sixteen species were not found in the current TDF and could thus not be involved in the IPA analysis.

Figure 5. PPS in current TDF relative to total PPS and total dry forest species.

As discussed in 2.1.3, four different methods were applied to test the new IPA analysis method proposed by the IAvH (2017). The section below provides the results of the IPA analysis per method. Detailed results are provided in Appendix 4. R scripts are given in Appendix 5.

A. Whole dataset

In analysis A, the whole dataset was used for analysis, including duplicate records. There were no buffers created around occurrences, so only point exactly falling inside TDF AUs were incorporated. This analysis results in 7 areas meeting the IPA criteria for 10 different species. There are 3 areas (43% ) that are protected, whereas 4 areas (57%) are unprotected (Figure 6) (Appendix 4 Table 4.1). A binomial test points out that this result is insignificant (p = 1.00) (Appendix 5A).

1174 TDF

species • 100%

37 PPS • 3,15 %

21 in

(11)

11

Figure 6. Protection status IPAs based on method A. Non-significant (p = 1.00).

B. No duplicates

In this analysis method, records with exactly the same coordinates were removed (Figure 7). Compared to method A, now 51,4% of the records were deleted. This gives other threshold values (Appendix 2) and also generates different results (Figure 8) (Appendix 4 Table 2). Now 8 IPAs were identified for 8 different species. Three of these (37,5%) are currently being protected, five (62,5%) are unprotected. This result is not significant (binomial test: p = 0.73) (Appendix 5B).

Protected 43% Unprotected

57%

PROTECTION STATUS IPAS A

7 IPAs 10 species 3 protected 4 unprotected

Cleaned

dataset

• 5916 records

Colombia

• 824 records (14%)

TDF

• 59records (1%)

Whole

dataset

• 11508 records

Colombia

• 2523 records (22%)

TDF

• 396 records (3%) Removing duplicates: - 51,4 %

(12)

12

Figure 8. Protection status IPAs based on method B. Non-significant (p = 0.73).

C. Buffers

In analysis C, a buffer was created around each occurrence, causing more records to fall in the TDF layer. This also gives much more IPAs. In total 46 IPAs were identified for 14 different species. Eight areas (17,4%) are protected, 38 (82,6%) are unprotected (χ² = 19.57, df = 1, p = 9.72*10-6_{) (Figure 9)} (Appendix 4 Table 3).

Figure 9. Protection status IPAs based on method C (p = 9.72*10-6_).

D. Area

Just as in method C, buffers were created around each record, but now the total surface area of each occurrence was used as calculation method instead of counts. This generated again different threshold values (Appendix 2) and gave quite different results as well (Appendix 4 Table 4). Now 12 IPAs could be identified for 7 species. Note that in this case only A1 species appear in the list. Two of these IPAs (16,7%) already have a protection status, whereas ten areas (83,3%) are unprotected (χ² = 5.33, df = 1, p = 0.021) (Figure 10).

Protected 37% Unprotected

63%

PROTECTION STATUS IPAS B

Protected 17%

Unprotected 83%

PROTECTION STATUS IPAS C

8 IPAs 8 species 3 protected 5 unprotected 46 IPAs 14 species 8 protected 38 unprotected

(13)

13

Figure 10. Protection status IPAs based on method D (p = 0.021)

3.1.1 Integrated

Each of the four different methods generates quite different results with some similarity (Table 1). The IPAs that are shared between the different methods (22, 546, 670, 745 & 1363) and particularly the ones that are unprotected (22, 670, 1363) may deserve most attention for now. These areas are highlighted in Figure 11.

Table 1. Difference and similarities in results between the 4 different IPA methods.

IPA A B C D Total Protected?

22 x x x x 4 No 546 x x x x 4 Yes 670 x x x x 4 No 745 x x x x 4 Yes 1363 x x x x 4 No 375 x x x 3 No 889 x x 2 No 4 x x 2 No 118 x x 2 No 21 x x 2 No 1080 x x 2 No 1152 x x 2 No 1616 x x 2 No 1623 x x 2 No 1625 x x 2 No 1530 x 1 No 6 x 1 No 121 x 1 No 126 x 1 No 353 x 1 No 357 x 1 No Protected 17% Unprotected 83%

PROTECTION STATUS IPAS D

12 IPAs

7 species

2 protected 10 unprotected

(14)

14 361 x 1 No 451 x 1 No 453 x 1 No 459 x 1 No 553 x 1 No 556 x 1 No 576 x 1 No 693 x 1 No 776 x 1 Yes 777 x 1 Yes 844 x 1 No 851 x 1 Yes 882 x 1 Yes 1094 x 1 No 1107 x 1 No 1108 x 1 No 1110 x 1 No 1111 x 1 Yes 1153 x 1 No 1393 x 1 No 1537 x 1 No 1613 x 1 No 1614 x 1 No 1628 x 1 No 1655 x 1 No

(15)

15

Figure 11. The three IPAs shared between the different IPA methods. Highlighted: IPA 670: Cesar (top), IPA 22: Santander (middle), IPA 1363: Cundinamarca (bottom)

(16)

16

3.2 Land use and threats

The GIS analysis on land use and threats was performed on the three highlighted areas: 1. IPA 670 Cesar

2. IPA 22 Santander 3. IPA 1363 Cundinamarca. 3.2.1 Land use

In the second sub-question I asked: If PPS are not sufficiently protected, in what type of habitat do these unprotected PPS occur?

Although the landcover map reflects the situation in 2008, analysis gives a good impression of the land use in and around the three unprotected IPAs (Figure 12). Results of the spatial analysis for land use point out that only small parts are actually assigned as natural forest, whereas major part consists of pastures (Cundinamarca), annual or transitional crops (Santander) and grassland (Cesar). Also, all IPAs consist of or are surrounded by secondary vegetation, which occurs on lands that have been deforested and that are now recovering. Especially in Santander, the IPA lies very close to an urban area. All these land use types point out that these dry forest areas are being exploited quite intensively, or at least they have been in the past. This poses a threat to the PPS within these IPAs. Other threats for PPS, both within and outside protected areas are discussed in the next section.

3.2.2 Threats

In the third sub-question I asked: What might be threats to populations of PPS both in and outside protected areas? This question was answered through literature research and a GIS analysis of which the results are presented in this section.

(17)

17 3.2.2.1 Protection

Several categories of protected areas are found in Colombia that conserve nature in slightly different ways. For the TDF fragments that fall within protected areas these are biosphere reserves, national parks and reserves protected by civil society. According to the IUCN categories of protected areas, biosphere reserves are category 1 protected areas that aim to protect biodiversity and geomorphology with the least human intervention. Biosphere reserves are usually not freely accessible and economic activities are strictly controlled. National parks are categorized as category 2 protected areas, that receive somewhat less strict measures. National parks aim to protect biodiversity as well as to promote education and recreation. However, much more human intervention is allowed than in category 1 areas (IUCN, 2017). The last category involves the privately owned biodiversity areas that are protected by civil society. Within the IUCN system these areas fall in the category of least protection (category 6 – multi-use) (IUCN, 2017). However, especially for small, fragmented areas such as the remaining dry forest patches, this type of protection is very important and effective. In these reserves biodiversity is protected by the local community, combined with sustainable resource extraction and ecotourism (Parques Nacionales Naturales de Colombia, 2017).

Although the system of protected areas is quite extensive, contradicting information is published when it comes to the effectiveness of protection within these areas. According to Murcia et al. (2013), actual protection is often weak, due to the restricted human and economic resources that are available for it. However, Bruner et al. (2001) concluded that national parks are more effective in protecting tropical biodiversity than is generally thought. National parks scored especially good on prevention of land clearing compared to the surroundings. However, logging, hunting, fire and grazing still turned out to be a threat within the boundaries of the parks, although less than outside. The actual protection status of IPAs categorized as ‘protected’ is thus somewhat unsure and needs more research.

3.2.2.2 Global threats and climate change

On a higher level, globalization, increasing human population and changing consumption patterns pose great risks to tropical biodiversity. More ecosystem goods and services are demanded both for local and global use, which increases pressures on ecosystems. Lands are mainly cleared for agricultural and mining purposes (Murcia et al., 2013). Tropical dry forest soils are especially suitable and easy to clear for agriculture and cattle pastures (Murphy & Lugo, 1986). Much of the TDF in Colombia has already been transformed into cattle pastures and farmland and the remaining patches are still being cleared for this purpose (Murcia et al., 2013).

Climate change is another threat that poses a risk for both protected and unprotected IPAs. Miles et al. (2006) found that 37% of TDF in South America is at severe risk of the projected drier climate. These changing environmental conditions could result in even more loss of TDF, however also other areas might become suitable instead. Species may therefore move out of protected areas to new habitats (Hannah et al., 2007). Still, the question is if rare PPS are able to do this before they go extinct, since the rate of future climate change is likely to exceed the migration rate of most plant species (Neilson et al., 2005). Although the exact risk of climate change is challenging to predict, the expected and possible effects make it important to incorporate consequences of climate change into future conservation planning of TDF (Jones et al. 2016). For the three focus IPAs that emerged from this study the vulnerability for projected climate change is significant (Figure 13). From the figure can be deduced that projected vulnerability for climate change (2011-2040) is especially high or even very high in

(18)

18 Cundinamarca (IPA 1363) (A). In Cesar (IPA 670) (B) climate vulnerability is medium to high and in Santander (IPA 22) (C) it ranges between low and high.

3.2.2.3 Mining

The coal mining industry in Colombia belongs to the largest of the world and is still expanding (Huertas et al., 2011). More and more land is being cleared for the mining activities, which threatens ecosystems and impacts local communities (Moor & van de Sandt, 2014). To analyse the threat of coal mining on PPS in the TDF, a layer showing the extend and intensity of coal mining in 2016 was overlaid with the layer showing the three focus IPAs (Figure 14). Coal mining concentrates mostly around the departments La Guajira and Cesar, which is where IPA 670 occurs. IPA 1363 lies right next to a coal mining area. Although the focus IPAs do not fall right into mining areas, many TDF patches do and coal mining is expected to increase. Therefore coal mining must be considered as serious threat for TDF-PPS.

Gold mining is another source of deforestation and pollution, mainly due to mercury dumping that is related to this mining practice (Morales, 2017). The spatial analysis shows that the three IPAs do not directly fall in areas where gold is actively being mined (Figure 15). However, since gold mining is largely an illegal business, it is not sure whether the representation is reliable.

(19)

19

(20)

20 3.2.2.4 Coca production

Other threats for biodiversity that are of special concern in Colombia are mainly illegal practices related to the long-lasting conflict period, logging, oil dumps and large scale coca production. Also illegal gold mining belongs to this list. These practices formed a major source of income for armed groups, but have at the meantime been a source of water pollution, soil degradation and deforestation (Morales, 2017). In this section I highlight coca production, since this is the largest illegal agribusiness in the world, with Colombia as number 1 producer. Moreover, illicit coca plantations have devastating effects on forests and local communities (Rincón-Ruiz & Kallis, 2013).

Forest clearing for coca production has severely increased during the period of armed conflict. Apart from deforestation, the production of cocaine uses harmful chemicals that end up in the water, the soil and impacts the remaining, adjacent ecosystems (Morales, 2017). Also, formerly local markets expanded to global export, which greatly increased the pressure on biodiversity (Fjeldså et al, 2005; Davalos et al., 2011). Although coca production forms a major threat for tropical biodiversity, for the

(21)

21 areas highlighted in this study it might be less important. In figure 16 can be seen that coca production focusses in other areas than where the IPAs of Cesar, Santander and Cundinamarca occur. However, for other remaining dry forest areas, coca production must still be considered as a threat, since this is such an important source of deforestation.

Although not studied yet, the recent peace agreement of the Colombian government and the Revolutionary Armed Forces of Colombia – Popular Army (FARC-EP), can be expected to decrease the urge illegal practices such as for coca production. However, since the markets have expanded enormously, eradicating production will be a hard task, if not impossible. For the TDF PPS in particular, official protection is important to decrease threats from illegal practices. Deforestation for coca plantations was found to be quite effectively prevented by official biodiversity protection (Davalos et al., 2011).

3.2.2.5 Peace

Many biodiversity areas have been impacted because of the decades of conflict, especially through illicit logging, mining and coca cultivation (Morales, 2017). However, also in many areas biodiversity has profited from the fact that people were displaced in areas occupied by the FARC-EP and other guerrillas and paramilitaries. Recently a peace agreement between the Colombian government and the FARC-EP has been signed. Although the peace will have some positive effects for biodiversity,

(22)

22 an ecological downside of the agreement might also exist. The expected post-conflict development in formerly abandoned areas might lead to exploitation of relatively conserved areas after re-settlement. Socio-economic activities, such as agricultural practices, mining and logging may form a threat to PPS in areas that were formerly abandoned from society and now become available again for development (Negret et al., 2017; Morales, 2017).

One essential key to successful environmental protection from economic development seems to be involvement of the local community. Active participation of local communities stimulates sustainable use and management of environmental resources (Andrade & Rhodes, 2012). So although areas protected by civil society are categorized as the least protected areas by IUCN, they could in fact largely contribute to conservation of IPAs. In an additional study I elaborate on the possible ‘threat’ of post conflict development for TDF-PPS, in which also the active participation of local communities is incorporated (van Tol, 2017).

4. Discussion

4.1 IPA method

One interesting finding in my study is that the newly proposed method for identifying IPAs can be interpreted and used in several ways (method A-D). Each method generates quite different results and conclusions would thus depend on the method that is applied. Therefore I start first with an argumentation to choose the best method to draw conclusions from.

First of all, it is not sure whether duplicate records (as in analysis A) represent populations or errors. Therefore, it might be useful to eradicate them, since then it is at least sure that overlap between databases is cleared and all species are treated equally. For that reason, analyses methods B-D would be more accurate. Creating buffers around occurrences seems to be a good option to correct for uncertainties in georeferencing and data collection, as well as to incorporate some degree of dispersal ranges. It is still unclear how large these buffers must be for a dataset with multiple species. Probably sensitivity analyses could be used to select the most appropriate buffer size. Also more specific methods, using dispersal information of each species could be used to come to more optimal buffer sizes. These techniques laid beyond the scope of this research, but would be valuable for follow-up studies. Still, the arguments to use buffers hold and thus leave methods C and D as best approaches. Then, for this dataset it seems to make more sense to use a counting method (C) than a surface area calculation method (D), since analysis units were in most cases too small to ever meet the IPA criteria for surface area.

The highly fragmented state of the forest gives some additional implications to the proposed method. Because the forest is so fragmented, analysis units are generally very small. This resulted in very few IPAs in the end, especially because threshold values often were hard to meet due to the limited size of patches. Moreover, areas that do meet the criteria and are thus important for conservation, are very small. It would be a valuable addition to use larger analysis units. Larger IPAs wold also improve conservation of the forest.

From this study I would recommend method C, in which duplicates are removed, buffers are created and thresholds are calculated based on occurrence numbers, as the best approach. However, for other datasets and other biomes, other methods might be more appropriate. To optimize the method as proposed by the IAvH, more test studies are needed.

4.2 Identifying threats

Results of my study show that several threats exist for TDF-PSS. Current land use practices already pose a threat to the tropical dry forest and its PPS and these practices are expected to expand in the

(23)

23 future. However, to make a prediction of this threat for the future, a more detailed study on the expected land use change is needed. The identified threat of climate change is very important, but also very hard to handle, since there is no direct influence on it. Evacuating PPS to another area that would be less impacted, might be a possibility. To better understand the threats of mining and coca production, on site studies would be needed. Since these practices occur illegal on a large scale, it is hard to identify the threat from a distance. Also more attention is needed for the potential threats that emerge from socio-economic development related to the post-conflict period. For areas that are formally protected, it is important to do on-site studies to the effectiveness of this protection.

4.3 Data availability & Accessibility

Due to the fact that TDF is scientifically quite unknown, not much data is available on this ecosystem type. I had to work with the data that is available, which has some implications for this research. First of all, since PPS are (critically) endangered or vulnerable, they will also be rare and hard to find. For sure, not all existing PPS are therefore present as a record. Also, the datasets provided present-only records; it is therefore not sure that PPS are not present in areas where they were not found. Moreover, many of the plant occurrence records did not have a date or were dated before 1950. It is very likely that these occurrences are no longer accurate, since deforestation took place in many places or plants just died naturally. I did not eradicate old data points because many records lacked an event date. If I would have eradicated all undated occurrences, the dataset would have become too small for proper analysis. However, the dry forest GIS layer that I used is much more up to date and indicated quite well where PPS occurrences were likely to be still accurate.

Besides that, 16 out of the 37 PPS were left out of this study in the end since they do not occur in the Colombian TDF. This is quite an alarming finding. This would mean that these species might be even more endangered than the ones focussing on now, since they apparently have no proper habitat left. It could also be due to the fact that they were simply not sampled and/or georeferenced, which made them fall out of the dataset. For the three species that had no records at all in Colombia, the situation might even be more urgent.

Another implication is the data accessibility. First of all, I experienced that many databases and information sources in Colombia have limited access due to local server problems that could not be solved. For that reason I used less data than that was theoretically available. This would be something to improve in the future to make conservation studies in Colombia more accurate and complete. Also, plant occurrence data is very likely to be influenced by the relative inaccessibility of the areas occupied by the FARC-EP or other armed groups, during the years of conflict. I expect plant occurrences in former conflict territories therefore to be underrepresented or relatively old. However, this was not yet verified by scientific research. For the reliability of conservation studies or other studies based on spatial occurrence records in Colombia, it would be very useful to carry out such research.

5. Conclusions & Recommendations

5.1 Conclusions

The main question of this research was: Are current protected dry forest areas in Colombia preserving the most important populations of PPS for conservation?

Based on analysis method C, which is most appropriate for this dataset, I can conclude that current protected areas in Colombia are not preserving the most important populations of PPS for conservation. From the 46 IPAs that were identified, only 8 areas (17,4%) are protected, while 38 (82,6%) are unprotected. The other analysis methods point towards the same direction and thereby

(24)

24 support this conclusion. Three of the unprotected areas (IPAs in Cesar, Santander and Cundinamarca) were not only found through method C, but are shared between methods A-D and might therefore deserve most attention for now.

To come back to the second sub-question, in which was asked: If PPS are not sufficiently protected, in what type of habitat do these unprotected PPS occur? I can conclude that PPS in the three highlighted, unprotected IPAs mainly occur in areas that, apart from being characterized as forest, are intensively being used for cattle grazing and agricultural practices. Also urban areas are close by (Cesar, Santander). Moreover, the fact that secondary vegetation exists in and around the priority IPAs points out that these areas already have been impacted by deforestation before. Because of globalization, post-conflict development and a growing human population, agricultural practices are expected to expand in the future. Land use must therefore be considered as an important threat to TDF-PPS.

The literature research and GIS analysis for the third sub-question - that was: What might be threats to populations of PPS both in and outside protected areas? - pointed out that for PPS inside currently protected areas it is important to know how effective this protection is. This depends on the type of protection under which an IPA falls as well as potential illegal practices that take place in an area. On-site studies are needed to verify the effectiveness of protection.

Both protected and unprotected IPAs face the threat of the growing world population and globalization, which increases the pressure on ecosystems through food production and economic activities. Also the projected climate change forms a big threat for the TDF and PPS in general and s hard to tackle. In the three highlighted areas, projected vulnerability for climate change, likely being expressed in an even drier climate, is significant. Especially in Cundinamarca the risk is high to very high, which might result in loss of suitable TDF habitat for the PPS occurring here. Although other areas might become suitable for dry forest species instead, it is questionable if plant species are able to migrate and disperse fast enough. Extirpation (local extinction) or for some species complete extinction are more likely consequences.

For the three focus areas there is no direct threat of coal and gold mining. However, since these practices do occur in other parts of the tropical dry forest and are expected to expand in the future, they still form an important threat for TDF-PPS. Also there is no direct impact of coca production in the three focus areas. Though, since coca production is a big industry in Colombia and occurs illegally on a large scale, coca production might form a threat to TDF-PPS. The expected post-conflict development after a long period of civil war might form another threat for TDF-PPS, since socio-economic development will likely be in conflict with the conservation of nature.

5.2 Reflection & Recommendations

In the light of former studies that identified the conservation studies of the Tropical dry forest in Colombia, my conclusions are not very surprising, though not less important. It was already known that the tropical dry forest of Colombia is highly degraded and fragmented and that the protection status is very poor. Results of my study add to this that the few areas that are being protected are not sufficiently protecting the most important vulnerable and endangered plant species and that they face several threats that stress the need for protection even more.

As the result of this study shows, three unprotected IPAs (in Cesar, Santander and Cundinamarca) emerge from all methodological approaches being tested. I would suggest that immediate protection is initiated to preserve the PPS occurring here, while optimizing the method for further IPA selection.

To improve the method proposed by IAvH, more test studies are needed to come with a more specific approach. I would suggest to test the counting method with buffered occurrences records, that came as best method out of this study, on other datasets, including other biomes. I would also suggest

(25)

25 that very small analysis units are combined into larger ones, to improve the conservation ability of the TDF.

For conservation practices in general, is important to close the so called ‘knowing-doing gap’ (Knight et al., 2008). This study and many more studies on (plant) conservation in Colombia and in general, have been carried out and are available for implementation in actual biodiversity protection, but most studies never reach the actual implementation phase. If the rich biodiversity of Colombia is really to be protected, action is needed here.

Acknowledgements

First of all, I would like to thank my supervisor Dr. Suzette G.A. Flantua (UvA) for offering me the opportunity to do this project under her professional supervision. Also I am very thankful for her support and feedback where needed, which made me critically think about and reflect on my work. Her enthusiasm for the project and for her work in general has strongly motivated me to make the best out of it. Also it was an honour for me to present my thesis at the symposium that she organized around my bachelor projects and other studies in Colombia.

Secondly I would like to thank leader of the project, Dr. Mauricio Diazgranados that allowed me to contribute to the Colombia Bio programme that is currently being carried out by Kew Royal Botanic Gardens and the Humboldt Institute for biological research in Colombia (IAvH). It is an honour to me to contribute to this project.

Additionally I would like to thank Dr. Daniel Kissling for examining my thesis with his expert knowledge and supportive criticism.

References

Anderson, S. (2002). Identifying important plant areas. London: Plantlife International.

Andrade, G., & Rhodes, J. (2012). Protected areas and local communities: An inevitable partnership toward successful conservation strategies? Ecology and Society, 17(4).

Armenteras, D., Gast, F. & Villareal, H. (2003). Andean forest fragmentation and the representativeness of protected natural areas in the eastern Andes, Colombia. Biological conservation, 113(2), 245-256. Banda, K., Delgado-Salinas, A., Dexter, K. G., Linares-Palomino, R., Oliveira-Filho, A., Prado, D. et al. (DRYFLOR group) (2016). Plant diversity patterns in neotropical dry forests and their conservation implications. Science, 353(6306), 1383-1387.

Botanical Information and Ecology Network (BIEN) data base (2017). Accessed online via

http://biendata.org/ April 2017.

Brooks, T. M., Mittermeier, R. A., Mittermeier, C. G., Da Fonseca, G. A., Rylands, A. B., Konstant, W. R. et al. (2002). Habitat loss and extinction in the hotspots of biodiversity. Conservation biology, 16(4), 909-923.

Bruner, A. G., Gullison, R. E., Rice, R. E., & Da Fonseca, G. A. (2001). Effectiveness of parks in protecting tropical biodiversity. Science, 291(5501), 125-128.

Catalogador de Información Biológica - IAvH (CEIBA) (2013). Plantas vasculares que se han

reportado en los bosques secos de Colombia. Accessed online via

(26)

26 Dávalos, L. M., Bejarano, A. C., Hall, M. A., Correa, H. L., Corthals, A., & Espejo, O. J. (2011). Forests and drugs: coca-driven deforestation in tropical biodiversity hotspots. Environmental science & technology, 45(4), 1219-1227.

Del Valle E, Maldonado J, Sainz H, and Sánchez de Dios R. (2004). Áreas importantes para la flora amenazada española. Atlas y Libro Rojo de la Flora Vascular Amenazada de España Dirección General de Conservación de la Naturaleza Madrid, España:1069.

Fjeldså, J., Álvarez, M. D., Lazcano, J. M., & León, B. (2005). Illicit crops and armed conflict as constraints on biodiversity conservation in the Andes region. AMBIO: A Journal of the Human Environment, 34(3), 205-211.

GBIF Backbone Taxonomy (2016). GBIF Secretariat. Checklist Dataset http://doi.org/10.15468/39omei

accessed via GBIF.org on 2017-04-23.

Hannah, L., Midgley, G., Andelman, S., Araújo, M., Hughes, G., Martinez-Meyer, E. et al. (2007). Protected area needs in a changing climate. Frontiers in Ecology and the Environment 5:131–138. Huertas, J. I., Huertas, M. E., Izquierdo, S. & González, E. D. (2012). Air quality impact assessment of multiple open pit coal mines in northern Colombia. Journal of environmental management, 93(1), 121-129.

Instituto Alexander von Humboldt (IAvH) (2017). Preliminary methodology for the identification of TIPAS in Colombia (Draft report).

International Union for Conservation of Nature (IUCN) (2017) Protected areas categories. Accessed online on 4/5/2017 via: https://www.iucn.org/theme/protected-areas/about/protected-areas-categories

Jones, K.R., Watson, J.E.M., Possingham, H.P. & Klein, C.J. (2016). Incorporating climate change into spatial conservation prioritisation: A review. Biological Conservation 194:121–130. 10.1016/j.biocon.2015.12.008

Knight, A. T., Cowling, R. M., Rouget, M., Balmford, A., Lombard, A. T., & Campbell, B. M. (2008). Knowing but not doing: selecting priority conservation areas and the research–implementation

gap. Conservation biology, 22(3), 610-617.

Miles, L., Newton, A. C., DeFries, R. S., Ravilious, C., May, I., Blyth, S., et al. (2006). A global overview of the conservation status of tropical dry forests. Journal of Biogeography, 33(3), 491-505. Mittermeier, R. A., Gil, P. R., Mittermeier, C. G. (1997). Megadiversity. Earth's biologically

wealthiest nations. Conservation International, Washington, D.C.

Meyers, N., Mittermeier, R. A., Mittermeier, C. G., Da Fonseca, G. A., & Kent, J. (2000). Biodiversity hotspots for conservation priorities. Nature, 403(6772), 853-858.

Moor, M. & van de Sandt, J. (2014). The dark side of coal. Paramilitary Violence in the Mining Region of Cesar, Colombia. Published by PAX for Peace, Utrecht, The Netherlands.

Morales, L. (2017) Peace and Environmental Protection in Colombia. Proposals for Sustainable Rural Development; http://go.nature.com/2l1gIRz

Murcia, C., Kattan, G. H., & Andrade-Pérez, G. I. (2013). Conserving biodiversity in a complex biological and social setting: The case of Colombia. Conservation biology: voices from the tropics. John Wiley & Sons, Ltd., Hoboken, New Jersey, 86-96.

(27)

27 Negret, P. J., Allan, J., Braczkowski, A., Maron, M., & Watson, J. E. (2017). Need for conservation planning in postconflict Colombia. Conservation Biology.

Parques Nacionales Naturales de Colombia (2017). Reservas Naturales de la Sociedad Civil. Accessed online on 5/8/2017 via http://www.parquesnacionales.gov.co/portal/es/sistema-nacional-de-areas-protegidas-sinap/reservas-naturales-de-la-sociedad-civil/

Pizano, C. & García, H. (Editors) (2014). El Bosque Seco Tropical en Colombia. Instituto de Investigación de Recursos Biológicos Alexander von Humboldt (IAvH). Bogotá, D.C., Colombia. Portillo-Quintero, C. A. & Sánchez-Azofeifa, G. A. (2010). Extent and conservation of tropical dry forests in the Americas. Biological Conservation, 143(1), 144-155.

Rincón-Ruiz, A., & Kallis, G. (2013). Caught in the middle, Colombia’s war on drugs and its effects on forest and people. Geoforum, 46, 60-78.

Sandvik, B. (2009). TM_world_borders-0.3. Thematic Mapping. Downloaded via

http://thematicmapping.org/downloads/world_borders.php (April 2017).

Särkinen, T., Iganci, J. R., Linares-Palomino, R., Simon, M. F., & Prado, D. E. (2011). Forgotten forests-issues and prospects in biome mapping using seasonally dry tropical forests as a case study. BMC ecology, 11(1), 27.

Schulman, L., Toivonen, T., & Ruokolainen, K. (2007). Analysing botanical collecting effort in Amazonia and correcting for it in species range estimation. Journal of Biogeography, 34(8), 1388-1399. Sistema de Información Ambiental de Colombia (SIAC). Catalogo de Mapas. Accessed online via

http://www.siac.gov.co/Catalogo_mapas.html (April 2017)

Sistema de Información Geográfica para la Planeación y el Ordenamiento Territorial (SIG-OT). Mapas temáticos predifinidos – nacional. Accessed online (June 2017) via

http://sigotn.igac.gov.co/sigotn/frames_pagina.aspx

United Nations Office on Drugs and Crime (UNODC) (2015). Colombia. Coca cultivation survey 2014. Provided by the Governement of Colombia. Accessed online on 8/6/2017 via

https://www.unodc.org/documents/crop-monitoring/Colombia/censo_INGLES_2014_WEB.pdf

Wilson, E.O (editor), Janzen, D.H.(author) (1988). Biodiversity. Ch 14: Tropical Dry Forests: The Most Endangered Major Tropical Ecosystem. National Academy Press, Washington D.C.

Front figure: http://www.humboldt.org.co/es/investigacion/proyectos/en-desarrollo/item/158-bosques-secos-tropicales-en-colombia

(28)

28

Appendix 1. Used GIS layers

Table 1.1. GIS layers used for IPA assessment

Name Description Source

BosqueSecoTropical_100K Shows the most up to date extend of TDF in Colombia (2014).

IAvH; SIAC

Parques Natural parks of Colombia

declared at a national scale

Sistema de Parques Nacionales de Colombia (SPNN); SIAC Parques natural regional Natural parks of Colombia

declared at a regional scale

Sistema de Parques Nacionales de Colombia (SPNN); SIAC Reserva sociedad civil Civil association/society

reserve

Sistema de Parques Nacionales de Colombia (SPNN); SIAC

Reserva de la biosfera “Biosphere reserve” Ministerio de Ambiente y

Desarrollo Sostenible (MADS); SIAC

Protected Areas Protected areas of Colombia. World Database of Protected

Areas (WDPA); SIAC

TM_world_borders-0.3 World borders Sandvik, 2009

Table 1.2. GIS layers used for analysis of threats

Name Description Source

BosqueSecoTropical_100K Shows the most up to date extend of TDF in Colombia (2014).

IAvH; SIAC

Cobertura de la Tierra Landcover 2008 Instituto Geográfico Augustín

Codazzi (IGAC); SIG-OT Producción de Carbón Production of carbon (tonnes)

in 2016 per municipality

SIG-OT Produccíon Minera de Oro Gold mining (kg) per

municipality

SIG-OT Protected areas merged All layers showing protected

areas (SIAC, table 1.1) merged into one layer.

SIAC, merged by myself

Vulnerabilidad Ambiental del Territorio Colombiano en el periodo 2011-2040

Vulnerability for projected climate change from 2011 – 2040.

SIAC

(29)

29

Appendix 2. IPA thresholds

FI D sp ec ie s Ca te go ry O cc g lo ba l_ DOc c n ati on al D 1% th re sh ol d_ D 10 % th re sh ol d_ D #o cc g lo ba l N D # na tio na l N D1% th re sh ol ds _N D 10 % th re sh ol d_ N D Ar ea G lo ba l ( km2 ) 1 ,5 k m bu ff er 1% th re sh ol d (A 1) Ar ea n ati on al 1 ,5 k m bu ff er 10 % th re sh ol d (A 3) 0 Ac oe lo rr ap he w rig hti i A3 110 0 1 NA 100 0 1 NA 647 6, 47 NA NA 1 An ib a pe ru til is A3 194 121 2 12 116 81 1 8 553 5, 53 391 39 ,1 2 Ap he la nd ra fl av a A1 17 12 1 1 16 11 1 1 106 1, 06 78 7, 8 3 Ap he la nd ra p ha ra ng op hi la A1 19 17 1 2 9 7 1 1 43 0, 43 35 3, 5 4 As pi do sp er ma p ol yn eu ro n A3 650 460 7 46 197 53 2 5 1275 12 ,7 5 352 35 ,2 5 Att al ea a my gd al in a A3 51 51 1 5 10 10 1 1 71 0, 71 71 7, 1 6 Ba ctr is g as ip ae s A3 810 230 8 23 236 72 2 7 1499 14 ,9 9 483 48 ,3 7 Be le nc ita n emo ro sa A1 46 35 1 4 37 26 1 3 228 2, 28 170 17 8 Bu ln es ia a rb or ea A3 138 80 1 8 63 35 1 4 442 4, 42 316 31 ,6 9 Ca lip hr ur ia su be de nta ta A1 7 7 1 1 5 5 1 1 35 0, 35 35 3, 5 10 Ca pp ar id as tr um ma cr op hy llu m A1 98 20 1 2 62 18 1 2 380 3, 8 117 11 ,7 11 Ca pp ar is fl ex uo sa A1 72 1 1 1 57 1 1 1 401 4, 01 7 0, 7 12 Ca rin ia na p yr ifo rmi s A3 129 113 1 11 79 69 1 7 433 4, 33 381 38 ,1 13 Ca tt le ya q ua dr ic ol or A3 2 0 1 NA 2 0 1 NA 15 0, 15 NA NA 14 Ca va ni lle si a pl ata ni fo lia A1 144 49 1 5 81 31 1 3 447 4, 47 179 17 ,9 15 Ce dr el a od or ata A3 2690 231 27 23 1828 159 18 16 11288 11 2, 88 1015 10 1, 5 16 Cy no ph al la fl ex uo sa A3 2212 150 22 15 1291 49 13 5 8895 88 ,9 5 358 35 ,8 17 El ae is o le ife ra A3 865 110 9 11 172 29 2 3 1116 11 ,1 6 170 17 18 Eu ch ar is ca uc an a A1 2 2 1 1 2 2 1 1 14 0, 14 14 1, 4 19 Ga ya mu tis ia na A1 6 6 1 1 5 5 1 1 28 0, 28 28 2, 8 20 Gu ai ac um of fic in al e A3 241 27 2 3 137 12 1 1 905 9, 05 114 11 ,4 21 Li ca ni a ar bo re a A3 728 33 7 3 488 15 5 2 2511 25 ,1 1 111 11 ,1 22 Li ca ni a pl aty pu s A3 493 1 5 1 294 1 3 1 1672 16 ,7 2 7 0, 7 23 M ay te nu s c or ei A1 3 3 1 1 2 2 1 1 21 0, 21 21 2, 1 24 M el oc ac tu s c ur vi sp in us A1 117 27 1 3 85 15 1 2 465 4, 65 77 7, 7 25 O xa nd ra e sp in ta na A1 553 341 6 34 110 10 1 1 494 4, 94 54 5, 4 26 Pa ch ira q ui na ta A3 443 61 4 6 215 27 2 3 1297 12 ,9 7 220 22 27 Pa ch ira su ba nd in a A1 2 0 1 NA 2 0 1 NA 7 0, 07 NA NA 28 Pa rin ar i p ac hy ph yl la A3 56 26 1 3 32 15 1 2 269 2, 69 138 13 ,8 29 Pa ss ifl or a ma gd al en ae A3 47 47 1 5 20 20 1 2 73 0, 73 73 7, 3 30 Pe ltogy ne p ur pu re a A3 223 82 2 8 56 4 1 1 266 2, 66 87 8, 7 31 Pi tc ai rn ia ste no ph yl la A1 11 8 1 1 5 3 1 1 30 0, 3 14 1, 4 32 Pl ag io lir io n ho rs ma nn ii A1 1 1 1 1 1 1 1 1 7 0, 07 7 0, 7 33 Sp ath ip hy llu m gr an di fo liu m A1 53 19 1 2 29 9 1 1 185 1, 85 63 6, 3 34 Ste no ce re us h umi lis A3 40 40 1 4 11 11 1 1 112 1, 12 112 11 ,2 35 Sy ag ru s s an co na A3 235 112 2 11 60 16 1 2 411 4, 11 134 13 ,4 To ta l 11508 2523 5915 824 36641 5432

(30)

30

Appendix 3. PPS list

Table 3. List of PPS species including threat category and precense in the TDF of Colombia

Scientific name Category Present in TDF Colombia

Aphelandra flava A1 Yes

Aphelandra pharangophila A1 Yes

Aspidosperma polyneuron A3 Yes

Attalea amygdalina A3 Yes

Bactris gasipaes A3 Yes

Belencita nemorosa A1 Yes

Bulnesia arborea A3 Yes

Capparidastrum macrophyllum A1 Yes

Cavanillesia platanifolia A1 Yes

Cedrela odorata A3 Yes

Cynophalla flexuosa A3 Yes

Elaeis oleifera A3 Yes

Guaiacum officinale A3 Yes

Maytenus corei A1 Yes

Melocactus curvispinus A1 Yes

Oxandra espintana A1 Yes

Pachira quinata A3 Yes

Parinari pachyphylla A3 Yes

Peltogyne purpurea A3 Yes

Stenocereus humilis A3 Yes

Syagrus sancona A3 Yes

Acoelorraphe wrightii A3 No Aniba perutilis A3 No Caliphruria subedentata A1 No Capparis flexuosa A1 No Cariniana pyriformis A3 No Cattleya quadricolor A3 No Eucharis caucana A1 No Gaya mutisiana A1 No Libidibia ebano A3 No Licania arborea A3 No Licania platypus A3 No Pachira subandina A1 No Passiflora magdalenae A3 No Pitcairnia stenophylla A1 No Plagiolirion horsmannii A1 No Spathiphyllum grandifolium A1 No

(31)

31

Appendix 4. IPA results

A.

Table 4.1. IPAs for method A including threat category, threshold value, number of occurrences in TDF and protection status

B.

Table 4.2. IPAs for method B including threat category, threshold value, number of occurrences in TDF and protection status

TDF AU Species Category Threshold Occurrences Protected?

375 Peltogyne purpurea A3 1 1 Yes 546 Cavanillesia platanifolia A1 1 3 Yes 745 Belencita nemorosa A1 1 1 Yes 4 Oxandra espintana A1 1 1 No 22 Aphelandra flava A1 1 1 No Capparidastrum macrophyllum A1 1 2 118 Parinari pachyphylla A3 2 2 No 670 Cavanillesia platanifolia A1 1 2 No 1363 Melocactus curvispinus A1 1 5 No TDF

AU Species Category Threshold Occurrences Protected?

546 Aspidosperma polyneuron A3 46 170 Yes Bulnesia arborea A3 8 15 Cavanillesia platanifolia A1 1 3 Peltogyne purpurea A3 8 51

745 Belencita nemorosa A1 1 1 Yes

889 Cynophalla flexuosa A3 15 43 Yes

22 Aphelandra flava A1 1 1 No

Capparidastrum macrophyllum A1 1 2

670 Cavanillesia platanifolia A1 1 3 No

1363 Melocactus curvispinus A1 1 6 No

(32)

32

C.

Table 4.3. IPAs for method C including threat category, threshold value, number of occurrences in TDF and protection status

TDF AU Species Category Threshold Occurrences Protected?

546 Cavanillesia platanifolia A1 1 3 Yes

851 Guaiacum officinale A3 1 1 Yes

882 Guaiacum officinale A3 1 1 Yes

889 Aspidosperma polyneuron A3 5 6 Yes

1111 Melocactus curvispinus A1 1 1 Yes

4 Oxandra espintana A1 1 1 No 6 Aphelandra flava A1 1 1 No 21 Aphelandra flava A1 1 1 No 22 Aphelandra flava A1 1 1 No Capparidastrum macrophyllum A1 1 3 118 Parinari pachyphylla A3 2 2 No 121 Parinari pachyphylla A3 2 2 No 126 Parinari pachyphylla A3 2 2 No 353 Cavanillesia platanifolia A1 1 1 No 357 Cavanillesia platanifolia A1 1 1 No 361 Cavanillesia platanifolia A1 1 1 No 375 Cavanillesia platanifolia A1 1 4 No Peltogyne purpurea A3 1 1 451 Belencita nemorosa A1 1 1 No 453 Belencita nemorosa A1 1 1 No 459 Belencita nemorosa A1 1 1 No 553 Guaiacum officinale A3 1 1 No 556 Cavanillesia platanifolia A1 1 2 No 556 Guaiacum officinale A3 1 1 No 576 Cavanillesia platanifolia A1 1 2 No 670 Cavanillesia platanifolia A1 1 4 No 693 Cavanillesia platanifolia A1 1 2 No 844 Belencita nemorosa A1 1 1 No 1080 Aphelandra pharangophila A1 1 1 No Capparidastrum macrophyllum A1 1 1 Stenocereus humilis A3 1 1 1094 Capparidastrum macrophyllum A1 1 2 No Stenocereus humilis A3 1 1 1107 Melocactus curvispinus A1 1 1 No

(33)

33 1108 Melocactus curvispinus A1 1 1 No 1110 Stenocereus humilis A3 1 1 No 1152 Maytenus corei A1 1 1 No 1153 Aphelandra flava A1 1 1 No 1363 Melocactus curvispinus A1 1 6 No 1393 Belencita nemorosa A1 1 1 No Oxandra espintana A1 1 1 1537 Attalea amygdalina A3 1 2 No 1613 Aphelandra pharangophila A1 1 2 No 1614 Aphelandra pharangophila A1 1 2 No 1616 Aphelandra pharangophila A1 1 2 No 1623 Aphelandra pharangophila A1 1 2 No 1625 Aphelandra pharangophila A1 1 2 No 1628 Aphelandra pharangophila A1 1 2 No 1655 Stenocereus humilis A3 1 1 No

D.

Table 4.4. IPAs for method D including threat category, threshold value, number of occurrences in TDF and protection status

TDF

AU Species Category Threshold Area Protected?

546 Cavanillesia platanifolia A1 4.47 1.809.827 Yes

745 Belencita nemorosa A1 2.28 3.507.927 Yes

21 Aphelandra flava A1 1.06 1.353.143 No 22 Aphelandra flava A1 1.06 156.987 No Capparidastrum macrophyllum A1 3.8 8.461.954 375 Cavanillesia platanifolia A1 4.47 7.255.728 No 670 Cavanillesia platanifolia A1 4.47 1.499.127 No 1080 Aphelandra pharangophila A1 0.43 0.517917 No 1152 Maytenus corei A1 0.21 1.193.042 No 1363 Melocactus curvispinus A1 4.65 216.979 No 1616 Aphelandra pharangophila A1 0.43 0.762582 No 1623 Aphelandra pharangophila A1 0.43 0.577183 No 1625 Aphelandra pharangophila A1 0.43 0.951147 No

(34)

34

Appendix 5. R scripts

A.

PPS_within <- 3 PPS_outside <- 4 alt.binom <- (binom.test(PPS_within,(PPS_outside+PPS_within))) p_value.binom = as.numeric(alt.binom[3]) p_value.binom 1

B.

PPS_within <- 3 PPS_outside <- 5 alt.binom <- (binom.test(PPS_within,(PPS_outside+PPS_within))) p_value.binom = as.numeric(alt.binom[3]) p_value.binom 0.7265625

C.

PPS_within <- 8 PPS_outside <- 38 PPS_total <- sum(PPS_within,PPS_outside) chi_matrix=matrix(NA,nrow=3,ncol=4) chi_matrix[1:3,1] = c("obs","exp","chi") chi_table <- data.frame(chi_matrix,row.names=1,stringsAsFactors=FALSE) colnames(chi_table) = c("within","outside","total") chi_table[1,1] = PPS_within chi_table[1,2] = PPS_outside chi_table[1:2,3] = PPS_total chi_table[2,1:2] = PPS_total / 2

for(n in 1:2){ chi_table[3,n] = ((as.numeric(chi_table[1,n]) - as.numeric(chi_table[2,n])) ^ 2) / as.numeric(chi_table[2,n])} chi_table[3,3] = sum(as.numeric(chi_table[3,1:2]))

p_value = pchisq(as.numeric(chi_table[3,3]),df=1,lower.tail=FALSE)

(35)

35

Within Outside Total

Obs 8 38 46 Exp 23 23 46 Chi 9.78260869565217 9.78260869565217 19.5652173913043 p_value 9.722321e-06

D.

PPS_within <- 2 PPS_outside <- 10 PPS_total <- sum(PPS_within,PPS_outside) chi_matrix=matrix(NA,nrow=3,ncol=4) chi_matrix[1:3,1] = c("obs","exp","chi") chi_table <- data.frame(chi_matrix,row.names=1,stringsAsFactors=FALSE) colnames(chi_table) = c("within","outside","total") chi_table[1,1] = PPS_within chi_table[1,2] = PPS_outside chi_table[1:2,3] = PPS_total chi_table[2,1:2] = PPS_total / 2

for(n in 1:2){chi_table[3,n] = ((as.numeric(chi_table[1,n]) - as.numeric(chi_table[2,n])) ^ 2) / as.numeric(chi_table[2,n])} chi_table[3,3] = sum(as.numeric(chi_table[3,1:2]))

p_value = pchisq(as.numeric(chi_table[3,3]),df=1,lower.tail=FALSE)

chi_table

Within Outside Total

Obs 2 10 12

Exp 6 6 12

Chi 2.66666666666667 2.66666666666667 5.33333333333334

p_value 0.02092134

(36)

36

Appendix 6. Team of collaborators

Collaborator Current position Affiliation Responsibilities/contributions

Reineke S. van Tol Bachelor student Future Planet Studies / Biology University of Amsterdam Planning Research proposal Data analysis Writing report Presentation Dr. Suzette G.A. Flantua Paleoecology & Landscape ecology group; Data analyst Computational Geo-Ecology group IBED University of Amsterdam Daily supervision Dr. W. Daniel Kissling Associate professor IBED University of Amsterdam Examiner Dr. Mauricio Diazgranados

Kew Royal Botanic Gardens, Richmond

Project leader Data provider

Identifying the protection status of vulnerable and endangered plant species in the Tropical Dry Forest of Colombia