INTRODUCTION

Mangrove forests are composed of trees and shrubs that thrive in intertidal zones between tropical and subtropical latitudes. These forests provide ecosystem services such as protecting against coastal erosion and providing nursery grounds for commercial species (^{Ximenes et al. 2023}). Field-based mangrove forest variables, such as basal area and stem density, are of utmost importance for estimating structural complexity and aboveground biomass (^{Tran et al. 2022}) and corroborating data from remote sensors (^{Vizcaya-Martínez et al. 2022}).

The differences among mangrove field-based inventory variables depend on the species, geographic region, and forest type (^{Wang et al. 2018}). For example, mangrove forests in tropical zones with higher annual rainfall rates tend to present higher basal areas and less stem density than mangroves in arid or semiarid regions (^{Valderrama-Landeros et al. 2022}). The logistics of fieldwork in mangrove forests are much more complicated and time-consuming than in other terrestrial forests (^{Salum et al. 2020}). For instance, the coastal environments where mangroves thrive are arduous for fieldwork, mainly due to tidal fluctuations and muddy soil (^{Ferreira et al. 2022}). In addition, the soil of degraded mangrove forests tends to be less compact than that of mangroves living under optimal conditions, increasing the difficulty of fieldwork (^{Flores-Verdugo et al. 2015}). Thus, it is necessary to identify the ideal method to efficiently carry out mangrove field-based surveys for each species within a mangrove forest.

There are several ways to perform field-based inventories, including distance-based techniques (plotless approaches), such as the point-centered quarter method (PCQM), and area-based approaches (square and circular plots). Overall, fixed square plots of up to 0.16 ha (40 × 40 m) in which all stems within the area are counted are generally preferred for long-term monitoring purposes (^{Villeda-Chávez et al. 2018}, ^{Valderrama-Landeros et al. 2020}). However, since these field-based observations are restricted to only a few permanently installed plots, the acquired data are spatially limited (^{Flores-de-Santiago et al. 2023}). For this reason, when spatially analyzing a mangrove system to corroborate biomass estimations from remote sensors, including synthetic aperture radar and unmanned aerial vehicles, or existing metrics, such as leaf area indices from the literature, it is better to cover multiple small stations such as circular plots (^{Flores-de-Santiago et al. 2013}, ²⁰²⁰). The PCQM approach utilizes transects that traverse forests, thereby eliminating the need to designate specific sampling areas. However, the effectiveness of this method depends on the number of stations with transects. Moreover, the size of the forest may not be sufficiently large for PCQM.

The objective of this study was to determine the degree of error among traditional field-based survey methods for mangrove forests in a semiarid system of the Gulf of California using a field-based approach. The working hypothesis was that the optimal method would depend on the stem density of the mangrove species. Due to the gaps within the plots, we expected an inverse estimation error relationship between stem density and basal area.

MATERIALS AND METHODS

Study area

The Urias coastal lagoon (Fig. 1) is a small (18 km²), shallow (<4 m depth) water body located on the semiarid eastern coast of the Gulf of California, Mexico (23°09′06″ N, 106°19′57″ W). Three mangrove species partially border the coastal lagoon: Rhizophora mangle L. (RM), Laguncularia racemosa (L.) Gaertn (LR), and Avicennia germinans (L.) L. (AG) (^{Valderrama-Landeros et al. 2018}, ²⁰²¹). The short rainy season and tendency of the coastal lagoon to exhibit hypersaline conditions have resulted in extensive shrub-like AG mangrove communities in areas with low water exchange (^{Flores-Verdugo et al. 2015}).

Figure 1 Locations of the field-based mangrove inventories in the southwestern Gulf of California in the Mexican Pacific. The bold line indicates the USA-Mexico international border, and the red circle depicts the location of the Urias system. Each colored rectangle indicates the survey area of 0.16 ha of each mangrove class over a false-color composite (near infrarred, red, green) GeoEye image. 

RESULTS

All mangrove classes presented some multi-stemmed trees (Fig. 2). For the square plot method, the spatial distribution of the stems varied according to the species; for example, the RM forest presented low density (444 stems/ha), scattered trees, and low basal area (4.8 m²/ha) (Table 1). The LR forest exhibited a higher density (3,225 stems/ha) of multi-stem trees with a relatively high basal area (13.4 m²/ha). In contrast, the AG forest contained scattered trees (1,069 stems/ha) with a small basal area (5.7 m²/ha). Finally, AG shrub presented high density (6,500 stems/ha) and a basal area of 14.6 m²/ha distributed in dense clusters and spaces among the analyzed stems. Due to the high density of smaller-sized stems, LR and AG shrub exhibited different densities when considering all data but with a basal area that remained relatively unchanged. For example, the LR forest increased from 3,225 to 4,500 stems/ha, and its basal area increased from 13.4 to 13.7 m²/ha. On the other hand, the AG density in AG shrub increased from 6,269 to 6,500 stems/ha, while its basal area only increased from 14.5 to 14.6 m²/ha.

Figure 2 The spatial distribution of mangrove stems according to the diameter at breast height (DBH; small circles) measured for the 4 mangrove classes (Rhizophora mangle, Laguncularia racemosa, Avicennia germinans, and Avicennia germins shrub). Each square measured 40 × 40 m. The 3 blue circles indicate the locations of the circular plots (0.04, 0.03, and 0.015 ha). The purple lines indicate the locations of the 3 point-centered quarter method (PCQM) transects. 

The time required for 4 people to complete a field-based mangrove inventory within a square plot (0.16 ha) was 1 h and 20 min for RM, 1 h for LR, 20 min for AG, and 2 h for AG shrub. The DBH distribution of the 4 mangrove classes revealed noticeable differences. For example, most RM stems exhibited DBH frequencies between 8 and 10 cm, followed by a second group composed of frequencies between 12 and 14 cm. The LR forest exhibited high density for stems less than 4 cm. The AG forest exhibited a normal distribution with frequencies between 4 and 6 cm, while AG shrub presented much higher densities between 4 and 6 cm.

The RM class showed no significant differences in basal area (F _(4,116) = 2.6, P = 0.04) and stem density (F _(4,116) = 4.1, P = 0.004) among the 4 sampling methods when compared with those of the square method. The 0.04-ha circle and PCQM approaches resulted in the smallest estimated error. The PCQM approach also resulted in the lowest CV for basal area and stem density; however, PCQM required more time to complete than the circular area plot method. The LR class (DBH > 2.5 cm) showed significant differences among the methods in terms of basal area (F _(4,799) = 184.2, P = 0.0), but the post-hoc analysis revealed no significant differences between the 0.04-ha and 0.03-ha circles (P = 0.43). Nonetheless, all methods showed significant differences (F _(4,799) = 145.9, P = 0.0) in stem density for the LR class. Moreover, PCQM showed the highest basal area and stem density errors, and we recorded the lowest estimated errors with the 0.015-ha circle.

The AG class showed significant differences in both basal area (F _(4,285) = 25.1, P = 0.0) and stem density (F _(4,285) = 21.3, P = 0.0), but the post-hoc analysis indicated that there were no significant differences between the square plot and PCQM (P = 1) and the circular plot of 0.04 ha (P = 0.98) for both variables. All methods showed significant differences in basal area (F _(4,1543) = 41.3, P = 0.0) and stem density (F _(4,1543) = 244.2, P = 0.0) in the AG shrub class. However, we estimated a minimum error for both variables within the circular plot of 0.015 ha. The statistical results for LR (all stems) and AG shrub (all stems) did not vary compared to the results of the previous analyses with stems larger than 2.5 cm.

DISCUSSION

Given that costs are high, surveying an entire 0.16-ha plot may not be ideal except for continuous or long-term monitoring (^{Villeda-Chávez et al. 2018}). It is important to select the easiest, fastest, and most reliable method to obtain field-based mangrove structural data with as few people as possible, especially when working in large areas with many field stations (^{Flores-Verdugo et al. 1992}). The circular plot method can be conducted quickly, while PCQM is usually more labor-intensive and requires at least one individual tree in each quarter. On the other hand, PCQM depends on a linear or semi-linear transect based on the configuration of the mangrove forest (^{Flores-Verdugo et al. 1990}). In our analyses, the maximum number of PCQM points varied as follows: 5 in RM, 10 in LR, 8 in AG, and 14 in AG shrub. Hence, covering the minimum number of 20 sampling points along the transect for PCQM (^{Cottam and Curtis 1956}) is often impossible because the forest likely includes other species. For example, selecting several parallel, field-based circular plots may be a more feasible option, as the fringe mangroves of the Urias system extend approximately 50 m from the edge of the tidal channel.

Regarding basal area and stem density, our results indicate differences between PCQM and the 3 circular plots, except for the RM class. The PCQM method overestimated the basal area by 34% in some mangrove classes, which was likely the result of high stem density and sparse distributions (e.g., LR class). ^{Hijbeek et al. (2013)} also reported large stem density differences when employing PCQM at other sites due to errors that arose when 2 mangrove classes were mixed within the same transect. Our semiarid mangrove classes exhibited a high stem density that could result in similar errors. However, other mangrove conditions exist in tropical areas with much larger basal areas and lower stem densities (^{Kovacs et al. 2010}). Thus, the PCQM method could be suitable in those large mangrove forests, as ^{Araújo and Shideler (2019)} and ^{Dookie et al. (2022)} suggested.

The distribution and composition of our mangrove classes were similar to those found in other regions of the Mexican Pacific under arid and semiarid conditions (^{Kovacs et al. 2013}). Therefore, the results of this field-based mangrove inventory analysis can be used in other sites in this region or in other arid or semiarid zones worldwide. The method that yielded the best results was the circular plot of 0.04 ha in places where tree density was less than 3,500 stems/ha (RM and AG) and the circular plot of 0.015 ha, which contained higher tree density. Moreover, the required sampling time directly increased with the amount of surveyed area (LR and AG shrub). However, using sample area plots may not be feasible in field-based inventories when evaluating sparse or open-canopy mangrove forests, such as those found in degraded environments (^{Villeda-Chávez et al. 2018}). Therefore, selecting the optimal method will benefit field-based estimates of aboveground biomass with allometric equations designed for each mangrove species in natural and rehabilitated forests. Moreover, our results provide critical additional information to improve mangrove restoration and rehabilitation strategies and provide complementary information to develop conservation policies (^{Ávila-Flores et al. 2020}). Further validation of this structural method comparison is needed in other mangrove sites with diverse physiognomic compositions, including petenes (karst soil), overwashed areas (e.g., islands), and riparian systems. The results of additional validation studies will provide a more comprehensive understanding of the applicability of the methods to different mangrove ecosystems.