*1.2. Study Area*

La Paz Bay is in the southwest of the Gulf of California, at the southern portion of the Baja California Peninsula, in Mexico (Figure 1). This bay has diverse biotic and abiotic characteristics that made it a remarkable site with diverse ecosystems, such as beaches, dunes, seagrass beds, rocky reefs, and mangroves [17]. Its coastal zone harbors 17 mangrove communities, 16 of them within the proximity of La Paz city, and one more at the Espíritu Santo Archipelago. These mangroves are within protected areas such as the Espíritu Santo Archipelago National Park, the Balandra Flora and Fauna Protection Area, and the Islands of the Gulf of California Flora and Fauna Protection Area. They are also under the category of international protection, such as the Ramsar Sites "Humedales Mogote-Ensenada de La Paz" (Ramsar site no. 1816) and "Balandra" (Ramsar site no. 1767).

**Figure 1.** Location of La Paz Bay in the southern end of the Peninsula of Baja California, Mexico.

The 14 communities belonging to the Ramsar site No. 1816 are: Centenario-Chametla, Comitán, El Conchalito, El Mogote, Enfermería, Eréndira, Estero Bahía Falsa, Estero El Gato, La Paz-Aeropuerto, Palmira, Playa Pichilingue-Brujas, Unidad Pichilingue UABCS, Salinas de Pichilingue, and Zacatecas. Two of these remaining communities belonging to Ramsar site no. 1767 are Balandra and El Merito. Finally, the mangrove Espíritu Santo Archipelago is located within another protected area, it is in the national park category and its name is homonymous to that community (Figure 2).

**Figure 2.** Location of mangroves placed in La Paz Bay, Mexico.

#### **2. Materials and Methods**

## *2.1. Study Design*

The development and implementation of the Mangrove Conservation Status Index (MCSI) consisted of five phases carried out between January and September 2019. First, the remnant vegetation index (RVI), Delphi method survey (DMS), and the rapid assessment questionnaire (RAQ) were conducted at the 17 mangrove areas during January–February 2019, followed by the use of the analytical hierarchy process (AHP) to estimate the weights of each one of these indicators. We conducted a spatial analysis on open-source GIS software QGIS (version 3.4.4) to calculate the RVI, designed a questionnaire for the RAQ, and applied surveys to local mangrove experts for the DMS. Finally, we estimated the Mangrove Conservation Status Index for all the mangroves.

#### *2.2. Index Development and AHP*

For the construction of the MCSI, three components were selected, as these are widely used in various environmental analyses (RVI, DMS, and RAQ). For example, several studies have carried out comparisons between mangrove cover between different years, globally and nationally and locally. Also, the application of the Delphi method in environmental modeling is considered as a useful tool, with various studies focused on mangroves. Finally, the rapid assessment tool has also been used in forest and wetland analysis. Once the index components were selected, the following formula was generated:

$$\mathbf{MCSI} = (\mathbf{RVI})\mathbf{W1} + (\mathbf{DMS})\mathbf{W2} + (\mathbf{RAQ})\mathbf{W3}.$$

The weights of each component were determined using the analytical hierarchy process (AHP) method developed by Thomas L. Saaty (see Appendix A). For this, paired combinations were made between the three components using a pairwise comparison matrix (Table 1). For example, since the RVI is a quantitative value that reflects the loss or gain of cover in a given period, it was considered of greater relevance than the RAQ and DMS components. In the same way, among these last components, the DMS was considered of greater importance than RAQ. DMS is the result of the opinion of several experts (which includes years of experience) in comparison to RAQ, which takes information from a single field visit.


**Table 1.** Pairwise comparison matrix (PCM).

#### *2.3. Remaining Vegetation Index (RVI)*

The value of RVI of each mangrove community was calculated by considering the vegetation cover obtained in 2018 as the present vegetation area (PVA), divided by the original vegetation area (OVA), which corresponded to the data of the year of 1973. The result was multiplied by 100 to obtain a comparable value on a scale of 0/100. This index was used for the first time in a case study in Colombia [18] following this formula:

#### **RVI** = [(PVA)/(OVA)] × 100.

We obtained the vegetation area from scanned aerial photographs and Landsat satellite images. We consulted CONABIO's database. We used the oldest image available from the sources mentioned above for the calculation of the RVI for each mangrove community. In this case, we obtained an aerial photograph from 1974 that captured mangroves, except those at Espíritu Santo Archipelago, in La Paz bay at the Autonomous University of Baja California Sur library's archive. We downloaded Sentinel images (10 m, 20 m, and 60 m, avoiding cloud interference) for May 2018 from the Earth Explorer platform (USGS) to calculate present vegetation cover for the calculation of the RVI.

To digitalize the aerial photographs, we scanned them with the highest available resolution (10,200 × 14,028 pixels). We georeferenced images using geomorphological land references by the control point method. Subsequently, we extracted the sections corresponding to mangroves on QGIS and obtained pixels (1 m × 1 m) by a resampling process. The resampling process did not allow for a higher pixel resolution; nevertheless, it provided better contrast between neighboring pixels. Therefore, observations allowed the precise definition of mangrove areas (Figure 3).

Together, the field data collection and the georeferenced aerial photographs allowed the confirmation of the presence of mangroves and the obtention of the polygons containing mangroves by the use of the on-screen scanning facility in the QGIS software. We transformed the satellite image from geographic coordinates to metrics. For optimal use, we created a composite of bands 4, 5, 3, and panchromatic (Band 8) to increase image resolution and facilitate vegetation recognition [19]. The generation of the base project in the QGIS platform integrated the resulting images in raster format. Once the properties of the images (pixel size, georeference) were validated, we calculated the mangrove cover for each of the 17 sites. We obtained 16 mangrove polygons for 1974 and 17 for 2018 using the manual digitizing technique which has been used by different authors [20–23] and estimated the vegetation cover area. We used the resulting areas to calculate the RVI.

**Figure 3.** Treatment of spatial images in the QGIS software.

#### *2.4. Delphi Method Survey*

We applied interviews with regional mangrove experts following the Delphi method, which is a structured way to obtain information and knowledge on a particular topic [24]. This method provides both qualitative and quantitative data (see Appendix B), and it can be adapted for rapid assessments, such as the one implemented during this study [25]. We contacted a total of ten people, but only seven answered the survey. Of these, four were researchers, two worked for governmen<sup>t</sup> agencies, and one collaborated with non-governmen<sup>t</sup> organizations. We conducted interviews in person or remotely via electronic media such as Skype or video conference. The interview consisted mostly of open questions, as well as closed questions or fixed-alternatives. Interviewees answered the open question freely with no limit on time. The fixed-alternative questions were formulated to be answered in a scalar way using the Likert measurement tool (Table 2), which consists of obtaining a degree of conformity determined by a range of values. Table 3 shows the main question used in the survey.



Note: Bold scores represent increases on mangrove cover.


**Table 3.** Key question applied to experts in the Delphi Method Survey component.

#### *2.5. Rapid Assessment Questionnaire*

The rapid evaluation is a reliable and timely estimation method, which allows an approximation of the magnitude and characteristics of a problem. It marks the line to define needs or tasks to consider during a subsequent evaluation [26]. This type of assessment provides complementary information to other sources, in a simple, fast, and flexible way. In the case of the mangroves of La Paz Bay, we visited 17 sites, which were selected according to the managemen<sup>t</sup> plans of the protected area (Balandra) and Ramsar site (Humedales Mogote-Ensenada de La Paz No. 1816). To evaluate each of the mentioned mangroves, we created a rapid assessment questionnaire (*RAQ*) based on di fferent surveys developed by academics and decision-makers from the region. The *RAQ* considered specific environmental indicators, divided thematically (water, air, soil, flora, fauna, and waste), and used qualitative indicators to assess impacts observed at each mangrove site during the field visits. The values of the *RAQ* ran from 0 to 1; the closer the value to 1, the more impacted the site was. We recorded our observations at the site, and photographic evidence is available from the authors upon request (see Appendix C).

#### *2.6. Application of the Integrative Mangrove Conservation Status Index*

We calculated the *MCSI* using the scores of each one of the components of the index, RVI, RAQ, and DMS, and following the formula:

$$\mathbf{MCSI} = (\mathbf{RVI})(0.62) + (\mathbf{RAQ})(0.25) + (\mathbf{DMS})(0.13).$$

We classified mangrove sites depending on their MCSI score following an adapted classification of the IUCN Red List of Ecosystems (see Appendix D). This scale considers eight categories of risk for the earth's ecosystem (Figure 4). Three of them contemplate quantitative thresholds: critically endangered (CR), endangered (EN), and vulnerable (VU)—together, the IUCN describes these ecosystems as threatened. There are several qualitative categories to include: (1) ecosystems that fail to meet the quantitative criteria for the threatened ecosystem categories (NT, near threatened); (2) ecosystems that unambiguously meet none of the quantitative criteria (LC, least concern); (3) ecosystems with poor data (DD, data deficient); and (4) ecosystems that have not been assessed (NE, not evaluated). An additional category (CO, collapse) is assigned to ecosystems that have collapsed throughout their distribution, the analogue of the extinct (EX) category for species [27].

**Figure 4.** Categories of the IUCN Red List of Ecosystems. Source: IUCN, 2019.
