Comparing Particulate Carbon Fluxes in Tropical Karst Lakes with Different Trophic Statuses

Abstract

Human activities have led to an increased influx of carbon into lakes due to changes in land use that result in higher erosion rates, eutrophication, and the introduction of organic matter. This, in turn, causes greater carbon exports and carbon accumulation in sediments. In our study, we estimated the fluxes of total particulate carbon (F_TPC), particulate organic carbon (F_POC), and particulate inorganic carbon (F_PIC) in three lakes with different trophic statuses. Two lakes, one eutrophic (Bosque Azul) and one mesotrophic (San José), are in the anthropically impacted zone of the plateau. In contrast, an oligotrophic lake (Tziscao) is in the mountainous, pristine area of “Lagunas de Montebello” National Park, a tropical karst lake district in Chiapas, Mexico. Our findings revealed that the highest F_POC values were observed in the eutrophic lake (0.47 ± 0.2 g m⁻² d⁻¹), while the highest F_PIC were observed in the mesotrophic lake (1.11 ± 0.8 g m⁻² d⁻¹). In contrast, the oligotrophic lake exhibited the lowest fluxes. Eutrophication increased the levels of F_POC, while deforestation and erosion contributed to the rise in F_PIC. Eutrophication and erosion in the lakes of LMNP led to five-, two-, and sixteen-fold increases in the F_TPC, F_POC, and F_PIC, respectively.

1. Introduction

The carbon export in inland waters has become increasingly important because these water bodies have a significant capacity for carbon deposition. The accumulation of total particulate carbon (TPC) in lakes is especially noteworthy, with a particular focus on the organic fraction known as particulate organic carbon (POC). This is essential due to the large amount of sediment that lakes accumulate and the high concentration of carbon that flows into them from their drainage basins [1].

Carbon accumulation in lakes is comparable in magnitude to carbon burial in ocean sediments despite the significant differences in the surface areas occupied by these two ecosystems [2].

Human activities have led to an increased influx of carbon into lakes, primarily through land-use changes that destabilize soil carbon and result in higher erosion rates. Additionally, alterations in nutrient levels (eutrophication) and organic matter inputs have increased POC flux to the sediments [3]. Generally, an increase in nutrient supply results in greater primary production, which leads to higher exports, an accumulation of POC, and increased rates of POC burial. If POC is of allochthonous and recalcitrant origin, it may represent a significant long-term burial [2].

Therefore, the POC flux to sediments becomes relevant, as the increase in both autochthonous (from within the waterbody) and allochthonous (from external sources, drainage basin) production contributes to high burial rates in lakes associated with anthropogenic changes [4,5,6,7]. However, the relationship between carbon fixation and preservation remains uncertain in tropical environments. Lakes in tropical regions exhibit high sedimentation rates and significant emissions per unit area, indicating that these lakes are actively processing carbon [1].

There is a limited understanding of the extent of carbon fluxes to sediments in tropical aquatic ecosystems and how biogeochemical processes influence these fluxes. Most research has focused on temperate regions [8,9]. Nonetheless, even within these temperate environments, carbon export to sediments, while recognized as significant, has not been thoroughly examined, and only a few estimates are available [2].

Estimates of carbon stocks and fluxes in Mexico’s epicontinental water bodies are scant, especially given the country’s vast size and diverse climate and topography. This lack of information is particularly pronounced compared to other countries in North America [3]. There is an urgent need to assess carbon export rates in tropical lakes, as these ecosystems play a critical role in regional and global carbon cycling contributions. Furthermore, tropical lakes are expected to be significantly affected by eutrophication and climate change in the near future [1].

In southern Mexico, “Lagunas de Montebello” National Park (LMNP) features a landscape characterized by Lower Cretaceous limestone [10]. The region has developed an extensive karst landscape, creating one of the largest lake districts in the country with 139 lakes [11]. LMNP is known for the clarity of its water bodies, which exhibit vibrant colors ranging from pale blue to turquoise and light emerald. Surrounding the park is a lush and dense forest.

The lakes can be classified into two groups based on their location. The first group consists of those lakes developed on the plateau area in the northwest region; they are fed mostly underground but receive surface waters from the Río Grande de Comitán. These lakes are hydrologically connected through artificial channels and flood in the rainy season [10,12]. The second group includes lakes found at higher altitudes in the southeast mountain region that are primarily groundwater-fed. Most human development and activities are concentrated in the plateau area, while the mountain zone is predominantly forested and still pristine.

However, in 2003, some lakes in the plateau area exhibited a bright green hue, accompanied by a “rotten egg” smell and cases of fish mortality [12]. Although this change became noticeable in 2003, disturbances began when the Tojolabales and Tseltales, people of Mayan origin, arrived between 1200 and 1521. They dug canals to connect several lakes in the plateau, facilitating the movement of people and goods by canoe among these water bodies. The area was affected as early as the 1950s, with significant increases in activity occurring during the 1980s when the Programa Nacional de Desmontes (PRONADE, National Deforestation Program) deforested a large portion of the PNLM between 1972 and 1983, and Mayan people from Guatemala (Chuj, Cakchiqueles, and K’anjobales) settled in the region, as indicated by paleolimnological studies (e.g., [13]).

Water pollution in the Río Grande de Comitán basin, particularly in the LMNP, has been linked to changes in land use. These changes include converting forested areas into agricultural land and pasture for cattle and urban development. Moreover, the discharge of domestic and industrial wastewater has exacerbated the issue. The alteration of natural water flow—through modifications to channels, wells, and water wheels—has further contributed to the decline in water quality [14,15]. Additionally, groundwater in the northwest section of the LMNP and around the Rio Grande de Comitán is likely contaminated with nutrients due to infiltration from agricultural fields and urban wastes from the upper basin [16]. The LMNP now features lakes with different trophic statuses, ranging from ultraoligo/oligotrophic (to the southeast) to eu/hypereutrophic (to the northwest).

In this work, our main aim was to compare the effects of eutrophication on particulate carbon fluxes to sediments in tropical karst lakes. To accomplish this objective, we estimated the fluxes of total particulate carbon (F_TPC), particulate organic carbon (F_POC), particulate inorganic carbon (F_PIC), seston (F_S), abioseston (F_AB), bioseston (F_BS), and chlorophyll-a concentration (F_Chl) in three lakes with different trophic statuses (oligo, meso, and eutrophic). We anticipated that fluxes to the sediments would increase from oligotrophic to eutrophic status.

2. Materials and Methods

2.1. Study Area

This study was conducted in “Lagunas de Montebello” National Park, situated in Chiapas, southeastern Mexico, near the Guatemalan border (16°04′–16°10′ N and 91°37′–91°47′ W) (Figure 1). The LMNP lies in the lower portion of the Rio Grande de Comitán sub-basin and is part of the larger Grijalva-Usumacinta hydrological region (RH30). The climate is classified as Cb(m)(f)ig [17], indicating a long, humid, and cool summer characterized by a typical summer precipitation pattern.

The area’s average annual temperature is 18.7 °C, and the annual average precipitation is 1960 mm (as recorded by the CONANP Automatic Meteorological Station N15DA7496 at 16°06′52.5″ N, 91°43′48.2″ W) (Figure 2). The dominant vegetation consists of temperate forests, with coniferous forests being the main plant associations [18].

The LMNP is in a developed karst area characterized by folded morphologies, mesiform structures, and vertical karst development. It features low elevations with slopes ranging from 11° to 57° and a karst plateau [19]. Although the lake district receives surface inflows from the Río Grande de Comitán, the primary water source for all lakes is groundwater, which has formed dolines, uvalas, and poljes through the dissolution of carbonate rocks and the structural configuration resulting from the tectonics of the area (i.e., NW-SE fault systems) [10,14]. The LMNP lies within the transcurrent fault province of the Chiapas fold-and-thrust belt, whose tectonics are closely related to the triple junction of the North American, Cocos, and Caribbean plates [20]. The local groundwater flow network in the NW region of the LMNP has a SW-NE orientation and discharges into the Río Grande de Comitán and Lake San Lorenzo [16].

The study was conducted in three lakes: Tziscao (Tz), San José (SJ), and Bosque Azul (BA) (Figure 1;

Table 1. Location and main morphometric characteristics of the studied lakes [21]. (BA: Bosque Azul; SJ: San José; Tz: Tziscao; Z_max = maximum depth; Z_mean = mean depth.)

Lake	Latitude	Longitude	Altitude	Area	Volume	Zmax	Zmean
	(°N)	(°W)	(m a.s.l)	(ha)	(km3)	(m)	(m)
BA	16.120–16.131	91.729–91.739	1458	52.5	0.01050	58	20.0
SJ	16.106–16.119	91.738–91.750	1454	60.6	0.00623	30	10.3
Tz	16.075–16.093	91.665–91.696	1490	306.6	0.08852	86	28.9

). The three selected lakes are warm-monomictic and represent a trophic gradient found in the lakes of PNLM. We chose BA in the northwest plateau, where agricultural fields have replaced forested land, and the largest anthropogenic activities occur. As a result, the plateau lakes have experienced eutrophication and are now eutrophic. Additionally, BA is a lake with no access restrictions for sampling, unlike most other lakes situated on communal lands. At the opposite end of the spectrum, we selected Tz, a mountain lake in the southeast region surrounded by forests and distant from significant anthropogenic activities. Consequently, the mountain lakes have remained pristine and exhibit an oligotrophic state. Lastly, SJ is a lake that, although situated in the plateau area, has an intermediate, mesotrophic status since it serves as a source of drinking water for the nearby community and possesses a greater degree of protection than the other lakes in the plain.

BA is part of a series of lakes artificially connected by canals. BA receives water from Lake San Lorenzo to the northwest and discharges into Lake Encantada in the southeast. All these interconnected lakes are eutrophic. In the eastern part of Tz, there is a small inflow from Lake Patianú. Both lakes are oligotrophic. Lastly, SJ has no inflows.

As mentioned, the three lakes are tropical warm-monomictic. They have an extended thermal stratification season and a brief circulation period associated with the hemispheric winter. Lakes in the northwestern region are impacted, and those in the southeast are pristine, displaying different limnological characteristics (

Table 2. Minimum (MIN), maximum (MAX), average and standard deviation (SD) of water column values (X) of the limnological variables in BA, SJ, and Tz, LMNP. Data taken from [11]. (Z_SD = Secchi disk depth, Z_EU = euphotic zone depth, T = temperature, DO = dissolved oxygen, K₂₅ = electrical conductivity, TP = total phosphorus concentration, TN = total nitrogen concentration, Chl = chlorophyll-a concentration, TSS = total suspended solids concentration.)

Variable	BA		SJ		Tz
	MIN–MAX	X ± SD	MIN–MAX	X ± SD	MIN–MAX	X ± SD
ZSD (m)	0.3–1.1	0.8 ± 0.3	1.8–4.1	3.2 ± 0.9	5.0–10.0	7.3 ± 1.3
ZEU (m)	1.6–4.8	2.7 ± 1.0	6.7–25.2	13.7 ± 7.2	21.1–28.5	25.3 ± 2.7
T (°C)	17.3–26.4	18.9 ± 1.8	17.2–24.9	20.9 ± 2.2	17.5–23.8	20.0 ± 1.9
DO (mg L−1)	0–14.7	1.3 ± 2.5	0–8.5	5.1 ± 2.8	0–8.6	5.7 ± 2.8
pH	6.8–9.3	7.7 ± 0.6	7.0–9.1	8.1 ± 0.5	7.0–9.3	8.2 ± 0.5
K25 (µS cm−1)	356.1–649.9	489.4 ± 59.3	286.6–511.3	337.7 ± 37.5	220.8–412.2	248.5 ± 19.9
TP (µmol L−1)	0.8–11.3	4.8 ± 2.6	0.5–2.3	1.3 ± 0.5	0.1–15.7	2.1 ± 3.0
TN (µmol L−1)	39.8–221.4	124.1 ± 49.9	23.2–118.3	46.1 ± 21.9	11.8–285.7	63.5 ± 52.8
Chl (µg L−1)	0.9–86.6	23.1 ± 20.7	0.2–5.7	1.2 ± 1.2	0.1–2.3	0.5 ± 0.4
TSS (mg L−1)	0.9–27.5	5.9 ± 5.0	1.0–11.4	5.2 ± 3.2	0.1–5.2	1.2 ± 0.9

), explained in detail by [11].

2.2. Fluxes

Fluxes of TPC (F_TPC), POC (F_POC), PIC (F_PIC), seston (F_S), abioseston (F_AB), bioseston (F_BS), and chlorophyll-a (F_Chl) were measured using moorings composed of one KC-Denmark sediment trap (–2-tube station) with two acrylic tubes (diameter, Ø = 80/74 mm; length, L = 450 mm; volume, V = 1.9 L; ratio L:Ø = 6) (KC Denmark, Silkeborg, Denmark). The traps were placed two meters above the bottom (at the hypolimnion) at the depths of 45 m in Tz, 14 m in SJ, and 35 m in BA to prevent capturing resuspended sediment [22]. The zone of maximum depth in each lake represents only a small fraction of the total lake bottom area. Therefore, the locations of the traps (Figure 1) were selected in areas where the depth, though below the maximum, surpassed the average depth, making them representative of each lake. Sediment traps were recovered over three consecutive periods: from 8 March to 21 June (period 1), from 21 June to 10 September (period 2), and from 10 September to 1 December (period 3) of 2023. To prevent bacterial degradation during trap exposure [23,24], a 0.005% HgCl₂ (Merck KGaA, Darmstadt, Germany) preservative (50 mg L⁻¹) [25] was added to the traps.

In each deployment, samples were taken using a 5 L UWITEC sampling bottle to determine the initial concentrations (T1) of TPC, POC, PIC, seston, and chlorophyll-a (Chl). The final concentrations (T2) were measured in the sediments recovered from the traps. An aliquot of 70–100 mL was filtered for the initial concentrations of TPC, POC, and Chl and one liter for seston; while for the final concentration, the aliquot was 1–2 mL for TPC, POC, and Chl and 10 mL for seston. The fluxes of the different components were calculated by considering the differences between the initial (T1) and final (T2) concentrations, the internal diameter of the trap, and the duration for which the traps remained in place.

After collecting the material from the traps, it was filtered the same day through 13 mm diameter Whatman GF/F filters for TPC, POC (pre-combusted at 550 °C for 4 h), and Chl, as well as 47 mm (pre-combusted at 550 °C for 4 h) filters for seston. TPC and POC were measured using a Carlo Erba NC2100 elemental analyzer (CARLO ERBA Reagents, Cornaredo, Italy) at the University of North Carolina at Wilmington. The particulate inorganic carbon (PIC) was calculated by subtracting TPC from POC.

The determination of Chl followed the Environmental Protection Agency (EPA) method 445.0 [26]. Seston was determined by gravimetry and divided into abioseston and bioseston. Bioseston (organic matter) was measured as the weight lost from the sample after calcining the filters at 550 °C for 4 h, while abioseston (inorganic matter) was calculated by subtracting bioseston from seston.

Each time the sediment traps were placed, temperature (T, °C) and dissolved oxygen (DO, mg L⁻¹) profiles were measured along the water column (resolution of 1 m) using a Hydrolab DS5 multiparameter water quality sonde (OTT Hydromet GmbH, Kempten, Germany).

2.3. Statistical Analyses

Descriptive statistics (mean, median, and standard deviation) were computed for each variable. The Shapiro–Wilk test [27] was carried out to assess the normality of the dataset (F_TPC, F_POC, F_PIC, F_S, F_AB, F_BS, and F_Chl) by the shapiro.test() function in the “stats” package [28] since the number of observations (n) was less than 50. The results revealed that the dataset did not conform to a normal distribution. Consequently, a non-parametric alternative, the Kruskal–Wallis test, assessed differences between groups (periods, lakes, and fractions). This rank-based test does not assume normality, making it appropriate for analyzing skewed or heteroscedastic data. The tests were performed by the function kruskal.test() in the “stats” package [28]. A significance level of p < 0.05 was used to determine the differences between the datasets. The variations among factor levels were tested using Dunn’s post hoc test by the function dunnTest() in the “FSA” package [29]. Correlation analyses were also conducted to obtain the Spearman correlation coefficient using the functions cor() and cor.test() to assess significance. All calculations, data handling, plotting, and statistical testing were carried out in RStudio (2024.09.0 + 375).

3. Results

3.1. Temperature and Dissolved Oxygen

The temperature profiles of the three lakes during the studied period (from March to December) showed clear thermal stratification (Figure 3). The highest temperature gradients at the thermocline were found in the eutrophic BA, while the lowest were in the oligotrophic Tz. Moreover, the thinnest Z_MIX (epilimnion) was found in the eutrophic BA, while the widest was in the oligotrophic Tz.

The DO concentration profiles were clinograde, with the development of a deep anoxic layer that encompassed the entire hypolimnion during the whole sampling period in the eutrophic BA, from June to September in the mesotrophic SJ (periods 2 and 3), and from September to December (period 3) in the oligotrophic Tz (Figure 4).

The lower thermal gradients in the thermoclines during March and December and the shorter duration of hypolimnetic anoxia for SJ and Tz suggest that the circulation period (i.e., deep mixing) likely occurred sometime between January and February.

The lower thermal gradients in the thermoclines during March and December and the shorter duration of hypolimnetic anoxia for SJ and Tz suggest that the circulation period likely occurred sometime between January and February.

The T and DO profiles indicate that the lakes’ productivity (oligo, meso, eutrophic) influences the depth of the thermocline and the intensity (°C m⁻¹) of the thermal gradient (i.e., shallower thermoclines are associated with higher temperature gradients in eutrophic lakes). Moreover, the duration and intensity of anoxia (wider anoxic layers and prolonged anoxia in eutrophic lakes) developed in the hypolimnion of these lakes is largely determined by the lakes’ productivity. Further details on the thermal structure and other limnological characteristics of the lakes are provided by [11].

3.2. Fluxes

3.2.1. Carbon

The three lakes significantly differed (p < 0.05) in the total C and C fraction fluxes. In contrast, carbon fluxes remained consistent throughout the sampling period with no significant temporal differences (p > 0.05). This is likely because the lakes were stratified for the entire sampling duration.

F_TPC was higher for BA and SJ, whereas for Tz, it was significantly lower (p < 0.05). F_POC was greater in BA and lower in Tz (p < 0.05), with SJ showing no significant difference compared to both BA and Tz (p > 0.05). F_PIC was higher in SJ (p < 0.05), while BA and Tz were lower and comparable (p > 0.05) (Figure 5,

Table 3. Minimum (MIN), maximum (MAX), median values (MED), and standard deviation (SD) of carbon fluxes (F_TPC: total particulate carbon flux; F_POC: particulate organic carbon flux; F_PIC: particulate inorganic carbon flux) in the studied lakes (BA: Bosque Azul, SJ: San José, Tz: Tziscao) during three periods. n = number of observations.

Flux	BA			SJ			Tz
	n	MIN–MAX	MED ± SD	n	MIN–MAX	MED ± SD	n	MIN–MAX	MED ± SD
FTPC (g m−2 d−1)	11	0.28–0.91	0.61 ± 0.2	9	0.28–4.31	1.50 ± 1.2	11	0.11–0.48	0.30 ± 0.1
FPOC (g m−2 d−1)	10	0.20–0.71	0.47 ± 0.2	11	0.12–1.82	0.23 ± 0.6	11	0.11–0.43	0.22 ± 0.1
FPIC (g m−2 d−1)	10	0.01–0.24	0.08 ± 0.1	8	0.07–2.63	1.11 ± 0.8	11	0.02–0.14	0.07 ± 0.1

).

3.2.2. Seston

Similar to C fluxes, there were differences in seston fluxes among the lakes (p < 0.05), though no significant temporal differences were observed (p > 0.05). Likewise, significant differences were identified for seston fluxes across their various fractions (p < 0.05); F_AB was greater than F_BS in all three lakes. SJ exhibited the highest F_S and F_AB, followed by BA, and lastly, Tz, which had the lowest fluxes (p < 0.05); F_BS was highest in BA, followed by SJ and Tz (p < 0.05) (Figure 6,

Table 4. Minimum (MIN), maximum (MAX), median values (MED), and standard deviation (SD) of seston fluxes (F_S: seston flux; F_BS: bioseston flux; F_AB: abioseston flux) in the studied lakes (BA: Bosque Azul, SJ: San José, Tz: Tziscao) during three periods. n = number of observations.

Flux	BA			SJ			Tz
	n	MIN–MAX	MED ± SD	n	MIN–MAX	MED ± SD	n	MIN–MAX	MED ± SD
FS (g m−2 d−1)	12	3.10–8.79	5.61 ± 1.5	12	4.25–40.8	9.10 ± 16.5	12	1.17–3.31	1.81 ± 0.7
FBS (g m−2 d−1)	12	1.27–2.87	1.82 ± 0.4	12	1.21–8.08	1.37 ± 2.8	12	0.35–0.90	0.53 ± 0.2
FAB (g m−2 d−1)	12	2.56–6.56	3.79 ± 1.4	12	2.96–35.0	7.82 ± 13.8	12	0.69–2.41	1.31 ± 0.5

).

3.2.3. Chlorophyll-a

The F_Chl showed variations among the three lakes (p < 0.05), with the highest fluxes in the eutrophic BA, followed by the mesotrophic SJ, and the lowest in the oligotrophic Tz. In contrast to C and seston fluxes, the lakes exhibited significant temporal differences in F_Chl (p < 0.05). In the three periods, BA presented the highest F_Chl and Tz, the lowest (p < 0.05). SJ displayed intermediate F_Chl, but in period 3, the value was similar to that of Tz (p < 0.05) (Figure 7,

Table 5. Minimum (MIN), maximum (MAX), median values (MED), and standard deviation (SD) of chlorophyll-a fluxes (F_Chl) in the studied lakes (BA: Bosque Azul, SJ: San José, Tz: Tziscao) during three periods. n = number of observations.

Flux	Period	BA			SJ			Tz
		n	MIN–MAX	MED ± SD	n	MIN–MAX	MED ± SD	n	MIN–MAX	MED ± SD
	1	6	0.35–0.96	0.65 ± 0.3	6	0.21–0.30	0.26 ± 0.0	6	0.07–0.11	0.09 ± 0.0
FChl	2	6	2.43–3.18	2.77 ± 0.3	6	0.43–0.75	0.57 ± 0.1	6	0.06–0.10	0.08 ± 0.0
(mg m−2 d−1)	3	6	1.40–1.89	1.71 ± 0.2	6	0.17–0.35	0.24 ± 0.1	6	0.17–0.31	0.23 ± 0.1
	1–3	18	0.35–3.18	1.71 ± 0.9	18	0.17–0.75	0.30 ± 0.2	18	0.06–0.31	0.10 ± 0.1

).

Some of the fluxes presented positive and significant correlations (p < 0.05), as is the case of F_BS and F_Chl, which presented the strongest correlation (ρ = 0.6), followed by F_POC and F_Chl with a moderate correlation (ρ = 0.4), unlike F_BS and F_POC, which did not present a significant correlation (ρ = 0.1, p > 0.05). Regarding the inorganic flux fraction, F_AB and F_PIC showed a moderate and significant correlation (ρ = 0.4, p < 0.05).

4. Discussion

This study evaluated and compared the particulate carbon fluxes in three tropical karst lakes with varying trophic status (eutrophic, mesotrophic, and oligotrophic) in the LMNP, Mexico. Our results confirm that both eutrophication and erosion significantly influence the F_TPC to sediments, primarily through the differential contributions of F_POC and F_PIC.

From a quantitative perspective, F_TPC was highest in the mesotrophic Lake SJ (1.50 ± 1.2 g m⁻² d⁻¹), followed by the eutrophic BA (0.61 ± 0.2 g m⁻² d⁻¹), and lowest in the oligotrophic Lake Tz (0.30 ± 0.1 g m⁻² d⁻¹). However, this pattern conceals the contrasting composition of the carbon fractions: F_POC dominated in the eutrophic lake (85% of F_TPC), while F_PIC was predominant in the mesotrophic lake (83% of F_TPC). The oligotrophic lake recorded low values for both fractions but remained primarily dominated by F_POC (76%).

These findings qualitatively and quantitatively support the hypothesis that a higher trophic status promotes increased F_POC through enhanced autochthonous primary production, particularly phytoplankton growth, as evidenced by the positive correlation between F_POC and F_Chl (ρ = 0.4, p < 0.05). The eutrophic BA exhibited the highest F_Chl (1.71 ± 0.9 mg m⁻² d⁻¹), more than five times that of SJ and nearly seventeen times that of Tz, indicating substantial autochthonous organic carbon production primarily resulting from the dominant phytoplankton species, which consist of cyanobacteria, particularly Planktothrix suspensa [30], and are probably influenced by the abundant littoral macrophytes. These results align with observations from other eutrophic systems, where increased nutrient availability stimulates primary productivity, leading to greater organic carbon export to sediments [4,5].

Interestingly, the highest F_PIC values were recorded in the mesotrophic SJ (1.11 ± 0.8 g m⁻² d⁻¹), exceeding those in BA (0.08 ± 0.1 g m⁻² d⁻¹) by more than an order of magnitude. This indicates that inorganic carbon fluxes are not strictly tied to trophic status but rather influenced by catchment conditions, particularly erosion processes. The soils surrounding SJ—especially Phaeozem and Luvisol types—are known for their high erodibility and carbonate content [31,32,33,34]. Their exposure to deforestation and agricultural conversion facilitates the transfer of particulate inorganic matter to the lake. This is further supported by the high F_AB, which comprised 85% of the seston composition in SJ.

F_AB consistently exceeded F_BS across the three studied lakes, although their proportions varied according to trophic status. In the oligotrophic Tz and mesotrophic SJ, F_AB accounted for 70–85% of the seston, reflecting a substantial input of mineral particles. However, it is important to highlight that the methodology used to differentiate between F_BS and F_AB, the LOI approach, includes diatom frustules as part of the F_AB. This distinction is particularly relevant in SJ and Tz, where diatoms are the dominant component of phytoplankton [30], and their frustules tend to sediment more readily than those of cyanobacteria [35]. In contrast, the eutrophic BA exhibited a higher proportion of F_BS (~30%), associated with elevated autochthonous productivity. This pattern aligns with observations in eutrophic tropical systems, where increased phytoplankton biomass enhances the organic fraction of seston [5,36]. The strong correlation between F_BS and F_Chl (ρ = 0.6, p < 0.05) supports this relationship, indicating that bioseston reflects in-lake biological activity. Similar relationships have been documented in temperate and tropical lakes, such as Lake Malawi and alpine hardwater systems [9,37]. These results confirm that catchment erosion and internal lake productivity influence seston composition.

The observed patterns also reflect differences in lake morphology, hydrological connectivity, and human influence. BA is part of an artificially connected lake network, receiving nutrient-rich runoff and exhibiting persistent anoxia in its hypolimnion, conditions conducive to preserving organic matter [11]. SJ, although located in the same plateau region, is hydrologically isolated and primarily receives groundwater as its input, likely carrying nutrients along with substantial inorganic material. Furthermore, residents who rely on SJ for drinking water actively contribute to the lake’s protection by avoiding activities that may pollute it [10,12]. In contrast, Tz is a high-elevation, groundwater-fed lake surrounded by intact forests with limited human disturbance, which explains its low F_TPC and near-pristine oligotrophic state [11].

Compared to the literature values for other lakes, Tz exhibited F_TPC and F_POC values that were similar to those found in other tropical oligotrophic systems, such as Lake Alchichica in Mexico (F_TPC: 0.09–0.73 g m⁻² d⁻¹; F_POC: 0.12–0.77 g m⁻² d⁻¹ [38,39,40]), and temperate lakes like Maggiore and Mergozzo in Italy (F_POC: 0.08–0.17 g m⁻² d⁻¹ [41,42]). In contrast, the fluxes in BA and SJ clearly exceeded these values and aligned more closely with the karstic Puding Reservoir in China (F_POC: 0.32 ± 0.18 g m⁻² d⁻¹), which has a mesotrophic-to-eutrophic status [43], underscoring the significant impact of human activities.

Collectively, these results demonstrate that both eutrophication and erosion resulting from land-use change can significantly impact carbon dynamics in tropical karst lakes. While eutrophication increases F_POC due to enhanced in-lake production [4,5], deforestation and soil degradation elevate F_PIC through mineral particle input [7,44]. These synergistic processes contribute to notable increases in sediment carbon burial [6,7]. The F_TPC was five times greater in eutrophic/mesotrophic lakes than in the oligotrophic lake, F_POC was twice as high, and F_PIC increased up to sixteen-fold. These considerable differences emphasize the sensitivity of carbon fluxes to local environmental pressures and highlight the importance of land and watershed management in regulating lake biogeochemistry [2,3].

Importantly, the karstic nature of the LMNP lakes indicates a significant degree of hydraulic connectivity through subterranean pathways, as noted by [12]. Thus, disturbances in one lake or sector can spread throughout the aquifer system, increasing the risk of widespread eutrophication and carbon accumulation [16]. These findings have vital implications for conservation policy in LMNP, suggesting that restoration strategies should focus on reducing nutrient inputs and addressing deforestation and soil erosion across the basin.

Future studies should seek to quantify and evaluate long-term carbon burial efficiency, the diagenetic alteration of deposited material, and seasonal and interannual variability. Combining remote sensing with catchment-scale modeling could further elucidate the relationship between land-use dynamics and particulate carbon fluxes in tropical karst systems.

5. Conclusions

Our study enhances our understanding of how higher trophic status and catchment degradation jointly shape particulate carbon fluxes in tropical inland water ecosystems. We demonstrate that impacted tropical karst lakes can show up to a 16-fold increase in inorganic carbon fluxes to sediments compared to that in pristine systems, primarily due to deforestation and the erosion of carbonate-rich soils. While higher trophic status drives increased organic carbon export through enhanced autochthonous production, land-use change amplifies inorganic loading. Crucially, we show that trophic status influences the magnitude and composition of sedimented carbon. These findings highlight the close connection between landscape processes and in-lake carbon dynamics—even in groundwater-fed systems—underscoring the vulnerability of hydraulically connected basins. Our results provide essential baseline data for tropical karst lakes, a globally overlooked system in biogeochemical research. They urge integrated catchment management strategies that connect nutrient control, land-use planning, and long-term ecological monitoring across freshwater networks.