Sample preparation

The collected material was prepared in the Calcareous Microfossils laboratory at the Instituto de Investigaciones en Estratigrafía (IIES), Universidad de Caldas, following the decanting technique proposed by ^{Flores and Sierro (1997)}. This method requires weighing the amount of sediment to be decanted, which then allows a homogeneous distribution of the microfossils in an area of known dimensions, thus enabling the calculation of coccolith abundance per gram of sediment using the formula:

N= n x R² x V x r^-2 x g^-1 x v^-1.

Where n is the number of nannofossils counted in a given area, R is the radius of the Petri dish where the sample was decanted, V is the volume of water added to the dry sediment during preparation, r is the radius of the visual field of the microscope used in the count, g is the weight of the dry sediment, and v is the volume of solution added with the micropipette.

A Nikon Digital Sight DS-Ri1 camera with a DS-U2 controller attached to a Nikon Eclipse 50i POL petrographic microscope was used for the photographic records, using Nikon Imaging NIS-Elements software version 3.2. In order to obtain higher resolution and more detailed photographs, the most abundant samples were prepared for image taking using the Scanning Electron Microscope (SEM).

Qualitative and quantitative microscopic analyses

To study the coccolithophores, 39 slides were prepared and subsequently analyzed using a petrographic microscope with a magnification of 1000x. This analysis was divided into two parts. Initially, a qualitative estimate of the state of preservation of the coccoliths was carried out according to the criteria established by ^{Roth and Thierstein (1972)}; Roth (1983); ^{Raffi and Flores (1995)} where: (G) Good: The specimens present little or no dissolution and/or recrystallization. (G-M) Good-Moderate: Light to moderate dissolution and/or recrystallization and species and identification is limited. (M) Moderate: Specimens show moderate dissolution and/or excessive recrystallization. (P) Poor: The microfossils present extreme dissolution and/or excessive recrystallization.

Subsequently, all coccolithophores were taxonomically identified, following the ^{Young et al. taxonomic guide (2003)} for Pleistocene coccolith assemblages and the descriptions by ^{Perch-Nielsen (1985)} for Miocene-Pliocene reworked forms. Coccoliths of Gephyrocapsa, which were characterized by being smaller than 3 μmand not being recognized by the petrographic microscope due to their tiny size, were grouped under the name of Gephyrocapsa <3 μm. A list of the taxa found in the samples is presented in Appendix A.

Finally, a quantitative analysis was performed using two types of counts. The first was a count of 500 specimens per sample, regardless of the number of fields of view observed. In samples with a low number of coccoliths, a maximum of 400 visual fields were counted. This method guarantees, to a 95% confidence level, that specimens with 1% abundance are included within the assemblage (^{Dennison and Hay, 1967}; ^{Fatela and Taborda, 2001}). The second count was conducted in 40 visual fields, looking exclusively for those species that showed abundance <2%, in order to quantify the species underestimated or overestimated in the first count (Appendix B).

Statistical analysis

The abundance matrix used for statistical analysis was created by omitting samples with an abundance of less than 100 specimens and discarding species with abundances of less than 2% (^{Schneider et al., 2011}). This ensures that interpretations based on statistical results contain the most representative taxa of the assemblage (^{Dennison and Hay, 1967}; ^{Fatela and Taborda, 2001}; Schneider et al., 2011). Additionally, reworked forms of calcareous nannofossils were also excluded from statistical treatment in order to avoid alterations during processing.

A redundancy analysis (RDA) was performed to summarize the relationships between coccolithophores species and environmental parameters (^{Zar, 1996}; ^{Hammer et al, 2001}; ^{Whitlock and Schulter, 2009}; ^{Borcard et al, 2011}). Counts were standardized from a Hellinger-type transformation to allow low incidence taxa to be weighted in statistical analysis in the R core Team software, (2015) with the Vegan package. The significance of each environmental parameter was evaluated based on the length of the gradient in the management space and its correlation with the two main axes. The direction and size of the vectors are indicators of the influence of variables on the distribution of taxonomic groups.

Age

The presence of Emiliania huxleyi - whose appearance dates back to the Ionian (Middle Pleistocene) - in the assemblage marks the lower limit of the biozone NN21 (^{Martini, 1971}). This bioevent dates back 290,000 yrs. (^{Thierstein et al., 1977}; ^{Backman et al., 2012}). From the assemblage of coccoliths found, it is established that these sediments should be no older than the biozone NN21 of the Ionian (Middle Pleistocene).

Assemblage and abundance of coccolithophores

The results of the microscopic analysis showed that out of 39 samples analyzed, 9 samples - 8 from the Chocó Basin and 1 from the Tumaco Basin - were barren. The rest of the samples containing microfossils made it possible to identify 20 taxa of coccolithophores in both localities (Figure 4).

Figure 4 Species of coccoliths found in the analyzed samples. Images taken under Scanning Electron Microscope (SEM). A. Gephyrocapsa oceanica (10 μm). B. Gephyrocapsa oceanica (5 μm). C. Emiliania huxleyi (5 μm). D. Coccosphere of Gephyrocapsa spp. (5 μm). Images captured under petrographic microscope and enlarged by 1000x. Crossed Nicol (dark field). Parallel Nicol (light field). E. Gephyrocapsa oceanica. F. Gephyrocapsa muellerae. G. Emiliania huxleyi. H. Calcidiscus leptoporus. I. Helicosphaera carteri. J. Pontosphaera spp. K. Coccolithus pelagicus. L. Ceratolithus spp. M. Florisphaera profunda. N. Helicosphaera princei. O. Helicosphaera sellii. P. Helicosphaera wallichii. Q. Reticulofenestra pseudoumbilicus. R. Sphenolithus heteromorphus (lower) and Sphenolithus abies (higher). S-T. Discoaster variabilis.  

Among the taxa recorded, G. oceanica, G. muellerae, Gephyrocapsa <3 μm, E. huxleyi, C. leptoporus and H. carteri presented relative abundances greater than 2% (Figure 4 E-I; Table 2). In contrast species such as Ceratolithus spp., C. pelagicus, F. profunda, H. princei, H. sellii, H. wallichii y Pontosphaera spp. presented an abundance of less than 2% (Figure 4 J-P). Another common component of the assemblages were the reworked Miocene and Pliocene forms of calcareous nannofossils, mainly forms of Reticulofenestra spp. (Figure 4 Q). The sample PC-11-EGO was characterized by having the highest rework values, with a presence of D. variabilis, S. abies and S. heteromorphus. (Figure 4 R-T).

Table 2 Coccolithophores species with relative abundance >2%. 

Although the most modern specimens of coccolithophores were seen in both the Chocó Basin and the Tumaco Basin, there was a contrast in the number of coccoliths per gram (cc/g) calculated for the two regions (Table 1). Similarly, qualitative analyses of the degree of preservation of the coccoliths varied between the two basins, showing that in the Tumaco Basin, overall preservation was good, without excessive dissolution and/or recrystallization; whereas, the taxa of the Chocó Basin commonly presented moderate preservation (Table 1). However, recognition at species level was achieved in most of the specimens (Figure 5).

Figure 5 Example of the abundance and preservation of coccoliths. A. sample PC-19-MER; B. sample PC-15-EGO. Note the difference in abundance and preservation in the two samples. 

Calculations of coccoliths abundances per gram of sediment showed that the Tumaco Basin has the highest values, with an average of 1.2x107 cc/g. The highest abundances are found in the samples PC-11-EGO, PC-13-EGO, PC-14-EGO, PC-15-EGO and PC-16-EGO, with values of around 2.0x107 cc/g, while the samples with lower abundance correspond to PC-12-EGO and PC-18-EGO with values of 7.9x105 cc/g and 6.8x105 respectively. The points sampled from the Chocó Basin present lower values, with an average of 5.7x106 cc/g, where samples PC-02-MER and PC-05-MER present the highest abundance, with values of 1.1x107 and 1.5x107 cc/g respectively. A lower concentration is observed in the samples PC-15-MER and PC-17-MER with values between 1.2x106 and 1.7x106 cc/g (Figure 6; Table 2).

Figure 6 Geographic distribution and relative abundance maps of the most abundant (>2%) coccolithophores species found in the sediments. 

Multivariate analyses

Redundancy analysis RDA (^{Zar, 1996}; ^{Hammer et al., 2001}; ^{Whitlock and Schulter, 2009}; ^{Borcard et al., 2011}) provided significant information on the relationship between coccolithophore abundance, demonstrating that the most explanatory and/or correlated factor with respect to taxon distribution and abundance is related primarily to distance from the coastline (axis 1; 1.4276028; DC) compared to water column depth (axis 2; 0.5770010; WD) and sediment discharge (-0.4538382; T.a), (Table 3; Figure 7). These parameters explain 74%, 14%, and 12% respectively of the total variance. E. huxleyi, Gephyrocapsa <3 μm and G. muellerae reveal a trend towards axis 1 with negative values, while the vector of greatest magnitude on this axis is given by H. carteri, and to a lesser extent by C. leptoporus (Figure 7). As for the species G. oceanica, greater correspondence is shown with axis 2, the vector of this taxon being the closest to said axis, showing preferences for greater depth and affinity for high influence of sediments (Figure 7).

Figure 7 Trends in coccolithophore species with respect to variations with main axes, distance to coastline (DC), depth of water column (WD), and discharge of detritic material (T.a). 

Table 3 Correlation coefficient between the variables and the scores of the most representative coccoliths species in the redundancy analysis. The values in bold correspond to the parameters that presented the best correlation with the main axes. 

Abundance and preservation of coccoliths: Chocó vs. Tumaco

Descriptive and quantitative analyses of coccoliths assemblages extracted from the Chocó and Tumaco basins show a similarity in the composition of the nannoflora (Figure 4-6, Table 2). However, a difference was observed in the degree of preservation and abundance between the localities (Figure 5; Table 1). In the case of the Chocó Basin, its preservation was moderate and the number of coccoliths per gram was lower (5.7x10⁶ cc/g) compared to the Tumaco Basin. This value is related to the fact that the majority of samples presented very low abundances, 8 of 21 samples were catalogued as barren, and only 11 had a recovery higher than 100 coccoliths (Table 1). In contrast, for the Tumaco Basin, only one barren sample was obtained, estimating good preservation and an average abundance of 1.2x10⁷ cc/g (Table 1). In an attempt to explain this difference in abundance and distribution, a multivariate RDA statistical analysis was performed evaluating the available variables that may have controlled coccoliths production. This analysis revealed that fluctuations in coccoliths abundances >2% can be explained mainly by distance from the coastline (Table 3; Figures 6 y 7). The greatest abundances of E. huxleyi and Gephyrocapsa <3 μm were found in samples whose location was close to the coastline and in shallower areas (i.e., sampling points PC-15 MER, PC-19 MER, PC-01-EGO and PC-03-EGO). Although H. carteri showed low levels of abundance (Figure 6), the distribution and abundance of this species seem to be very controlled by the distance to the coastline as supported by the RDA analysis (Table 3) with the most abundance in the samples taken furthest from the continent (Figure 7). According to statistical processing, the variables with the lowest percentages of influence are: the amount of detritus per year and depth of the water column, these factors were not determinant in the abundance of coccoliths calculated for the samples examined.

Modern studies undertaken in the oceanic waters of the Chocó Basin indicate that the temperature is higher towards the coast, and higher compared to the temperature of the Tumaco Basin. In contrast, salinity is higher towards the distal zone and is more intense than that presented in the waters of the Chocó Basin (Figure 2) (CCCP, 2002; NOAA, 2017). These characteristics are closely related to the influence of ocean currents of greater magnitude, such as the Chocó Counter Current (with less cold water) in the Chocó area, and the Peru Current (colder water masses) in areas of Tumaco (CCCP, 2002). Although in general these physical-chemical characteristics may be reflected by the microfossils found in this study, the lack of quantitative data does not allow an evaluation of the impact of these variables on the abundance and preservation of coccolithophore assemblages.

Occurrence of reworking

Reworked forms were identified and quantified among the results acquired by identifying and counting the coccoliths of the 39 samples (Figure 4 Q-T). The most common reworked species were those belonging to the genus Reticulofenestra spp. that was seen in most of the samples. The PC-11-EGO sample, located south of the Tumaco Basin, showed the highest content and variety of reworked species (Figures 3 and 9).

Sphenolithus abies, Sphenolithus heteromorphus, Discoaster variabilis and Reticulofenestra spp. were the most abundant species in that sample (Figure 9). The presence of this assemblage in Pleistocene sediments enables us to infer the contribution of marine sediments from the Miocene-Pliocene to the study area. Based on the location of the PC-11-EGO sample in the slopes of a positive relief (Figure 3), it is proposed that the origin of these calcareous nannofossils may be related to processes of seabed removal, which would incorporate forms of Neogene in Quaternary sedimentation. This seems to be supported by the fact that this was the only sample with retrieval of the oldest specimens, noticing that the surrounding sites did not present forms of Sphenolithus and Discoaster. Additionally, although the course of the Mira and Patía rivers on the Neogene marine rocks exposed in Tumaco Bay may have constituted another origin of the reworked rocks (^{Borrero et al, 2012}; ^{Bedoya et al, 2013}), the exclusive concentration of Miocene-Pliocene nannofossils in the PC-11-EGO sample once again suggests more localized processes that would be providing older materials (Figure 3).