Application of the Compound-Specific Stable Isotopes (CSSI) Technique to Evaluate the Contributions of Sediment Sources in the Panama Canal Watershed

Abstract

Sedimentation processes have negative socioeconomic and environmental consequences. The Compound-Specific Stable Isotopes (CSSI) technique allows for the evaluation of sediment inputs associated with different land use changes in a study region. In the present work, this technique was used in the Alhajuela Lake sub-basin, within the Panama Canal Watershed. The role of the main soil contributors to the landscape (land uses, river, runoff, slope) relevant to the sediment load within the sub-basin of Alhajuela Lake was evaluated, and the relevant indicators in the landscape were selected in order to obtain the best representative sample. The contribution levels of three (3) representative land uses (Forest, Pasture and Sediment) in the study area were evaluated for the sediments present in sixteen (16) selected mixing points. The samples collected were subjected to the standard laboratory process to obtain the carbon chain isotopic values present in the fatty acids. The results of the determinations of the carbon chain fatty acid isotope ratios were evaluated using a Bayesian mixing model that takes into account the uncertainty present in the identified source values. According to the results obtained, the source identified as Sediment has a prominent contribution in most of the mixing points. The contributions of Forest land use are important in the mixing points located north of the study area. The contributions associated with Pasture land use are relevant in the points located in the proximity of this land use. The results suggest that landslides caused by high rainfalls events (Forest and Pasture sources) cause strong sedimentation to the north of Alhajuela Lake. At the same time, a high distribution of soils deposition is observed in the area surrounding Alhajuela Lake due to the strong presence of soils with Sediment source in these places. The results obtained are consistent with observations and measurements of the sediments accumulated in Alhajuela Lake between the years 2008 and 2012.

1. Introduction

Water erosion and the resulting sedimentation are natural processes caused by water, wind and ice. Some human activities such as deforestation, overgrazing, changes in land use, unsustainable farming practices and global climate change tend to accelerate land impoverishment. The result is landscape degradation, which has negative impacts on soil fertility, crop productivity, water, pollution, and potential effects on global climate, as well as sedimentation in lakes, reservoirs, etc. It is of great importance for sustainable land management to discriminate the main sediment sources in a watershed, thus identifying sites with high levels of erosion and water sedimentation.

Within the framework of an International Atomic Energy Agency (IAEA) regional project [1], it was possible to apply for the first time the technique of Compound-Specific Stable Isotopes (CSSI) analyzing the fatty acids of the carbon chains present in the watershed area of the Panama Canal. This novel technique allows estimating sediment load contributions associated with different land uses within the study region. Its application can be integrated with other nuclear techniques [2], such as Fallout Environmental Radionuclides (FRN), thus improving the tools needed to assess soil degradation.

Compound-specific stable isotope (CSSI) analysis based on the use of the stable carbon-13 isotope ($\delta$¹³C) signature of fatty acids or alkanes in soils and sediments has emerged as a very useful technique for tracing the sources and fate of eroded soils in the landscape and distinguish between types of land use, including even forested watersheds [2,3,4,5,6,7,8,9,10]. This technique was first introduced to study the origin of estuarine sediments in 2008 [11]. This sediment tracking method (CSSI) can be applied to various biomarkers (fatty acids, n-alkanes, etc.) and different isotopes ($\delta$¹³C and/or $\delta$²H) [6,7,8,10].

This CSSI technique to study the sediments origin in Latin America and the Caribbean region has been introduced in 2014 through a regional project sponsored by the IAEA [12,13]. As a result of this project, Bravo-Linares et al. [8] mapped sediment transport in the watersheds of hillside forests in Chile utilizing the CSSI technique, which permitted them to determine the principal sources of soil erosion. To localize the main sediment sources in a highly erodible agricultural watershed in Argentina, Torres-Astorga et al. [14] suggested the integrated application of two analytical techniques (CSSI and X-ray fluorescence) to choose environmental tracers. Later, Bravo-Linares et al. [9] combined FRN and CSSI techniques to reconstruct the 67-year sedimentation input history in a forested basin in central Chile.

The objective of this work is to preliminarily evaluate the role of major soil contributors to the landscape (land uses, river, runoff, slope) as potential of soil contribution relevant to the sediment load within the sub-basin of Alhajuela Lake. For the use of the CSSI technique, taking into account the use of selected landscape indicators, twenty-six (26) soils points selected among different places of relevance in the study area were sampled.

2. Alhajuela Lake

In the center of the Isthmus of Panama, within the Panama Canal Watershed, is located the study area of this work: the sub-basin of Alhajuela Lake. The selected area is of great economic importance at the national level because of its link to the Panama Canal economic zone. The Alhajuela Lake sub-basin, with a surface area of 1026 km², is the second largest within the Panama Canal Watershed and drains its waters to the reservoir of the same name.

Alhajuela Lake is, together with Gatun Lake, one of the reservoirs used to supply water to Panama Canal operation. Gatun Lake was created in 1910 to reduce the excavation and for flood protection against the flooding of the rivers that poured into the navigation channel. Alhajuela Lake was created in 1935 to support Gatun Lake and for hydro-generation. Figure 1 shows Gatun and Alhajuela Lakes with their respective sub-basins.

Since its construction, Alhajuela Lake has experienced sedimentation of eroded material from its upstream drainage. Figure 2 shows the sediment accumulation at the confluence of the Boquerón and Pequení rivers.

The climate in this area is sub-Equatorial with a dry season; this is the predominant climate throughout the Alajuela sub-basin and is characterized by annual temperature averages between 26.5 and 27.5 °C in the lowlands (at 0–200 m above sea level [m.a.s.l.]), and as elevation increases, the temperature decreases to 20 °C at 1000 m.a.s.l. Rainfall is high and ranges between 2500 and 3500 mm/year. In low regions such as Alajuela Lake, it reaches 2454 mm/year and increases towards higher elevations, exceeding 3000 mm/year. The climate has a short dry season of 3 to 4 months, lasting somewhat longer and more accentuated in the mountains and foothills [16].

Large rainfall events produce a peak in erosion and sedimentation processes. A particularly intense and long rainy event on 8 December 2010 produced landslides in the upper part of the Alhajuela sub-basin that completely filled the lake with suspended sediment for several days. Figure 3 shows the landslides, especially in areas of steep slopes, caused by the 8 December 2010 storm, and how these landslides filled the rivers with sediment.

Comparing the 69 m contour lines, the evolution of sedimentation over the years can be observed (Figure 4). As shown in Figure 4, sedimentation over time has been concentrated mainly in the northern area of Alhajuela Lake, at the confluence of the Boquerón and Pequení rivers.

The study area is located in the Alhajuela Lake sub-basin (1026 km²) within the Panama Canal Watershed (Figure 1). For the application of the CSSI technique, the main tributary rivers of Alhajuela Lake were sampled: Boquerón, Pequení, Chagres and Indio Este. Other points associated with small tributaries such as the Monocongo, La Fea, Benítez, La Palma and Quebrada Ancha rivers were also included.

The representative relief of the study area is undulating to slightly mountainous, with hills to the northwest and to the southeast of the area, between which there is an intra-mountainous depression of low altitude, as shown in Figure 5, which shows the differences in height and the relative position of the sampling points. The topography of the sub-basin is characterized by steep slopes and short riverbeds. It is made up of a mountainous terrain (500–1000 m.a.s.l.) located towards the north of the sub-basin, at the headwaters of the Chagres River, with steep and abrupt slopes with ranges greater than 45%. The hilly relief (350–500 m.a.s.l.), formed by medium peaks and summits, is located in the middle sections of the sub-basin, which have slopes of 20–45%. Finally, there is the relief of terraces (70–350 m.a.s.l.), which includes the plains and low hills around Alajuela Lake, which is more stable than the relief of mountains and hills, with slopes of 8–20% [16].

3. Materials and Methods

3.1. Sampling Strategy

The sampling strategy employed included the identification in the watershed of potential areas with established land uses that could contribute as sources of sediments and mixing points near to the lake that are of interest to assess the soil deposition (Figure 6).

Source identification considered the geographic and hydrological characteristics that allowed the selection of the different land uses and other zones in the study area, and that the sources could reach the deposition site (mixing points) in the downstream sedimentation zone.

Soil sampling at each selected point was carried out in an area of 100 m², where ten (10) soil samples of 2 cm depth were taken, from which the remains of roots and leaves were extracted, and the remaining soil was mixed in a container (composite sample). Wearing gloves to avoid contamination, a sub-sample (about 500 g) was taken and placed in a zip-lock bag from which air was extracted.

For the application of the CSSI technique, the main tributary rivers of Alhajuela Lake were sampled: Boquerón, Pequení, Chagres and Indio Este. Other points associated with small tributaries such as the Monocongo, La Fea, Benítez, La Palma and Quebrada Ancha rivers were also included; the points selected as sources are shown in red and those of mixtures in green (Figure 6).

To achieve the best representativeness of the samples to be collected in the study area, the most relevant landscape indicators were previously identified.

Table 1. Landscape indicators for soil and sediment sampling for CSSI.

Id	Indicators
1	Geomorphology (relief and slope).
2	Land use (demography, types of crops, grazing, mining).
3	Geology (formation changes and lithology).
4	Soil type (changes in formation and lithology).
5	Meteorological conditions (amount and type of precipitation, precipitation regime, temperature).
6	Type of climate.
7	Soil management (use of conservation measures, improvements, level of mechanization and vegetation cover).
8	Existence of flows (runoff, rivers).
9	Drainage network address.

shows these important indicators.

The selected landscape indicators ensure the appropriate selection of the type of sample to be collected (source or mixture), reducing the possible uncertainties that could cause a mixture between the contributing soil sources.

3.2. Laboratory Analysis

The samples collected were subjected to the standard laboratory process at the Isotopic Bioscience Laboratory of the University of Ghent, Belgium, to obtain the carbon chain isotopic values present in the fatty acids [17]. This included soil extraction, the methylation process, and final analysis of the compounds via isotope-ratio mass spectrometry (IRMS) of the methylated fatty acids (FAMEs).

Determination of the bulk organic carbon content and stable carbon isotopic composition ($\delta$¹³C) of source soils and sediment samples was performed employing an elemental analyzer (ANCA-SL, SerCon, Crew, UK) coupled to an isotope-ratio mass spectrometer (20–20, SerCon, Crew, UK). A 10–15 mg mass of the fine fraction (<0.1 mm) of each sample was weighed into tin capsules and loaded into an autosampler. Stable carbon isotope values were expressed as $\delta$¹³C values (in ‰ units) relative to the Vienna Pee Dee Belemnite (VPDB) international reference standard using the following equation:

$${\delta }^{13}{C}_{FA}=1000\times \left(\frac{Rsample}{Rstandard}-1\right)$$

(1)

where R is the molar ratio of the heavy-to-light isotopes (¹³C/¹²C) and the standard is the VPDB.

The extraction of lipids from the source material and sediment samples was carried out through accelerated solvent extraction (ASE 350, Dionex, Sydney, Australia) using dichloromethane (DCM). The volume of the total lipid extracts was reduced by evaporation at reduced pressure. Neutral and acidic compounds were separated using solid phase extraction on aminopropyl-bonded silica gel columns. The acid fraction was methylated with methanolic HCl of known carbon isotopic composition ($\delta$¹³C = −40.78 ± 0.33‰). An internal standard (MeC₁₇ FA, heptadecanoic acid methyl ester) was added to each sample prior to analysis. FAMEs were subsequently quantified using gas chromatography (GC, Trace GC, Thermo Scientific, Bremen, Germany) equipped with a flame ionisation detector. After this quantification, the solvent volumes of the samples were adapted for the most abundant FAMEs to match the ideal concentration range for capillary gas chromatography–combustion-isotope-ratio mass spectrometry analysis (GC–C-IRMS; Trace GC Ultra interfaced via a GC/C III to DeltaPLUS XP, Thermo Scientific, Bremen, Germany). FAMEs were identified based on their retention time, whereas peak purity and confirmation of identity was performed on selected samples using a parallel GC–MS measurement.

The obtained CSSI values of the FAMEs were corrected for carbon methylation during the process using the following mass balance equation:

$${\delta }^{13}{C}_{FA}=\frac{{\delta }^{13}{C}_{FAME}-\left(1-X\right){\delta }^{13}{C}_{Methanol}}{X}$$

(2)

where $X$ is the fractional contribution of free fatty acids to methylester, ${{ \delta }^{13}C}_{FA}$ is the isotopic value of fatty acids,${\delta }^{13}{C}_{FAME}$ is the fatty acid methylester isotopic value and ${\delta }^{13}{C}_{Methanol}$ is the isotopic value of methanol.

3.3. Mixing Model

The results of the determinations of the carbon chain fatty acid isotope ratios were evaluated using the MIXSIAR model ver. 3.1 [18,19] a Bayesian mixing model that takes into account the uncertainty present in the identified source values. This software allows estimating the contributions of the sources at each identified mixing point using the Markovian Chain Monte Carlo method. This model is implemented in the open-source program R and the runs were performed using RStudio 2022.07.2 Build 576 (© 2009–2022 RStudio, PBC)application. The final outputs of the program show the contribution levels of each source (probability density functions) at the mixing points and the correlations between the different sources present. For this work, the MIXSIAR model was run in normal mode with a chain length of 100,000 iterations, which improved the quality of the final estimates.

4. Results

The main characteristics of the twenty-six (26) evaluated points located in the main contributors and soil deposits near to the rivers and tributaries of the Alhajuela sub-basin are described in

Table 2. CSSI points sampled in Alhajuela sub-basin.

Sample	Point	Coordinate E (UTM)	Coordinate N (UTM)	Elevation (m)	Observation	Code
1	Escandalosa	656036	1042035	489	upper part	PAN-F01
2	Esperanza	680898	1040569	552	upper part	PAN-F02
3	Chamón	684799	1033081	656	upper part	PAN-F03
4	San Miguel	664225	1041656	532	upper part	PAN-F04
5	Dos Bocas	672246	1045265	-	upper part	PAN-F05
6	Chagresito	686106	1039070	461	upper part	PAN-F06
7	Quebrada Ancha River	658502	1029911	94	lower part	PAN-F07
8	Boquerón River Peluca Station	657947	1037415	104	upper middle part	PAN-F08
9	La Fea River	661522	1022680	128	lower part	PAN-F09
10	Chagres River CH1-2	662895	1023605	74	upper middle part	PAN-F10
11	Indio Este River INE 1	661468	1019588	75	INE-1 river mouth	PAN-M01
12	Chagres River CH1-0 Indigenous Village	662957	1024333	114	middle part	PAN-M02
13	Chagres River CH1	661670	1020488	72	CH-1	PAN-M03
14	Chagres River CDL-2	661239	1035829	78	CDL-2	PAN-M04
15	Boquerón River Mouth	658493	1034183	72	PEL-3	PAN-M05
16	Candelaria	662870	1037590	90	CDL-1	PAN-M06
17	Quebrada Bonita	653470	1032773	101	lower part	PAN-M07
18	CDL-3	659219	1034420	70	CDL-3	PAN-M08
19	Vistamares	675625	1021547	944	-	PAN-M09
20	Santa Librada	659803	1041130	136	-	PAN-M10
21	Guarumal	662593	1017814	119	-	PAN-M11
22	INE-3	675109	1017504	592	upper part	PAN-M12
23	Santo Domingo	653844	1036192	466	upper part	PAN-M13
24	La Palma River	662709	1022661	107	lower part	PAN-M14
25	Monocongo River	659952	1033962	90	lower part	PAN-M15
26	Benítez River	657478	1024878	114	lower part	PAN-M16

.

Table 3. Isotopic ratio measurements of carbon chain fatty acids.

Sample	d13C	C14:0	C15:0	C16:0	C17:0	C18:0	C19:0	C20:0	C21:0	C22:0	C23:0	C24:0	C25:0	C26:0	C27:0	C28:0	C29:0	C30:0	C31:0	C32:0	C33:0	C34:0
PAN-F01	−26.45	−35.31	−31.32	−30.5		−30.31	−32.72	−33.18		−35.97	−36.71	−35.78	−37.7	−36.21	−38.01	−35.91	−37.42	−35.5	−37.91	−36.03	−38.42	−35.87
PAN-F02	−23.16	−26.2		−25.98		−25.88		−24.77		−27.38	−25.73	−26.74	−29.41	−28.64	−31.5	−27.94	−33.12	−29.75	−33.36	−32.18	−33.28	−30.82
PAN-F03	−19.71	−18.81		−17.69	−19.46	−19.87		−20.3	−22.51	−23.66	−22.79	−24	−26.95	−27.17	−31.52	−27.95	−32.62	−29.11	−31.62	−30.8	−30.51	−28.09
PAN-F04	−18.22	−21.72		−19.14	−18.71	−20.35	−17.47	−20.12	−21	−21.25	−21.27	−22.46	−25.67	−25.28	−29.47	−26.14	−32	−27.53	−31.24	−29.04	−29.57	−26.28
PAN-F05	−21.28	−26.33	−25.54	−24.6	−24.06	−25.32	−24.31	−25.61		−27.15	−26.41	−26.67	−29.51	−27.71	−30.76	−27.5	−32.67	−29.09	−32.94	−31	−31.24	−28.68
PAN-F06	−23.13	−21.1		−19.81		−21.77		−23.05		−26.4	−25.48	−27.48	−29.81	−29.7	−34.09	−30.55	−35.57	−31.09	−35.83	−34.04	−35.98	−32.48
PAN-F07	−28.71	−33.31	−31.46	−30.1	−31.55	−29.42	−27.56	−34.15	−35.61	−34.17	−35.24	−32.45	−35.56	−33.38	−36.28	−35.18	−37.14	−36.88	−37.28	−36.57	−38.24	−35.64
PAN-F08	−26.36	−30.77	−29.82	−28.56	−29.83	−29.22	−31.4	−31.04	−35.36	−34.48	−34.22	−32.54	−34.31	−32.7	−34.44	−30.8	−34.78	−33.68	−35.47	−34.39	−36.51	−34.48
PAN-F09	−24.64	−27.45	−28.06	−25.85	−28.95	−27.51	−29.1	−29.25	-	−31.91	−31.36	−32.72	−34.58	−34	−36.26	−34.05	−36.53	−34.53	−36.28	−34.62	−36.11	−33.29
PAN-F10	−28.53	−33.89	−31.71	−29.05	−32.17	−29.99	−33.54	−35.05	−37.37	−36.2	−38.05	−35.92	−38.38	−36.6	−38.5	−36.67	−38.63	−37.57	−38.81	−37.56	−39.81	−37.92
PAN-M01	−28.46	−34.37	−32.86	−29.38	−33.22	−30.98	−35.02	−35.36	−36.68	−36.72	−37.84	−36.34	−38.08	−36.55	−37.87	−36.88	−38.78	−37.37	−39.43	−37.93	−40.54	−39.15
PAN-M02	−28.73	−31.96	−33.09	−30.2	−34.07	−31.16	−33.73	−36.8	−38.99	−36.99	−38.97	−36.96	−38.67	−36.34	−37.45	−36.36	−38.5	−37.31	−39.56	−37.9	−40.67	−39.07
PAN-M03	−28.85	−36.01	−35.62	−34.67	−36.01	−33.66	−35.02	−35.86		−36.18	−37.9	−35.26	−38.65	−36.86	−38.78	−37.23	−39.19	−37.8	−39.75	−38.14	−40.72	−39.5
PAN-M04	−29.44	−32.63	−34.13	−27.91	−33.95	−28.86	−32.71	−36.7	−38.63	−36.58	−38.81	−36.85	−38.97	−37.28	−37.92	−36.72	−38.56	−37.44	−38.56	−37.69	−40.37	−39.02
PAN-M05	−27.34	−31.84	−30.4	−26.58	−30.96	−28.64	−31.52	−32.16	−34.84	−34.55	−35.56	−34.47	−37.04	−36.08	−37.61	−35.88	−38.35	−36.2	−38.6	−36.46	−39.04	−35.85
PAN-M06	−25.35	−27.55	−28.33	−23.3	−29.09	−27.11		−29.07	−32.61	−32.9	−32.05	−31.94	−33.94	−32.13	−34.3	−31.72	−35.19	−33.55	−35.23	−34.16	−35.76	−32.76
PAN-M07	−29.29	−31.11	−31.25	−24.11	−31.91	−25.27	−33.39	−33.91	−37.69	−35.73	−37.54	−34.85	−38.58	−35.69	−37.91	−35.89	−38.13	−36.18	−38.63	−36.47	−40.2	−38.21
PAN-M08	−25.77	−28.73	−28.39	−22.85	−29.84	−26	−29.71	−30.06	−33.11	−32.56	−32.93	−30.94	−33.13	−31.22	−33.55	−30.67	−34.08	−31.93	−33.55	−31.53	−34.64	−30.85
PAN-M09	−27.32	−31.12	−30.29	−24.96	−31.8	−28.29	−33.05	−32.26	−35.57	−34.67	−35.66	−33.81	−36.87	−34.65	−37.07	−34.64	−37.56	−35.59	−37.64	−35.36	−39.25	−35.35
PAN-M10	−25.54	−30.54	−29.97	−28.04	−29.68	−29.68	−31.43	−31.69	−34	−34.28	−34.75	−33.37	−36.17	−34.48	−36.3	−33.93	−36.77	−35.54	−36.74	−35.73	−38.61	−36.83
PAN-M11	−28.69	−33.71	−32.67	−28.9	−33.02	−30.42	-	−35.27	−38.58	−35.73	−37.81	−34.97	−38.91	−36.54	−38.4	−36.64	−38.34	−37.06	−39.2	−37.33	−40.77	−38.37
PAN-M12	−28.44	−29.88	−30.11	−22.55	−30.84	−26.09	−31.73	−31.06	−38.38	−34.43	−34.81	−32.75	−36.41	−34.63	−37.17	−34.4	−36.74	−33.63	−37.32	−34.27	−40.03	−35.96
PAN-M13	−27.61	−30.88	−30.17	−27.81	−31.03	−31	-	−36.8	−37.2	−37.55	−38.85	−37.43	−39.02	−37.21	−38.59	−36.4	−38.97	−37.76	−40.26	−38.15	−40.96	−38.33
PAN-M14	−28.91	−32.47	−31.72	−27.96	−33.75	−32.22	−35.09	−36.9	−39.67	−37.1	−39.38	−36.68	−39.89	−37.21	−39.33	−37.19	−39.3	−38.05	−39.64	−37.37	−40.93	−39.15
PAN-M15	−29.16	−32.46	−32.89	−25.73	−32.66	−29.95	−46.04	−36.05	−37.93	−36.59	−38.93	−36.95	−40.02	−37.24	−38.75	−37.02	−38.87	−37.05	−39.25	−36.94	−40.5	−37.88
PAN-M16	−28.17	−31.26	−31.07	−25.19	−31.67	−28.86	−33.08	−34.62		−35.35	−37.48	−35.14	−38.04	−36.05	−38.03	−34.79	−38.13	−37.14	−39.35	−38.39	−40.73	−39.89

shows the results of the determinations of the isotopic ratios of the carbon chain fatty acids (C14 to C34) for each sampling point (PAN-F01 to PANF-10 being sources; PAN-M01 to PAN-M16 being mixtures), providing fundamental information for applying the CSSI technique.

For the application of the CSSI technique, it is essential to link the sampling points with zones representative of the predominant land uses in the study area that will allow us to subsequently determinate the contributions of the sources in the different mixing points identified.

To obtain detailed information on soil contributors in the study area, the different historical land uses reported for the region were processed using GIS tools and information downloaded from public databases [20], as shown in Figure 7 and reviewed and verified on the field studies.

A first analysis identified 10 sampling points as sources, associated with three (3) different land uses, four (4) rivers and other contributors where samples were taken. These soil contributors identified were “mature mixed broadleaf forest” in 60% of the sampled points, “water surface” in 20%, and finally “pasture” in 20% of the samples (

Table 4. Soil contributors identified as sources sampling points sampling points.

Point	Sample	Land Use
Escandalosa	PAN-F01	Mature mixed broadleaf forest
Esperanza	PAN-F02	Mature mixed broadleaf forest
Chamón	PAN-F03	Mature mixed broadleaf forest
San Miguel	PAN-F04	Mature mixed broadleaf forest
Dos Bocas	PAN-F05	Mature mixed broadleaf forest
Changresito	PAN-F06	Mature mixed broadleaf forest
Quebrada Ancha River	PAN-F07	Water surface
Boquerón River Peluca Station	PAN-F08	Pasture
La Fea River	PAN-F09	Pasture
Changres River CH1-2	PAN-F10	Water surface

).

Mature mixed broadleaf forest (re-labeled as Forest) covers more than 90% of the Alhajuela sub-basin, Pasture 4%, and water surface 4%; the remaining 2% is associated with population and other minor land uses. The water surface was also re-labeled as Sediment. In this work, the term “Sediment” for this source is not related to the soil deposition associated with the water erosion phenomena; Sediment land use represents sources sampling points located in the vicinity of water, end of catchments, preferential channels or soil deposition zones near surface water bodies (rivers or lake).

To support the representativity of the identified soil sources, the hydraulic isolation between these zones was also taken into account according to the delimitations of the micro-watersheds in order to avoid the process of negative mixing. The sources identified are significant at the study site: Forest is the main land use in the sub-basin, and Pasture is a relevant land use in the zones near the Alhajuela lake. Finally, Sediment, including the soil deposition in channels, riverbanks, etc., is a source which has been identified as important for soil contributions.

The adopted approach for the dominant potencial sources (Forest, Pasture and Sediment) was to take composite samples from each representative land-use or soil sources from several different subcatchments/micro-basins across the Alhajuela sub-basin. This procedure provides the information on variance required to assess the level of uncertainty in the modeling outputs.

The data from the carbon chain fatty acid isotope ratio determinations were used to perform the different runs using the MIXSIAR code ver. 3.1 [18,19].

The values of the isotopic ratios of C15, C17, C19, and C21 were eliminated from the analysis because the data were not complete at several source and mixture points, and because the last three (C17, C19, and C21) were associated with fatty acids produced by bacteria, not plants [17], which eliminates their usefulness for source determination from different land uses. For data analysis, the different sources points were grouped according to the identified land use. The longer chain length FA (C26,28,30) are produced by organisms other than plants, although, in this study was adopted the approach that the longer chain lengths (FA > C20) in order to enhance the discrimination capacity of the analysis and could be applied for CSSI techniques for sediment apportioning. Additionally, several studies, including the use of the longer chain length FA, have been taken into account. According to Gibbs [17], the C22:0 FAs and above are primarily of plant origin; therefore, these biomarkers should be considered more reliable. A recommendation, based on current information available in the literature, is to use very-long-chain saturated fatty acids and to avoid the use of the ubiquitous saturated fatty acids, C16 and C18 [21]. The ${{\delta }^{13}C}_{FA}$ values of C20, C22, C24, C26, C28, C30 and C32 clearly differentiated several land uses [22]. The chains C26, C28, C30 included in this research have been adopted successfully in similar studies [6,23].

Initially, the level of representativeness of the measured variables was analyzed using box plots to identify the contributions of the sources at each one of the mixing points. Figure 8 shows only one example of the box plots obtained in MIXSIAR for the mixing point PAN-M07. The criteria for this variable analysis (recommended or not) take into account whether the distribution of values is within or outside the range (mean, quartiles, upper and lower limits) of the sources and mixtures data. This analysis, carried out by the software using box plots, makes it possible to evaluate the principle of conservative behavior of the variables, to improve the possibilities of identifying the contributions of the soil deposition at the points and, additionally, to reduce the calculation times of the runs performed by discarding variables that do not allow determination of the contributions of the sources.

The range of values in the box plots for each source contributors’ type (forest, pasture, sediment) were developed using representative samples for each predominant contributors’ type. The MIXSIAR model includes these samples in order to define the source distribution library for the soil apportions calculations. The box plots include all FAs of the sampling points associated with each source contributors and these are compared with each mixing points allowing to obtain the most relevant FA chain-length.

From this initial analysis of the measured mixing samples, it was determined which FA carbon chain variables could be suppressed in the runs in order to evaluate the contributions of the sources at each one of the mixing points; these results are presented in

Table 5. Variables not recommended for the runs.

Mixture	Variables Not Recommended
PAN-M01	C14, C18, C20; C22, C24, C28, C29, C31, C32, C33, C34
PAN-M02	dC13, C16, C18, C20; C22, C23, C24, C25, C31, C32, C33, C34
PAN-M03	dC13, C14, C16, C18, C20, C25, C26, C27, C28, C29, C30, C31, C32, C33, C34
PAN-M04	dC13, C20, C22, C23, C24, C25, C26, C28, C32, C33, C34
PAN-M05	-
PAN-M06	-
PAN-M07	dC13, C25, C33, C34
PAN-M08	-
PAN-M09	-
PAN-M10	-
PAN-M11	C18, C20, C25, C31, C33, C34
PAN-M12	C33
PAN-M13	C18, C20, C22, C23, C24, C25, C26, C27, C29., C30, C31, C32, C33, C34
PAN-M14	dC13, C18, C20, C22, C23, C24, C25, C26, C27, C28, C29, C30, C31, C33, C34
PAN-M15	dC13, C20, C22, C23, C24, C25, C26, C27, C28, C29, C31, C33
PAN-M16	C31, C32, C33, C34

.

Mixing points 1 to 4 and 13 to 15 present the highest number of variables not recommended (out of range) for the analysis. This could be associated with the lack of capacity of these fatty acids for source determination for these points, and also to the lack of representativeness in the mixing points; all these factors could have an impact on the final uncertainty level of the results for these points.

In addition, as part of the exploratory study of the data, a principal component analysis was performed to assess the variability of the data set. For most of the samples, more than 96% of the values of the variables in the first two (2) components were satisfactorily explained. An example of the analysis is shown in Figure 9.

The final runs of the MIXSIAR model helped to identify the contribution levels of the different sources of land use in the selected mixing points. The program displays the results in textual and graphical form and includes probability density plots of the source inputs, histograms and correlation matrices between the different sources. Figure 10 presents an example of the graphical results.

The results of the runs for all the mixing points are summarized in

Table 6. Estimates of source contribution levels and correlations.

Mixture	Contributions of Land Uses to the Mixtures (%)						Strong Correlations	Correlation Index
	Forest	SD *	Pasture	SD *	Sediment	SD *
PAN-M01	0	0.8	2	1.7	98	1.8	Sediment/Pasture	−0.90
PAN-M02	0	0.9	2	2.0	98	2.0	Sediment/Pasture	−0.90
PAN-M03	0	0.6	1	1.3	99	1.5	Sediment/Pasture	−0.89
PAN-M04	3	1.8	3	3.2	94	3.7	Sediment/Pasture	−0.88
PAN-M05	10	4.3	20	9.7	70	6.5	Sediment/Pasture Pasture/Forest Sediment/Forest	−0.93 −0.89 0.59
PAN-M06	16	6.0	79	11.0	5	6.2	Sediment/Pasture Pasture/Forest Sediment/Forest	−0.90 −0.89 0.61
PAN-M07	4	4.9	16	9.3	80	7.0	Sediment/Pasture Pasture/Forest	−0.86 −0.67
PAN-M08	44	8.8	47	15.0	9	9.3	Sediment/Pasture Pasture/Forest	−0.84 −0.82
PAN-M09	7	5.3	31	12.2	62	8.3	Sediment/Pasture Pasture/Forest Sediment/Forest	−0.94 −0.84 0.60
PAN-M10	5	4.8	44	1.0	51	6.9	Sediment/Pasture Pasture/Forest Sediment/Forest	−0.91 −0.81 0.51
PAN-M11	1	0.8	1	1.9	98	1.9	Sediment/Pasture	−0.91
PAN-M12	27	9.1	20	18.6	53	12.2	Sediment/Pasture Pasture/Forest	−0.91 −0.83
PAN-M13	0	1.1	2	2.2	98	2.4	Sediment/Pasture	−0.88
PAN-M14	1	1.1	1	2.2	98	2.4	Sediment/Pasture	−0.89
PAN-M15	1	1.4	3	3.3	96	3.4	Sediment/Pasture	−0.91
PAN-M16	4	2.8	5	4.8	91	4.1	Sediment/Pasture Pasture/Forest	−0.81 −0.52

* SD, standard deviation.

, where the contributions of the three (3) land uses identified in the mixtures and the levels of relevant correlations identified between the sources are shown in percentages. The contributions of the source Sediment (identified in Table 4) are predominant in most (82%) of the evaluated mixing points, changing this behavior only for samples M06 and M08.

Correlations are significant (with negative sign) for Sediment/Pasture sources with index values greater than −0.80, followed by Pasture/Forest with index values greater than −0.50. When mixing points 5, 6, 9 and 10, the relationship for Sediment/Forest sources correlates significantly (with positive sign) with index values greater than 0.50.

A spatial representation of the source contributions for each mixing point is presented in a map that graphically incorporates the contributions estimated by the MIXSIAR model (Figure 11).

According to the results obtained, the land use identified as Sediment, has a prominent contribution in most of the mixing points, following a similar pattern of proportions in the points located near the Chagres River (M02, M03 and M14) and the points downstream near the Indio Este River (M01 and M11).

From a spatial point of view, the contributions of Forest land use are important in the mixing points located north of the study area in the Boquerón River (M05 and M10), in the upper and lower part of the Pequení River (M06 and M08) and in the upper part of the In-dio Este River (M09 and M12).

The contributions associated with Pasture land use are relevant in the points located in the proximity of this land use, represented in the upper and lower part of the Boquerón River (M05 and M10). The Pasture contributions also appear in the middle part of the Pe-quení River (M06), its mouth (M08) and in the upper part of the Indio Este River (M09 and M12).

It is interesting to note the change in the pattern of sediment contributions in the Indio Este River, with a balanced presence of the three land uses evaluated (Forest, Pasture and Sediment) in its upper part and the preponderance of the Sediment class in the lower part and at the mouth of the river. This change is similar to that which occurs in the Pequení River, with significant contributions of Pasture and Forest in its upper part and a majority presence of Sediment in its lower part and at the mouth of the Boquerón River. In the final part of the Pequení River, after joining the Monocongo River, when it flows into the lake, the nature of the contributions changes to a majority presence of Pasture and Forest.

The behavior of sediment contributions is different in the mixing points located near the Chagres River, where the use of Sediment soils always predominates in the upper and middle reaches and at the mouth of the river.

From the studies carried out, it can be determined that the contributions to the sediment loads in Alhajuela Lake are measured at the mixing points located in its surroundings are of Sediment in the middle and southern part of the lake, while in its northern portion there is a more balanced presence of the three land uses studied.

In the northern part of Alhajuela Lake, discharges from the mouth of the Pequení River are mainly associated with the sources Forest and Pasture, while in the discharges from the Boquerón River the source Sediment predominates followed by the source Pasture. The predominance of the Forest and Pasture sources in the Pequení River could be explained by the river’s centrality in the Alhajuela sub-basin. In general, the results point to explain what was observed in Figure 2 (confluence of the Boquerón and Pequení rivers) and Figure 3 (evolution of the 69 m contour line from 1928 to 2012), in the sense that landslides caused by high rainfall in the sub-basins of the Boquerón and Pequení rivers (Forest and Pasture sources) cause strong sedimentation of coarse material to the north of Alhajuela Lake. At the same time, there is a high distribution of fine sediments, coming mainly from erosion in the riverbeds, around and within Lake Alhajuela due to the strong presence of soils with Sediment source in these places. Figure 12 [15] sheds lighter on this phenomenon by showing the sediments accumulated in Alhajuela Lake between the years 2008 and 2012 (including the storm of 8 December 2010). Most of the sediments accumulated during this period are located in the northern part of Alhajuela Lake, near the confluence of the Boquerón and Pequení rivers (7.7 Mm³), while 7.3 Mm³ moved towards the central part of the lake and 3.3 Mm³ accumulated at the confluence of the Chagres and Indio Este rivers.

5. Conclusions

The CSSI nuclear technique was applied for the first time in Panama to evaluate preliminarly the impacts of soil deposition phenomena in a sector of the Panama Canal Watershed. Differentiated soil contributions were estimated for several contributors (the three (3) land uses, four (4) river and others, identified in the selected mixing points, characterizing their behavior. The results obtained show that Alhajuela Lake could receive the most soil from its middle and southern part, while in its northern portion there is a more balanced presence of the three (3) contributors sources studied. The sources of sediment to the terrestrial pool around the lake have been identified.

Future research will include the integration of these results using the CSSI technique with the FRN studies in order to evaluate the soil erosion rate and the redistribution in the landscale studies and the Isotope hydrology studies obtaining the water dynamic in the lake. Additionally, these studies should include direct measurements in the sediments of Alhajuela lake using drilling cores to obtain the sedimentation history in order to confirm the link with the results of preliminary soil contributions. In addition, the network of sampling points will be increased by identifying new distinctive areas that may contribute as significant sources of sediment (coffee, shrub vegetation, various crops, urban areas, etc.).