Forest Structure and Carbon Reserve in Natural and Replanted Mangrove Forests in Different Years in the Limpopo Estuary, Gaza Province, Mozambique

Abstract

The Limpopo estuary mangrove forest covers about 928 ha; however, 382 ha remain intact, and 546 ha were degraded after the 2000 floods. Mangrove replanting campaigns were carried out at the site. This study assesses the ability of restored forests to provide carbon storage functions. The results showed that A. marina was the dominant species in all study areas. The carbon reserve of living biomass above and below ground in the natural forest was 67.9 ± 100.9 MgCha⁻¹ and 65.0 ± 77.1 MgC ha⁻¹, respectively; in the planted forests (2016, 2014, 2010), it was 1.1 ± 0.5 MgCha⁻¹ and 2.1 ± 1.0 MgCha⁻¹, 1.8 ± 1.0 MgCha⁻¹ and 3.6 ± 2.0 MgCha⁻¹, 3.7 ± 2.0 MgCha⁻¹ and 5.3 ± 2.5 MgCha⁻¹. Soil carbon reserve was 229.4 ± 119.4 MgCha⁻¹ in natural forest and 230.3 ± 134.8 MgCha⁻¹, 234.8 ± 132.7 MgC ha⁻¹, 229.4 ± 119.4 MgCha⁻¹ in planted forests (2016, 2014, 2010). The total carbon reserve in the natural forest was 362.3 MgCha⁻¹; and 233.5 MgCha⁻¹, 240.2 MgCha⁻¹ and 246.4 MgCha⁻¹ in the planted forests (2016, 2014, 2010), respectively. Natural and restored forests had similar amounts of soil carbon, which reinforces the idea that soil is a stable carbon pool. Moreover, restored forests failed to store the same amount of live biomass (carbon), which supports the idea that it is better to prevent habitat degradation than to restore it.

1. Introduction

The mangrove ecosystem is a heterogeneous habitat, with an unusual variety of plants and animals [1] adapted to grow in the intertidal zone in tropical and subtropical regions [2,3,4]. They provide a range of ecosystem goods and services to human society and coastal systems [5,6,7], and in the context of climate change, they play a significant role in climate mitigation through carbon sequestration and storage [8].

Although they represent only 0.7% of the world’s rainforest area, mangroves have been shown to contain significant carbon storage, particularly in the soil. Mangroves can store around 1023 Mgha⁻¹ in the tropics, five times more carbon than seagrass beds (200 Mg ha⁻¹) and almost twice as much as salt marshes (600 Mg ha⁻¹) [3,9,10]. Mangroves are globally threatened by, among other causes, logging, urban expansion, aquaculture, and climate change [11]. It is estimated that the global mangrove carbon stock has reduced by 1.8% (the equivalent of 158.4 Mt) between 1996 and 2016 [12], while predictions indicate that, at the current deforestation rates, another 78.50 ± 151.32 Tg CO₂ will be released to the atmosphere from 2012 to 2095 [13]. In regards to climate change in particular, the predicted increase in frequency and intensity of extreme events will reduce mangrove carbon stocks and disrupt carbon cycling and export [11]. Mangrove restoration is therefore an important management tool for climate change mitigation in terms of increasing carbon sequestration and storage [11].

Mozambique has the third-largest mangrove area in Africa, with an estimated coverage of just over 300,000 ha [14]. The greatest extent is on the central coast, followed by the northern and southern coasts [15,16]. In Southern Mozambique, the Limpopo mangrove forest covers about 928 ha [17]. Research has revealed that after the 2000 flood, around 382 ha (41.2%) remained intact and 546 ha (58.8%) were degraded [17]. This led to the development of restoration initiatives in the degraded areas that began with the involvement of the Central for Sustainable Development—Coastal Zones (CDS-ZC, a government entity) and local communities [17]. The first mangrove restoration campaign was carried out in 2010, with five main species being planted. These were: Avicennia marina, Bruguiera gymnorrhiza, Ceriops tagal, Rhizophora mucronata, and Xylocarpus granatum [17]. Subsequent planting campaigns were conducted in 2014, 2016, 2018, 2020, and 2022.

Even though there are a number of mangrove restoration initiatives in the country, very few of them are monitored in terms of the return of ecological services. Carbon storage is undeniably one of the most important ecological services provided by mangroves, and understanding how this role can be repaired with forest restoration is of paramount importance to assessing the success of restoration [18,19]. Carbon quantification in the mangrove forests of Mozambique was carried out, among other places, in the Zambezi delta, where carbon reserves were estimated at 484 MgC ha⁻¹ [20] and in Maputo Bay, with 4.59 MgC ha⁻¹ [21]. The tallest mangroves in the country are found in the provinces of Gaza and Zambezia, where riverine forests reach up to 27 m in height. These areas also have high average biomass values per hectare of 207 MgC ha⁻¹ and 97 MgCha⁻¹ [22]. Based on these studies, one can compare the storage capacity of different types and conditions of mangrove forests, as, for instance, those forests in Maputo Bay are way more impacted than those in the Zambezi delta [21]. However, very little is known about the capacity of mangrove forests to sequester carbon, and in the context of mangrove restoration for climate change mitigation, this is a key indicator of success. In other parts of the world, different patterns were found. It is known that restored mangroves in the Philippines had less biomass than natural forests but stored similar amounts of carbon in the soil [10]. In Ecuador, on the other side, younger forests had significantly less carbon than older forests [23]. Another study in Australia also showed that the carbon sequestration rate and carbon storage capacity of restored forests aged 13 and 35 doubled with age [24].

To our knowledge, there is no study in Mozambique that specifically addresses the performance of restored forests regarding carbon sequestration and storage. However, this information is essential to assessing the success of forest restoration. Moreover, carbon storage and sequestration are increasingly becoming metrics for climate change actions, as blue carbon was recently included in the countries’ NDCs. Understanding how carbon storage changes with the age of a restored forest is also important to determine the average time needed for a restored forest to perform like a natural forest in this matter. In the Limpopo estuary, mangrove restoration campaigns were conducted in different years (2016, 2014, and 2010), making it possible to compare the forest structure and estimate the amount of carbon in biomass and soil reserves. The Limpopo also has almost intact, exuberant natural stands, which represent the benchmark for a fully restored stand.

This study provides valuable information for planning and managing mangrove restoration projects in Mozambique and other countries. It can also be used to develop carbon offset programs and attract funding for potential restoration projects.

2. Materials and Methods

2.1. Study Area

This study was carried out in the Limpopo estuary, located in the Limpopo district, Zongoene administrative post, Gaza province, in Southern Mozambique (Figure 1) [25]. The estuary covers about 505 km² of surface and has a coastline that extends up to 50 km. It is located between latitudes 25°18′ and 25°48′ S and longitudes 33°19′ and 33°48′ E [26].

The total population of Zongoene is 27.404 inhabitants (55% women), with agriculture and fishing being the main livelihood activities [26]. Five mangrove species occur in the Limpopo estuary mangroves, namely: Avicennia marina (the most abundant species), Rhizophora mucronata, Bruguiera gymnorrhiza, Ceriops tagal, and Xylocarpus granatum [17,26]. The mangrove forest covers about 928 ha, out of which only 382 ha (41.2%) remain intact [17]. Mangrove restoration with community involvement started in 2010, and so far more than 100 have been restored through planting, hydrological restoration, and passive restoration [17].

2.2. Methods

2.2.1. Structural Characterization, Floristic Composition, Regeneration, and Conservation Condition of the Mangrove Forest

Sampling was carried out in the Limpopo estuary, where the systematic stratified sampling method was used. The method consists of selecting strata through an established systematization scheme, aiming to cover the population in all its extensions and obtain a uniform model [7,20,27]. Ten quadrats (10 × 10 m²) were set in each sampling area (natural forest, forest planted in 2010, 2014, and 2016), totalling 40 quadrats. Within each quadrat, all trees with a diameter greater than 2.5 cm were classified as adults, and those below these values as juveniles [28]. All adult trees were identified to the species level, counted, and the following measurements were made: height estimation using a graduated stick and measurement of the diameter at breast height (DBH) [27]. Based on these data, structural parameters were calculated, such as species frequency, density, dominance, basal area, and importance value index (VI).

To determine the condition of the forests regarding cutting intensity and dieback, all adult individuals were classified into one of the five categories of cut, namely: intact, partial cut, severe cut, stump, and natural death [27]. To determine the quality of the poles, the main trunk of adult trees was classified into one of three categories: straight pole, semi-straight pole, or crooked pole [27].

Data on the natural regeneration pattern were obtained by the linear regeneration sampling method, where within the 10 × 10 m² quadrat, 5 × 5 m² sub-quadrats were set for identification, counting, and grouping of juveniles according to regeneration classes I (RCI), for seedlings with less than 40 cm in height; RCII, for seedlings between 40 and 150 cm; and RCIIII, for trees with a height between 150 cm and 300 cm (RCIII) [19,29].

2.2.2. Estimation of Carbon Reserve

Sampling on Live Trees

Based on the DBH and height of adult trees, the biomass in living trees above_(ABG) and below ground_(BGB) was determined by the indirect method from general allometric equations [30,31].

$$\mathrm{ABG}={0.0509}^{\ast }{\mathsf{\rho }}^{\ast }{\left(\mathrm{D}\right)}^{2\ast }\mathrm{H}$$

(1)

$$\mathrm{BGB}={0.199}^{\ast }{\mathsf{\rho }}^{0.899\ast }{\left(\mathrm{D}\right)}^{2.22}$$

(2)

where D is diameter at breast height (cm), H is height (m), and p is the wood density of the respective species, obtained by the World Agroforestry Database and other authors [29,30,31]. The aboveground and belowground biomass values were added to obtain the total average biomass for each study area (Mg ha⁻¹). A conversion factor was used to estimate the carbon reserve of the vegetation, which was multiplying the biomass by 0.5 kg/C, because the biomass is approximately 50% of the dry weight [32].

Soil Sampling

Soil samples were collected, one per quadrat, with a 1m soil probe. Four sub-samples were derived from each sample, corresponding to the depth intervals 0–15 cm, 15–30 cm, 30–50 cm, and 50–100 cm. At each depth interval, a 5 cm sample of the extracted core was taken, comprising at least 30 to 50 g of sample mass [33]. Samples were stored in individual zip-lock bags and transported to the laboratory in cooler boxes for further analysis.

Laboratory Analysis

Soil samples collected in the field were weighed (wet weight) and dried in an oven at a constant temperature of 60 °C for a period of 48 h until reaching constant weight (dry weight). This weight was used to calculate bulk density, which is determined using the following equation:

$$\mathrm{Bulk}\mathrm{density}{\left(\mathrm{gcm}\right)}^3=\frac{\mathrm{sample}\mathrm{volume}\left({\mathrm{m}}^3\right)}{\mathrm{dray}\mathrm{sample}\mathrm{mass}\mathrm{in}\mathrm{the}\mathrm{oven}\left(\mathrm{g}\right)}$$

(3)

Soil samples were then crushed and homogenized. Afterwards, 10 g of each sample was placed in porcelain crucibles of known weight and incinerated at 550 °C for 3 h. The final incinerated sample was weighed again (ash-free dry weight). The organic matter content was determined based on the mass loss of the incinerated residue. The formula below was used to calculate the organic matter content:

$$\mathrm{organic}\mathrm{matter}(\%)=\frac{(P-{(\mathrm{T}-C))}^{\ast }100}{P}$$

(4)

where P is the dry weight of the sample (g), C is the tare of the empty crucible (g), and T is the ash-free dry weight of the crucible (g).

The total soil carbon reservoir was determined by multiplying the soil horizon at depth intervals (cm), bulk density (g cm⁻³), and carbon content (%). The soil carbon per sampled depth interval was calculated using the following formula:

$$\begin{array}{c}\mathrm{soil}\mathrm{carbon}({\mathrm{Mgha}}^{-1})\\ =\mathrm{bulk}\mathrm{density}{({\mathrm{gcm}}^{-3})}^{\ast }\mathrm{soil}\mathrm{depth}\mathrm{interval}{\left(\mathrm{cm}\right)}^{\ast }\%\mathrm{carbon}\mathrm{concentration}\end{array}$$

(5)

The total carbon reserve was estimated by adding all component reservoirs (carbon from above- and below-ground living and soil carbon) [33].

$$Total carbon reserve \left(Mgh{a}^{-1 }\right)=AGB+BGB+Csoil$$

(6)

where ABG is aboveground biomass, BGB is belowground biomass, and C is carbon.

The annual carbon increment rate in the planted forests was estimated by calculating the difference in total carbon from one year to another and dividing by the life span of each forest [30,34].

2.2.3. Data Analysis

Frequencies of categorical variables and calculations of averages and standard deviations for numerical variables were made. For parametric data, an ANOVA test was used, and for nonparametric data, Kruskal–Wallis with a significance level of 5% was used to measure group differences. When significant differences were found, the Tukey HSD average comparison test was used to detect the origin of the difference in groups with a 5% significance level. Regression analysis was used to evaluate the relationship between height and diameter. Other additional tests, such as Pearson correlation, were performed.

3. Results

3.1. Floristic Composition

A total of 1075 adult trees were sampled in the natural forest and planted (2016, 2014, 2010) in the Limpopo estuary. In total, five species of mangroves were identified. Avicennia marina was the most frequently observed species (60%, 46.7%, 100%, and 50%), with greater dominance (83.7%, 82.5%, 100%, and 83.2%) and density (58%, 72.9%, 100%, and 77.1%) in the natural area, as well as in the areas planted in 2016, 2014, and 2010 (

Table 1. Relative values of frequency, density, dominance, and importance value index (IV) for species recorded in natural and replanted forests in different years.

		Species
Type of Forest	Relative Values	A. marina	R. mucronata	B. gymnorhiza	C. tagal	X. granatum
Natural forest	Frequency	60	33.3	6.7	-	-
	Density	58	34.4	7.6	-	-
	Dominance	83.7	15	1.3	-	-
	IV	201.7	82.7	15.6	-	-
Planted forest 2016	Frequency	46.7	26.7	6.7	20	-
	Density	72.9	19.8	1.4	5.8	-
	Dominance	82.5	8.7	1.5	7.3	-
	IV	202.1	55.2	9.6	33.1	-
Planted forest 2014	Frequency	100	-	-	-	-
	Density	100	-	-	-	-
	Dominance	100	-	-	-	-
	IV	300	-	-	-	-
Planted forest 2010	Frequency	50	16.7	11.1	16.7	5.6
	Density	77.1	15	4.4	3	0.5
	Dominance	83.2	11.5	2.6	2.4	0.3
	IV	210.3	43.2	18.1	22.1	6.4

). Xylocarpus granatum was the rarest species, being observed only in the planted forest in 2010. This corresponded to 5.6% of frequency, 0.3% dominance, and 0.5% density. Based on importance value indices, A. marina (201.7%) was the most important species in the natural area. The same scenario is observed in the areas planted in 2016, 2014, and 2010, accounting for 202.1%, 300%, and 210.3% of the importance value, respectively (Table 1).

3.2. Structural Attributes

The tallest trees were observed in the natural area, whose average height was 4.4 ± 1.9 m. In the forests replanted in 2016, 2014, and 2010, the average heights were 2.6 ± 0.6 m, 2.9 ± 0.7 m, and 2.8 ± 0.7 m, respectively (

Table 2. Structural attributes of mangroves in the study sites (average values). SD—Standard deviation.

Type of Forest	Species	Height ± SD	DBH ± SD	Basal Area ± SD	Density ± SD	Complexity Index
Natural	A. marina (91)	5.4 ± 3.7	18.1 ± 21.5	5.6 ± 0.1	114 ± 84	3.140
	B. gymnorrhiza (12)	3.6 ± 0.5	8.8 ± 3.9	0.1 ± 0.01	120 ± 120	0.004
	R. mucronata (54)	4.2 ± 1.6	9.6 ± 12.1	1 ± 0.1	90 ± 75	0.203
	Average (52.3)	4.4 ± 1.9	12.2 ± 12.5	2.2 ± 0.07	108 ± 93	1.111
Planted 2016	A. marina (151)	3.1 ± 0.5	4.6 ± 1.7	0.3 ± 0.002	215.0 ± 82.0	2.815
	B. gymnorrhiza (3)	2.0 ± 0.6	4.7 ± 0.4	0.01 ± 0	30.0 ± 30.0	0.0001
	C. tagal (12)	1.9 ± 1.1	4.9 ± 1.6	0.02 ± 0.001	40.0 ± 43.0	0.002
	R. mucronata (41)	2.0 ± 0.3	3.0 ± 0.5	0.03 ± 0	102.0 ± 62.0	0.024
	Average (51.7)	2.6 ± 0.6	4.3 ± 1.0	0.34 ± 0.003	96.0 ± 54.0	0.710
Planted 2014	A. marina (344)	2.9 ± 0.7	5 ± 5.6	1.8 ± 0.02	415.0 ± 214.9	1.953
Planted 2010	A. marina (283)	3.7 ± 1.02	5.1 ± 2.2	0.68 ± 0.002	314.0 ± 152.0	22.311
	B. gymnorrhiza (16)	1.9 ± 0.7	3.9 ± 1.3	0.02 ± 0.001	80.0 ± 56.0	0.005
	C. tagal (11)	1.7 ± 0.2	4.6 ± 1.4	0.02 ± 0.001	36.0 ± 37.0	0.001
	R. mucronata (55)	3.9 ± 1.2	4.5 ± 1.3	0.09 ± 0.001	102.0 ± 62.0	0.374
	X. granatum (2)	2.6 ± 0.5	4.0 ± 0.7	0 ± 0	183.0 ± 56.0	0
	Average (73.4)	2.8 ± 0.7	4.4 ± 1.4	0.81 ± 0.003	126.0 ± 64.0	4.538

). Significant differences were found between sites (p < 0.001), except for areas planted in 2016 and 2014, p = 0.07 (Table 2). As for the diameter distribution, the natural area had higher DBH values with an average of 12.2 ± 12.5 m in relation to the planted areas (2016, 2014, and 2010), whose average diameters were 4.3 ± 1 m, 5 ± 5.6, and 4.4 ± 1.4 m, respectively. These differences were significant (p < 0.001), except in areas planted in 2014 and 2016, p = 0.982 (

Table A1. Summary of statistical tests of structural taxes in the study sites.

Natural Forest vs. Replanted 2016
Variable	Test	p-Value	Alpha
Species density	T	0.360	0.05
DBH	Mann Whitney	<0.001	0.05
Height	Mann Whitney	<0.001	0.05
Natural Forest vs. Replanted 2014
Variable	Test	p-Value	Alpha
Species density	T	0.014	0.05
DBH	Mann Whitney	<0.001	0.05
Height	Mann Whitney	<0.001	0.05
Natural Forest vs. Replanted 2010
Variable	Test	p-Value	Alpha
Species density	T	0.191	0.05
DBH	Mann Whitney	<0.001	0.05
Height	Mann Whitney	<0.001	0.05
Forest Replanted 2010 vs. Replanted 2016
Variable	Test	p-Value	Alpha
Species density	T	0.360	0.05
DBH	Mann Whitney	<0.001	0.05
Height	Mann Whitney	<0.001	0.05
Forest Replanted 2010 vs. Replanted 2014
Variable	Test	p-Value	Alpha
Species density	T	0.235	0.05
DBH	Mann Whitney	<0.001	0.05
Height	Mann Whitney	<0.001	0.05
Forest Replanted 2014 vs. Replanted 2016
Variable	Test	p-Value	Alpha
Species density	T	0.079	0.05
DBH	Mann Whitney	0.982	0.05
Height	Mann Whitney	0.070	0.05

).

The average density of trees in the natural area (108.0 ± 93.0 trees/ha) was relatively lower when compared to the planted areas in 2014 and 2010, whose average densities were 415.0 ± 214.9 trees/ha and 126.0 ± 64.9 trees/ha. The tests indicate that there were no significant differences between the locations (Table A1). The average basal area was greater in the natural area (2.2 ± 0.1 m²) when compared to the planted areas (2016, 2014, and 2010), whose total basal area was 0.34 ± 0.003 m², 1.8 ± 0.02 m², and 0.81 ± 0.003 m², respectively. The forest planted in 2010 was the most complex (4538) in relation to the remaining study areas (Table 2).

3.2.1. Relationship between Height and Diameter

The regression graphs between height and diameter were presented only for the species with the highest importance value index (A. marina). The box plot displays the distribution of percentiles in each case. The ends of the boxes are positioned at the 25th and 75th percentiles that correspond to the maximum and minimum values observed in the data set (Figure 2). In the natural area, most of the A. marina trees had a diameter below 25 cm and a height between 2.5 and 7 m. The correlation between the two is moderate (R² = 0.6), with significant differences (p < 0.001). In all planted areas, however, the correlation was weak. In the area planted in 2016, A. marina had more trees with a diameter below 5.2 cm and heights ranging between 2.8 and 3.2 m. The R value was R² = 0.14 (p <0.001), indicating a weak correlation between height and diameter. In the area planted in 2014, most trees were below 5 cm in diameter and had heights between 2.8 and 3.2 m, again forming a weak correlation between the two parameters (R² = 0.17, p < 0.001). In the area planted in 2010, A. marina presented most of the individuals with a diameter below 6.5 cm and heights that vary between 2.5 and 4.5 m. The R value was R² = 0.22 (p < 0.001) (Figure 2).

3.2.2. Mangrove Logging in the Natural and Planted Area

Intact trees composed the majority of trees in natural and planted areas (2016, 2014, and 2010) (Figure 3). In the natural area, 36.8% of the sampled trees of A. marina were intact, followed by R. mucronata (35.7%). There were significant differences in the average density of intact trees in the natural forest when compared to the replanted forests (Kruskal–Wallis; p < 0.001) (Figure A1). Among the planted forests, there were no differences in the mean density of intact trees (p > 0.001). The species B. gymnorhiza had the highest percentage of partially (33.3%) and severely cut trees (33.3%). In the planted areas (2016, 2014, and 2010), there were considerable percentages of partially cut (12.5%, 31.8%, and 31.6%, respectively) and severely cut (18.2%, 21.1%, respectively) A. marina trees (Figure 3) (Kruskal–Wallis; p > 0.001) (Figure A2 and Figure A3).

As for the quality of the stems, in the natural forest, 42.9% of the sampled R. mucronata trees were semi-straight, followed by B. gymnorhiza (33.3%). Avicennia marina had the highest percentage of straight and crooked stems (37.5%; 37.5%, respectively) (Figure 4). In the replanted forests (2016, 2014, and 2010), most of the sampled trees had straight stems. Semi-straight stems composed about 34% of all stems (36.8%, 35.7%, and 34.6% for 2016, 2014, and 2010, respectively), and crooked stems composed between 28% and 36% of stems (36.8%, 28.6%, and 30.8% in the same years, respectively). There were significant differences in the densities of straight poles, except for the 2014 and 2016 planted forests (Kruskal–Wallis; p ˂ 0.001) (Figure A4, Figure A5 and Figure A6).

3.2.3. Pattern of Natural Regeneration in Natural and Replanted Forests

Regeneration was observed at all study sites. The average density of juveniles in the natural area was higher for RCI (2185 ± 2323 seedlings/ha), followed by RCII (1785 ± 2331 seedlings/ha) and RCIII (331 ± 654 seedlings/ha), with the ratio 7:5:1 (RCI:RCII:RCIII). In the areas planted in 2016, the average density of juveniles was higher for RCII (779 ± 927 seedlings/ha), followed by RCI (657 ± 736 seedlings/ha) and RCIII (414 ± 740 seedlings/ha), with the ratio 2:2:1. In the areas planted in 2014 and 2010, the average density of juveniles was higher for RCI (810 ± 1083 seedlings/ha; 867 ± 851 seedlings/ha), followed by RCII (420 ± 561 seedlings/ha; 687 ± 669 seedlings/ha) and RCIII (350 ± 740 seedlings/ha; 233 ± 253 seedlings/ha), the ratios being 2:1:1 and 4:3:1, respectively (

Table 3. Juvenile density by regeneration classes.

Density (Seedlings/ha)
Area	Species	RCI	RCII	RCIII	RCI: RCII: RCIII
		0–40 cm	40.1–150 cm	150.1–300 cm	7:5:1
Natural forest	A. marina	1267 ± 2129	250 ± 378	433 ± 873
	R. mucronate	2971 ± 2334	3100 ± 2525	243 ± 447
	Total	2185 ± 2323	1785 ± 2331	331 ± 654
Replanted 2016	A. marina	567 ± 641	217 ± 214	450 ± 753	2:2:1
	C. tagal	575 ± 675	725 ± 386	600 ± 1134
	R. mucronate	875 ± 1056	1675 ± 1338	175 ± 126
	Total	657 ± 736	779 ± 927	414 ± 740
Replanted 2014	A. marina	1012 ± 1128	425 ± 634	412 ± 824	2:1:1
	C. tagal	100 ± 141	400 ± 141	100 ± 141
	Total	810 ± 1083	420 ± 561	350 ± 740
Replanted 2010	A. marina	1040 ± 693	610 ± 536	290 ± 260	4:3:1
	C. tagal	833 ± 1443	1267 ± 1026	167 ± 289
	X. granatum	50 ± 71	200 ± 141	50 ± 71
	Total	867 ± 851	687 ± 669	233 ± 253

). Based on the statistical tests, there were no significant differences in the average density of juveniles between the sites in the natural forest when compared to replanted forests (Kruskal–Wallis; p ˃ 0.001) (

Table A2. Summary of statistical tests of regeneration in the study areas.

Natural Forest vs. Replanted 2016
Variable	Test	p-Value	Alpha
RCI seedling density	Kruskal–Wallis	0.941	0.05
RCII seedling density	Kruskal–Wallis	0.370	0.05
RCIII seedling density	Kruskal–Wallis	0.735	0.05
Natural Forest vs. Replanted 2014
Variable	Test	p-Value	Alpha
RCI seedling density	Kruskal–Wallis	0.941	0.05
RCII seedling density	Kruskal–Wallis	0.370	0.05
RCIII seedling density	Kruskal–Wallis	0.334	0.05
Natural Forest vs. Replanted 2010
Variable	Test	p-Value	Alpha
RCI seedling density	Kruskal–Wallis	0.941	0.05
RCII seedling density	Kruskal–Wallis	0.735	0.05
RCIII seedling density	Kruskal–Wallis	0.334	0.05
Replanted Forest 2010 vs. Replanted2016
Variable	Test	p-Value	Alpha
RCI seedling density	Kruskal–Wallis	0.941	0.05
RCII seedling density	Kruskal–Wallis	1.000	0.05
RCIII seedling density	Kruskal–Wallis	0.849	0.05
Replanted Forest 2010 vs. Replanted 2014
Variable	Test	p-Value	Alpha
Densidade de plântulas CR I	Kruskal–Wallis	0.941	0.05
Densidade de plântulas CR II	Kruskal–Wallis	0.370	0.05
Densidade de plântulas CR III	Kruskal–Wallis	0.849	0.05
Replanted Forest 2014 vs. Replanted 2016
Variable	Test	p-Value	Alpha
Densidade de plântulas da CR I	Kruskal–Wallis	0.941	0.05
Densidade de plântulas da CR II	Kruskal–Wallis	0.370	0.05
Densidade de plântulas da CR III	Kruskal–Wallis	0.956	0.05

)

3.3. Carbon Stored in Natural and Restored Forests

3.3.1. Biomass above and below Ground

Aboveground and belowground living biomass carbon stocks varied remarkably in the natural area when compared to the planted areas. The average aboveground and belowground biomass for living trees in the natural area was estimated at 135.8 ± 201.8 Mg ha⁻¹ and 130.0 ± 154.3 Mg ha⁻¹, respectively (

Table 4. Average values of above- and below-ground biomass and carbon in natural and planted forests.

	Aboveground		Belowground
	Biomass (Mg ha−1)	Carbon (Mg ha−1)	Biomass (Mg ha−1)	Carbon (Mg ha−1)
Area	Average	Average	Average	Average
Natural forest	135.8 ± 201.8	67.9 ± 100.9	130.0 ± 154.3	65.0 ± 77.1
Planted forest 2016	2.2 ± 1.1	1.1 ± 0.5	4.3 ± 1.9	2.1 ± 1.0
Planted forest 2014	3.6 ± 2.1	1.8 ± 1.0	7.1 ± 4.0	3.6 ± 2.0
Planted forest 2010	7.4 ± 4.0	3.7 ± 2.0	10.6 ± 4.9	5.3 ± 2.5

). In the planted areas, aboveground and belowground living biomass reserves at the four-year plantation (2016) were 2.2 ± 1.1 Mg ha⁻¹ and 4.3 ± 1.9 Mg ha⁻¹ (Table 4). Aboveground and belowground biomass increased to the six-year stands (2014) and was estimated at 3.6 ± 2.1 Mg ha⁻¹ and 1.8 ± 1.0 Mg ha⁻¹ (Table 4). After ten years, the biomass reserves were estimated at 7.4 ± 4.0 Mg ha⁻¹ and 10.6 ± 4.9 Mg ha⁻¹, respectively (Table 4). Based on the tests, there were significant differences between the study areas (p ˂ 0.001), except for areas planted in 2014 and 2010 (p ˃ 0.001) (Figure A7 and Figure A8).

3.3.2. Soil Carbon Reserve

Bulk density decreases with depth and organic matter content increases with depth at all study sites, and, based on the tests, there were statistically significant differences in bulk density and organic matter content in different depth intervals (p < 0.001) (Figure 5, Figure 6, Figure 7 and Figure 8). The average amount of carbon stored in the soil was higher in replanted forests than in natural forests, and this tends to increase with depth (Figure 5, Figure 6, Figure 7 and Figure 8). The average soil carbon was estimated at 229.4 ± 119.4 Mg ha⁻¹ in the natural forest, while in the replanted forests (2016, 2014, and 2010), it was estimated at 230.3 ± 134.8 Mg ha⁻¹, 234.8 ± 132.7 Mg ha⁻¹, and 237.4 ± 142.6 Mg ha⁻¹, for years 2016, 2014, and 2010, respectively (

Table 5. Summary of statistical tests of the total values of bulk density, organic matter, and soil carbon. Average ± Standard deviation; Kruskal–Wallis.

	Natural Forest	Planted 2016	Planted 2014	Planted 2010	p-Value
Bulk density (gcm3)	0.8 ± 0.5	0.5 ± 0.4	0.8 ± 0.5	0.7 ± 0.5	0.020
Organic matter (%)	18.6 ± 13.8	35.8 ± 32.3	21.0 ± 19.2	22.1 ± 18.9	0.030
soil carbon (Mg ha−1)	229.4 ± 119.4	230.3 ± 134.8	234.8 ± 132.7	237.4 ± 142.6	0.8

). There were no statistically significant differences in total soil carbon at all study sites (p ˃ 0.001) (Table 5).

3.3.3. Total Carbon Reserve

The natural forest had the highest amount of carbon with 362.3 Mg ha⁻¹, while in the replanted forests (2016, 2014, and 2010), total carbon was estimated at 233.5 MgCha⁻¹, 240.2 MgCha⁻¹ and 246.4 MgCha⁻¹, respectively (

Table 6. Total carbon values in natural and replanted forests and annual carbon sequestration rate.

Area	Total Carbon (MgCha−1)	Annual Carbon Sequestration Rate (MgCO2ha−1 per Annum)
Natural forest	362.3	-
Planted forest 2016	233.5	3.3
Planted forest 2014	240.2	1.5
Planted forest 2010	246.4	-

). Among replanted forests, the older the forest, the greater the carbon reserve. Based on the amount of carbon that was stored in each replanted forest, we estimated an average carbon sequestration rate of 2.45 MgCO₂ha⁻¹ per year (Table 6).

4. Discussion

4.1. Floristic Composition

Five mangrove species were found in the Limpopo estuary: A. marina, C. tagal, R. mucronata, B. gymnorhiza, and X. granatum, which was very rare with only a few individuals sampled. Avicennia marina was the species with the highest density, dominance, and importance value in the natural and replanted forests. This result was expected due to several reasons. Avicennia marina is a very common species in the mangrove forests of Mozambique, and in Southern Mozambique, it is usually dominant [35]. Additionally, the species is resilient and resistant to environmental stressors. The species has a high capacity to survive in conditions of salinity variation and high salinity, has a high regeneration capacity, and is resilient to extreme events such as floods, which are common in the area [17,35].

In the case of Limpopo, where 59% of the area was wiped away during the 2000 floods, A. marina survived, while other species became very uncommon and X. granatum was extinct from the area [17]. Moreover, A. marina was widely used in the assisted restoration, which is compatible with its dominance in these stands [17].

4.2. Structural Attributes

Trees with greater height and average diameters were observed in the natural forest, indicating that this is a mature forest, while short trees with smaller diameters were found mainly in the replanted forests. Planted and natural forests have specific characteristics at different stages of development [36], and growth in height and diameter for most mangrove species is influenced by factors such as forest age, disturbance, site conditions, and suppression by the dominant species [37]. In our case, the absence of large trees in the planted sites is natural, given the fact that the trees were recently planted, and more time is required for trees to grow wider. A similar result was found in Gazi Bay, where a 12-year-old restored stand had most trees with DBH up to 11 cm, while the majority of trees had a diameter between 6.1 and 7 cm [19]. These differences in the DBH will naturally reflect in the forest basal area, where mature forests will have wider trees and thus a bigger basal area, as was found in this study [38].

Stand density in mature and young forests also tends to be remarkably different: in young forests, the land is occupied by a high density of trees with reduced diameter and height, as was observed in the planted forests. As the forest matures, some trees will die, thus reducing the stand density, but such losses are compensated by the growth of large trees, as was observed in the older natural forests [38]. Given this, it is expected that the planted forests will become more similar to the natural forests with time.

4.3. Height-to-Diameter Ratio

The relationship between diameter and height varied in the natural forest and in the replanted forests. Metabolic ecology predicts that trees should increase in diameter faster than in height. It is known that the height–DBH relationship changes with time and stabilizes when adult trees reach maturity, justifying the moderate relationship between the two parameters in the natural forest [39]. According to [40], it is important to understand tree growth patterns, and this is often a necessary variable in tree volume and biomass models.

In the replanted forests (2016, 2014, and 2010), the correlation between diameter and height for A. marina was weak to very weak, with significant differences. These results were expected, as they are often observed in forests in formation [19]. The weak relationship between height and DBH in planted forests reflects the high spatiotemporal heterogeneity in the development and succession of planted mangroves and indirectly demonstrates the sensitivity of planting to fluctuations in environmental conditions, such as climate disturbances and pathogenic outbreaks, during succession [19].

4.4. State of Conservation in the Natural and Planted Areas

As for mangrove cutting, natural forests and replanted forests are dominated by intact trees. These results also coincide with what was found in Quirimbas National Park and other areas in Mozambique where mangrove exploitation is moderate and/or the quality of stems is low [7,41]. In the Limpopo estuary in particular, these results may also be associated with community efforts to restore and preserve the mangroves, as well as to educate neighborhood communities on the same matter. There are evident signs that the Limpopo estuary community is coming to a common understanding of the importance of preserving this ecosystem, and how well-structured community management systems can contribute to successful results [17].

The mangrove species that exhibited more partial and severe cutting in the natural and planted forests were A. marina, B. gymnorhiza, and R. mucronata. Similar results were also documented by [27]. In most of East Africa, A. marina wood is used for many purposes, such as charcoal and firewood production, boat building, and traditional drums, among other products [41,42]. Rhizophora mucronata, on the other hand, is mainly used for construction and charcoal production, while the wood of B. gymnorhiza, characterized by being heavy and durable but difficult to saw and work, is used for construction, furniture, house posts, and poles [43,44]. In the natural forest, there are more semi-straight poles, while in the forests planted in 2016, 2014, and 2010, the poles are straight. This result was expected since, according to [45], in planted forests, the spacing is mostly regular and closer, which induces the competitive interaction that may be responsible for the straightness of the main stem, thus having trees with narrower stakes in planted forests compared to natural forests. Pole quality also depends on site conditions and sivicultural treatments, such as, for example, pruning.

4.5. Regeneration Rate

All forests, except those planted in 2016, have more seedlings (RCI) than saplings (RCII) and young trees (RCIII). Having more RCI is a natural behavior of forests, as these are recently recruited small plants. Under ideal conditions, competition, salinity, and predation will kill about half of the seedlings and then of the sapling, which will lead to only one young tree surviving out of six seedlings. According to [28], the minimum stock density for a sustainable forest is 6:3:1 (RCI:RCII:RCIII). None of the study sites reached these numbers. The reason for this in the natural forest may be the shading effects created by the parental canopy that prevent light from reaching the ground, thus limiting seedling growth in RCI [46,47]. In the planted forests, the absence of enough adult trees to produce seeds could explain the results. However, based on seedling densities, mangroves can potentially be considered to have good regeneration capacity.

4.6. Biomass Carbon above and below Ground

The higher biomass and carbon above and below ground observed in the natural forest can be attributed to the relatively higher DBH and height values. It is known that height and DBH are extremely important variables in tree volume and biomass models [40]. The average biomass of living trees above and below ground found in this study (135.8 Mg ha⁻¹ and 130 Mg ha⁻¹) is above the limits found by [21] in Maputo Bay (4.59 Mg ha⁻¹). On the other hand, these results are below the limits found by [20] in the Zambezi delta in the highest classes for AGB (268.5 Mg ha⁻¹) and above the limits found in the highest classes for BGB (72.8 Mg ha⁻¹). These differences may be associated with tree density, species composition, height, and DBH values that have a great influence on carbon sequestration [48]. The average biomass of living trees above and below ground in planted forests varies with the age of the forest, that is, the older the forest, the greater the estimated carbon reserve (2.2 ± 1.1 Mg ha⁻¹ and 4.3 ± 1.9 Mg ha⁻¹, forest planted 2016; 3.6 ± 2.1 Mg ha⁻¹ and 7.1 ± 4.0 Mg ha⁻¹, forest planted 2014; 7.4 ± 4.0 Mg ha⁻¹ and 10.6 ± 4.9 Mg ha⁻¹, forest planted 2010), indicating that with age, trees invest more in the stem than in other components, consequently increasing their biomass [19]. Restoration studies in Indonesia also showed that the age of planted mangroves had a significant effect on aboveground and belowground carbon stocks, ranging from 2.13 Mg ha⁻¹ to 1.0 Mg ha⁻¹ in the youngest stands (4 years), 15.81 Mg ha⁻¹ and 3.98 Mg ha⁻¹ (5 years), 46.53 Mg ha⁻¹ and 10.26 Mg ha⁻¹ (7 years) [49].

4.7. Soil Carbon

The results of this study demonstrated that the largest carbon reserves were found in the soil. These results are similar to what was found in other tropical mangrove areas, where the largest carbon reserves were found in the soil, as shown in the following reports: Zambezi delta [20], Asia-Pacific region [10], and Micronesia [50,51].

The amount of total carbon stored in the soil in this study was higher in forests planted in 2010, 2014, and 2016 (230.3 ± 134.8 MgCha⁻¹; 234.8 ± 132.7 MgCha⁻¹, and 237.4 ± 142.6 MgCha⁻¹, respectively) compared to natural forests (229.4 ± 119.4 Mg ha⁻¹). These results are similar to those found on the East Coast of India [52], where replanted forest (151.5 ± 7.9 Mg ha⁻¹) had higher soil carbon than natural forest (143.4 ± 8.2 Mg ha⁻¹). These results may be associated with the silvicultural activities that take place regularly in the replanted areas, allowing the entrance of tidal water in each population in an adequate way, thus making the tidal flow more favorable and allowing good biological activity and greater accumulation of organic matter deposited in the soil [52]. On the other hand, it could be that the carbon stored in the soil remained stable in degraded areas.

4.8. Total Carbon Reserve

The results indicate that the natural forest (362.3 Mg ha⁻¹) has a higher amount of total carbon than the forests planted in 2016, 2014, and 2010 (233.5 Mg ha⁻¹; 240.2 Mg ha⁻¹; 246.4 Mg ha⁻¹). These results may be associated with the structural patterns of the forest, average tree density, basal area, and height, which have been relatively higher in the natural forest in relation to the replanted ones. This pattern of larger-area carbon reserves represented by large trees can also be explained by the combination of nutrient input from alluvial material and tidal action that allows the assimilation of carbon by mangrove plants [53]. However, our results show that overall carbon storage in the natural mangrove forest is lower compared to those reported by other authors, such as [54], who recorded an average of 534 Mg ha⁻¹ of total carbon in the Zambezi Delta. Other higher figures were found in Kenya with 560.22 Mg ha⁻¹ [55], Dominican Republic with 853 Mg ha⁻¹ [56], and Indonesia with 879 Mg ha⁻¹ [57]. These differences may be associated with variations in tree species composition, forest structure (tree density and average diameter), forest conservation status, carbon concentration, and soil water content in each region [58]. Additionally, this study only targeted the main carbon pools (soil and above and below biomass), while others also included minor pools such as litter.

The average annual rate of carbon increase is comparable to carbon sequestration rates found in the Sundarbans, where it was found an annual increase of 1.69 Mg C ha⁻¹ a⁻¹ for live biomass and 0.012 Mg C ha⁻¹ a⁻¹ for carbon in the sediment [58]. However, this number is way below that of the Philippines, where an annual increase of 10.2 MgCha⁻¹ year was found. Carbon sequestration rates vary according to several factors, including the natural conditions of the site, such as rainfall, species composition, and temperature.

5. Conclusions

This study looked at the forest structure, conservation condition, and carbon storage of natural and replanted mangrove forests in the Limpopo Estuary, Southern Mozambique. The results of this study indicate that there are still significant differences between the natural and replanted forests; however, some forest characteristics are similar. The natural forest has taller and wider trees, which results in higher above- and below-ground biomass as well as a higher complexity index. Even though soil carbon storage was similar at all sites, the natural forest has a higher capacity to store carbon as a reflection of its higher biomass. Despite the fact that these restored forests do not fully perform like natural forests, they still deliver a number of worthy services, particularly in terms of carbon sequestration and storage and climate change mitigation. This study reinforces the importance of drawing up conservation and management plans for mangroves, as restored forests require time to recover all the functions of a natural forest. However, when forest cover is lost, mangrove planting is a viable alternative to regain at least some of it.