Inferring Ecosystem Function from Dissolved Organic Matter Optical Properties: A Critical Review.

Over the last 30 years, the optical property community has shifted from conducting dissolved organic matter (DOM) measurements on new complex mixtures in natural and engineered systems to furthering ecosystem understanding in the context of past, present, and future carbon (C) cycling regimes. However, the appropriate use of optical properties to understand DOM behavior in complex biogeochemical systems is of recent debate. This critical review provides an extensive survey of DOM optical property literature across atmospheric, marine, and terrestrial biospheres using a categorical approach that probes each biosphere and its subdivisions. Using this approach, a rubric of ecosystem variables, such as productive nature, C cycling rate, C inputs, and water quality, sets the foundation for interpreting commonly used optical property DOM metrics such as fluorescence index (FI), humification index (HIX), and specific ultraviolet absorbance at 254 nm (SUVA254). Case studies and a meta-analysis of each biosphere and subdivision found substantial overlap and characteristic distributions corresponding to ecosystem context for FI, HIX, and SUVA254, signifying chromophores and fluorophores from different ecosystems may be more similar than originally thought. This review challenges researchers to consider ecosystem connectivity when applying optical property results rather than making traditional "if this, then that" results-style conclusions.

Introduction

Dissolved organic matter (DOM) represents the fraction of organic matter dissolved (nominally <0.7 μm) in water and is ubiquitous in environmental systems.[1−5] The variety of DOM sources and environmental transformation mechanisms results in diverse chemical functionalities that play major roles in engineered and natural systems. Therefore, significant emphasis has been placed on the development of analytical characterization techniques that can infer specific physicochemical properties of DOM.

Over the past 50 years, there has been significant work on the characterization of DOM using optical properties (i.e., absorbance and fluorescence spectroscopy). In fact, the use of these bulk optical properties represents the main method by which most scientists and engineers characterize DOM, in great part due to their simplicity. Optically active DOM functional groups cause a collection of chromophoric and fluorophoric signals, and the underlying molecular compounds can vary as a function of biogeochemical origin.[1,6−8] Only some DOM functional groups absorb and emit light in the ultraviolet (UV) and visible regions of the spectrum (see Table for some examples of compounds). Chromophoric DOM (CDOM) is the subfraction of the total DOM pool that absorbs light, and fluorophoric DOM (FDOM) is the subfraction of the CDOM pool that re-emits light as fluorescence; thus, [DOM] > [CDOM] > [FDOM]. Although identification of specific chromophoric and fluorophoric groups has been challenging, multiple lines of evidence suggest that aromatic, phenolic, and carbonyl-containing groups are broadly important components of CDOM.[9,10] Chromophoric amino acids (e.g., tryptophan and tyrosine), polyphenols,[11] and other organic acids[12] (e.g., functional groups containing NH, NH2, and C=O) are examples of specific compounds correlated to FDOM signals across diverse systems, some related to increased biological cycling.[13,14]

Table 1

Absorbance Ranges for Model Molecules That Absorb Energy in Ultraviolet and Visible Light Wavelengths Listed with Corresponding Molecular Structuresa

Adapted from ref (15). Copyright American Chemical Society 2022.

Although optical properties do not capture the entire chemical nature of DOM, they represent many chemically active moieties in the material. For example, it is expected that CDOM represents approximately 50% of the overall carbon mass in a given sample.[9,15] Intrinsically, DOM’s optical reactivity is a function of the proportion of aromatic or other sp2 hybridized carbon (C)-containing constituents in the mixture; however, externally, many biological, physical, and other chemical variables contribute to its turnover in natural and engineered systems. As the use of optical properties for DOM characterization continues to grow in diverse biosphere environments, e.g., atmospheric, marine, and terrestrial (continental aquatic, nonaquatic, and engineered systems), it has become increasingly common to evaluate DOM physicochemistry to help infer the role of C in ecosystem function. Given that CDOM and FDOM represent only small portions of the DOM pool, is it reasonable to justify varied inferences and extensions of C cycling in ecosystem function from optical property data alone?

Use of DOM Optical Properties for Ecosystem Assessment

The appropriate use of optical properties to follow DOM behavior in complex biogeochemical systems is of recent debate. Metrics and calculations available from absorbance and fluorescence data sets have been used to probe DOM intrinsic properties, and potential reactivity in the environment commonly includes the following: absorbance and fluorescence intensity and peak position data, specific UV absorbance at 254 nm (SUVA254),[9] E2/E3,[16,17] E4/E6,[18−20] Spectral Slopes (S),[21,22] Slope Ratio,[21] Fluorescence Index (FI),[23] Humification Index (HIX),[24,25] modified Biological Index (BIX),[26,27] and the respective changes of absorbance or fluorescence intensities as a function of time and/or space. Notably, analytical and statistical grouping or modeling of regions of fluorescent features, such as the Fluorescence Regional Integration (FRI) method[28] and Parallel Factor (PARAFAC) Analysis,[29−31] are also useful tools to help interpret excitation and emission signatures from excitation and emission matrices (EEMs) and follow DOM behavior across environmental gradients or within relatively broad types of ecosystems; however, here, we target the metrics most commonly used across broad scientific disciplines in order to target the recent debate in the optical property community.

Over the last 30 years, the optical property community has seen a shift from conducting DOM optical property measurements on new complex mixtures in natural and engineered systems to furthering ecosystem understanding in the context of past, present, and future C cycling regimes. Thus, optical property measurements on DOM have become welcomed additions of larger ecology thinking with sustainability, predictability, and impact at the forefront. The important concept to remember is that optical property analyses of CDOM and FDOM determine its intrinsic properties, and this information cannot be used solely to infer the fate of that material in the environment. The role and reactivity of DOM in an ecosystem are influenced by its internal chemistry and external environmental variables such as the defining characteristics of an ecosystem or ecosystems. The extent to which DOM optical property data and metrics are useful for ecosystem interpretations is dependent on the environmental context in which the observed optical behaviors are considered.

Environmental context and a recognition of diverse production, transformation, transport, and storage mechanisms are needed to infer the role of C and/or understand ecosystem function from DOM optical property measurements. Put simply, it should complement the environmental context and not contrast it.

Scope of Review

This review positions the diverse reach of DOM optical property measurements in the spotlight, paired with environmental information, classifications, and common ecological traits to critically discuss mechanisms for ecosystem understanding and optical property relationships among diverse ecosystems from DOM characterizations. Here, we define the term “ecosystem” broadly as a compartment or region characterized by a dominant C cycle within a solid, liquid, or gas phase biosphere. This review combines the wealth of DOM optical property data across diverse ecosystems and analyzes how ecosystems are understood from a joint physicochemical and ecological perspective. We also present case studies and metadata analyses from published works that report similarities of DOM optical properties across diverse ecosystems. Using this approach, we created a rubric of ecosystem variables at the outset, such as productive nature, C cycling rate, C inputs, and water quality, to set the foundation before interpreting commonly used optical property DOM metrics (e.g., SUVA254, FI, and HIX). Notably, many researchers use absorbance and fluorescence spectroscopy without metrics reporting; just because a study reports on CDOM and FDOM does not guarantee the inclusion and interpretation of optical property metrics. This work is a compilation of the published products across diverse ecosystems that use FI, HIX, and SUVA254, to better understand DOM sources and properties. The results showcase what ecological evidence or measurement is needed to guide more accurate interpretations from DOM optical property data and where misconceptions may arise. Given the existing and comprehensive body of reviews and books on DOM optics, the goal of this review is to look forward by addressing the recent ecosystem understanding debates and challenges regarding optical property causes, their subsequent interpretations, and provide recommendations for furthering ecological understanding from bulk optical DOM measurements.

Case Studies: Natural and Engineered Ecosystem DOM and the Continuum of Optical Properties

Biosphere Ecosystem Case Studies: How Do We Understand C Cycling and Ecosystem Function from DOM Optical Property Measurements?

Optical property DOM case studies presented in previous reviews are commonly organized by environmental categories (e.g., freshwater, marine, terrestrial, or natural versus engineered).[32−39] These categories either contain or have the potential to contain and/or receive complex DOM mixtures from in situ processes or external inputs. Therefore, the DOM composition continuum of each category is complex but varies as a function of source and ecosystem processes. Within the last two decades, the application of optical property measurements for DOM characterization has extended to more diverse C reservoirs within each general category (Figure ). Here, we present case studies that represent low and high C concentration environments of three main biosphere DOM pools: atmospheric, marine, and terrestrial (continental aquatic, nonaquatic, and engineered). The objective is to highlight the broad applications of DOM optical property measurements and discuss the “truths” associated with relying on absorbance and/or fluorescence data to interpret the role of C in ecosystem function.

Figure 1

Schematic diagram of biosphere dissolved organic matter (DOM) pool categories (atmospheric, marine, and terrestrial) and subdivisions of each accompanied by their defining ecosystem characteristics from the perspective of carbon cycling. Ecosystem characteristics include relative terms for productive nature, carbon cycling rate, donor control (relating to inputs received) and extent, and water quality (WQ). Note: subdivision categories that share relatively similar ecosystem characteristics are grouped together (e.g., lakes and ponds).

We use the phrase “truths associated with absorbance and fluorescence data” as a reminder for researchers to think about the limitations of each. Figure is organized as a conceptual classification tree (as viewed from above), with branches describing major ecosystem classes and types. Circling the biosphere DOM pools and branched subdivisions are general ecological traits that further describe each ecosystem type, such as C productivity, C cycling rate, C inputs, and water quality. Relative terms (e.g., high, low, slow, fast) are provided to help researchers make connections across their DOM optical property data with ecosystem context at the outset. Although all subdivisions branch from individual biosphere DOM pools, we recognize the interactive nature of natural and engineered systems and therefore do not define robust boundaries among them. Carbon cycling rates are defined by biotic transformations, though we acknowledge that photochemical transformations play a role in C cycling. Therefore, we note in subsequent sections which environments are more susceptible to photochemical transformations that contribute to the rate of C cycling. Four water quality descriptors are provided: “blue” (low sediment, low algae, and low DOM and/or particulate concentrations), “green” (low sediment, high algae, and high DOM and/or particulate concentrations), “brown” (low sediment, low algae, and high DOM and/or particulate concentrations), and “murky” (high sediment, low algae, and high DOM and/or particulate concentrations). For atmospheric aerosols and terrestrial soils, the water quality descriptors represent the water extractable or water-soluble organic C fractions from the respective ecosystem subdivision. Donor control represents low or high C inputs from surrounding environmental features or other ecosystems, and the extent describes the level of variability or diversity of such inputs.

Atmospheric

Within the atmospheric research community, DOM is alternatively referred to as water extractable organic carbon (WEOC), water-soluble organic carbon (WSOC), brown carbon aerosols (BrC), and humic-like substances (HULIS). Common optical property measurements within the past decade have included absorbance at 254 and 300 nm (A254 and A300), MAE365, AAE330–400,[40,41]S, SUVA254, BIX, HIX, FI, and fluorescence intensities of naturally occurring fluorophores with EEMs and PARAFAC analysis.[42−49] To that effect, atmospheric CDOM and FDOM signatures are used as chemical markers prior to aquatic- and terrestrial-zone mixing and evaluated as a potential C processing stimulant.[50] Moreover, the atmosphere is not a closed system of each listed subdivision, so these characteristics mix often but were separated in Figure for simplicity to illustrate where differences among atmospheric DOM may arise. Atmospheric C reservoirs include materials produced from terrestrial and marine ecosystems and, therefore, are considered to have high donor control and extent. In these environments, C production is relatively low, and cycling rates are slow. Although the biotic C cycling rate is slow overall, these environments experience rapid photochemical transformations.

Aerosols, dust, clouds, and fog were grouped together according to characteristically similar DOM properties including brown water quality due to high WSOC and/or particulate concentrations. Aerosols are commonly measured for particle abundance (absorption coefficients), solar/optical extinction, scattering, absorbance, and secondary organic aerosol formation.[51] Natural terrestrial and anthropogenic WSOC material contains humic-like substances and thus has higher SUVA254 and HIX values (Figure ).[52−55] FI values range from low to high but are mostly centered around 1.6, describing more microbial contributions; low FI values are the result of blue shifts in fluorescence associated with microbially derived DOM[23] (Figure a). Researchers working on aerosol WSOC focus on major atmospheric reservoirs (e.g., marine sea spray, Asian yellow dust, Arctic circle dust, anthropogenic emissions, urban aerosols) and the WSOC influences on aeolian transformations (secondary organic aerosols, irradiation) and local watershed chemistry.

Figure 2

Atmospheric biosphere subdivision box plots and distributions of their dissolved organic matter calculated optical properties of (a) fluorescence index, (b) humification index, and (c) Specific UV Absorbance at λ = 254 nm (SUVA254). The number of data points is indicated next to each distribution (n = #). Data is presented for subdivisions in which peer-reviewed values were available.

Precipitation reservoirs of WSOC show different optical properties, which may be linked thermodynamically to the favorability of gas phase WSOC to condense into a liquid droplet or snow crystal. The DOM precipitation optical properties depend on the WSOC aerosols in atmospheric reservoirs and the source of such material. Precipitation (snow and rain) reservoirs are characterized by low HIX and FI. The FI average value was <1.5, indicating terrestrial influences (Figure a). Notably, these variables fluctuate with proximity to terrestrial environments and anthropogenic pollutants.[42−48]

The atmospheric community has seen a surge in DOM optical property publications most recently. Of all biosphere groups, it is the youngest community to use optical properties for DOM characterization. FI and HIX were reported across both atmospheric subcategories, yet only SUVA254, to date, has been reported for aerosols, dust, clouds, and fog (Figure a–c). This result can arise for potentially two reasons: 1) the atmospheric community routinely uses longer wavelengths for absorbance data (e.g., ≥300 nm), and thus SUVA254 values are not reported, and 2) samples may be optically dilute at cuvette pathlengths of 1 cm for absorbance readings greater than the blank samples. The FI distributions covered the same ranges for both atmospheric subcategories; however, the data was skewed to lower FI for precipitation and higher FI for aerosols. This suggests a larger contribution of microbially influenced DOM in aerosols (note that DOM in aerosols can also be created by oxidation of other atmospheric components) than in the precipitation sampled, which show more terrestrial character. Interestingly, HIX was lower for precipitation than aerosols, dust, clouds, and fog, though more data exists for studies capturing precipitation. Low HIX values indicate DOM that is more hydrogen saturated than material that is humified. In the humification process, a molecular composition shift occurs where hydrogen saturated molecules decrease and aromatic nature increases, which can be measured with DOM optical properties, i.e., HIX, with characteristic shift to longer wavelengths.[25] Our findings show that the aerosol subcategory may contain more humified DOM versus after condensation and precipitation occurs.

Marine

In Figure , marine environments are broken into three subdivisions: open ocean, sea ice, and coastal. The variety of open ocean environments is represented by four subcategories: surface waters and photic zones, dysphotic and aphotic zones, ocean floor sediments, and ocean floor seeps. The open ocean category is characterized by having a high productive nature but a slow C cycling rate overall.[56−58] The donor control and extent of inputs are low, with the DOM chemical nature dominated by in situ processes.

CDOM in the open ocean is strongly influenced by local production and degradation processes[59] and absorbs more light in the UV range in the photic zone, and SUVA254 values can be high with depth and in ocean sediments depending on existing materials and microbial processes.[60,61] The open ocean is characterized as an optically active, microbially dominated C reservoir, with less than 10% of its DOM comprised of terrestrially derived materials.[59,62] Thus, the high FI values reflect a microbially dominated pool[8,26,63−67] with relatively low HIX. Open ocean water quality is considered to be green in areas with high nutrient, algal rich waters and blue for algal-poor waters, representing a variety of DOM composition from simple to complex structures. The complex structural definition here refers to microbial degradation products and recalcitrant DOM, persisting in the water column and sediments on long-term scales.

Sea ice has characteristically low C productive nature and slow cycling rates. As it forms, it has high donor control and low extent of input materials from surrounding waters or precipitation. Sea ice DOM has low SUVA254 and low HIX.[68−71] Müller et al. (2011, 2013) described the selective incorporation of organic materials (i.e., DOM, microbes, nutrients) into sea ice, which are responsible for the low molecular weight and amino acid-like DOM optical properties in this reservoir, compared to the surrounding waters. Thermodynamic favorability causes more simple materials to be included within frozen lattices, while more complex, larger DOM molecules are excluded.[71] This phenomenon has also been observed using the optical properties of DOM for lake ice formation.[72] Sea ice has a relatively blue water quality, comprised of more simple DOM chemical structures and low concentrations of particulates and nutrients.

Coastal ecosystems are highly productive mixing zones with fast C cycling rates.[73−76] On the coastlines, donor control and extent of inputs are high, which creates DOM of high variety with high SUVA254 and HIX in considerably brown and/or murky waters. Here, DOM originates from terrestrial and microbial sources; therefore, FI values fall in the middle.[7,77−85] These general DOM characteristics incorporate the variety of coastal ecosystem types (e.g., forest, mangroves, rivers, lakes, seas, wetlands, cryosphere-ice, etc.), as well as their related water quality (e.g., blue, brown, and murky). Green water quality describes eutrophic coastal zones, and blue water quality describes cryosphere-marine mixing zones, where we expect FI values to be higher than in lower latitude regions.

Of all the biosphere communities, marine researchers have been using optical properties to study DOM and C concentrations as well as marine C character for decades.[58,60,86−88] The marine community pioneered the use of EEMs to characterize CDOM and FDOM, that then successfully extended to the terrestrial (soil), freshwater, and atmospheric communities. The FI, HIX, and SUVA254 distributions across each marine subdivision are quite varied, showcasing diversity and providing insight into which optical property metric may be most used of the three (Figure a–c). Moreover, the data reflects C sources and trends useful to help understand dominant C cycles in different parts of the ocean. The FI distributions are quite diverse for all subcategories, yet their averages fall near the original published boundaries for geochemical endmembers (FI ranging from 1.2 to 2.0).[23] Terrestrial inputs contribute most at the surface waters and in coastal zones, whereas microbial influences dominate in the dysphotic zone and in sediments and seep areas (Figure a). The most humified DOM and the greatest distribution of HIX were observed for coastal zones, which, out of all the subcategories, is the ecosystem compartment that is linked to a terrestrial source of humified material. In fact, humification can also take place within the coastal zones, therefore contributing to its large HIX range. All other subcategories had low HIX values, describing environments where humification may not be a dominant process. Reports for sea ice, thus far, have not included FI nor SUVA254. SUVA254 values for coastal zones followed the same trend as FI and HIX, highly diverse, likely from its diverse inputs (high donor control and extent). Interestingly, marine sediments and seep areas had a broad distribution of SUVA254, describing high concentrations of C and aromatic nature. However, coupling this information with low HIX and high FI suggests that microbial degradation in the sediments does not result in more humified material. Therefore, we must remember that the aromatic nature can increase without increases in humified products: higher SUVA254, low HIX, and high FI.

Figure 3

Marine biosphere subdivision bos plots and distributions of their dissolved organic matter calculated optical properties of (a) fluorescence index, (b) humification index, and (c) Specific UV Absorbance at λ = 254 nm (SUVA254). The number of data points is indicated next to each distribution (n = #). Data is presented for subdivisions in which peer-reviewed values were available.

Terrestrial

Terrestrial ecosystems are divided into three main subdivisions, soil, wetland, and aquatic (Figure ), highlighting the large variety of freshwater influenced ecosystems (different than the coastal subdivision of the marine DOM pool) that routinely use absorbance and fluorescence to characterize DOM. Soils are large C reservoirs with low C productivity and varying cycling rates.[89−93] Soil subcategories (natural and human manipulated) differ in donor control and extent, whereas human manipulated soils receive their materials from various external resources (agriculture/fertilizers, landfills, and other urban land-use and development infrastructures). Natural and human impacted soils, as well as their depth profiles, contain diverse FI, HIX, and SUVA254 values, though the general ranges presented here reflect the dominant presence of higher plant material. The water quality is noted as brown to represent the tannin-rich nature and structural complexity.[25,94−105] We recognize that soils are comprised of active microbial communities with rapid initial growth and degradation rates which produce DOM with characteristic absorbance and fluorescence properties. The FI values for these environments fall just below 1.5, reflecting the dominant nature of terrigenous C, while the higher values (FI > 1.5) insinuate more dominant microbial activity.

Wetlands and peatlands are highly productive C reservoirs with slow and fast cycling rates and varying donor control and extent depending on wetland type.[106−110] Wetland DOM waters are characterized by high SUVA254 values, higher FI, and relatively low HIX.[9,79,81,111−114] Water quality in these ecosystems is green and/or brown describing tannin- and nutrient-rich waters with a range of simple to complex chemical structures. Of course, the gradient of wetland type is extensive (e.g., marsh, fen, bog, lawn, swamp), existing both inland and on coastlines. However, here we generalize them all as high C environments with a range of complex DOM compositions.

Aquatic environments encompass a vast range of freshwater ecosystems including frozen/cryosphere, engineered, and aqueous (temporary, groundwater, rivers, canals, streams and creeks, and lakes and ponds). These three subdivisions represent major DOM categories combining characteristic productive nature and characteristic optical properties for simplicity. With low productive nature and slow C cycling rates, cryosphere ecosystems receive their DOM from snowfall, aeolian deposition, and their terrestrial and/or marine surroundings, which have blue water quality and are characterized by low SUVA254 and HIX, largely due to low C concentrations, microbial influences, and preferential incorporation of simple molecular structures into frozen layers.[23,72,115−120] FI values are centered near 1.5; however, they have a larger distribution greater than 1.5, describing a microbially dominated environment. FI values were distributed widely; the spread of values highlights the presence of DOM from atmospheric, terrigenous, and marine sources. Notably, terrigenous influences in cryosphere environments increase with greater proximity to terrestrial sources (e.g., Greenland is surrounded by continents, whereas Antarctica is surrounded by ocean).

Engineered aquatic ecosystems, i.e., treated drinking water and wastewater facilities, are ranked as highly productive and cycle C rapidly, producing DOM with unique character.[13,32,121−127] Donor control and extent are high, describing the continual input of diverse human-manipulated materials. While engineered ecosystems are dominated by microbial DOM signatures (FI: high), they experience variable HIX and SUVA254 depending on the source and reactivity of DOM and the treatment process applied. For example, membrane[128] and oxidative[129] processes reduce the presence of larger, aromatic moieties in DOM, decreasing HIX postprocessing. Alternatively, preferential uptake of labile-DOM by microbial processes leaves complex, aromatic moieties behind,[130] increasing overall HIX. Engineered ecosystems allow for a unique look into DOM character which exhibits more complex chemical species than typically seen in characteristically green waters.

The aqueous subdivision contains the most subcategories and further divisions (Figure ) to clearly differentiate between the varying ecosystems. Temporary aqueous ecosystems include flash flood and urban runoff ephemeral waters with low residence times, low C productivity, and overall slow cycling rates. Based on small or large quantities of input waters, these ecosystems have high donor control and high extent, receiving a variety of materials that contribute to its flowing waters (e.g., Figures 1 and 4 of urban catchment schematic diagrams in Fork et al. (2018)).[131] The DOM from temporary aqueous environments has high SUVA254 and HIX and low FI, describing complex molecular moieties in blue, brown, and murky waters.[132−136] Groundwater is listed as a separate category based on spatial separation from surface waters such as rivers, canals, streams, and creeks. Yet, some DOM characteristics overlap and therefore are grouped together in Figure to show such similarities. Groundwater, canals, and oligotrophic rivers have a low productive nature but fast cycling rates due to microbial turnover (groundwater) and, without obstruction, photochemical transformations (rivers and canals). Donor control is high, but extent is low; these waters have very low nutrient concentrations. Eutrophic traveling waters have high productive nature and fast cycling rates with abundant nutrients. Donor control and extent is high for eutrophic conditions since these waters gain subsidies from diverse microbial communities, higher plants, and soils.[137−141] With diverse inputs, we expected a great variety for SUVA254, FI, and HIX. However, the data showed FI values centered around 1.5, low HIX, and high SUVA254. Hence, the distribution of data is diverse but not the general characteristics. These waters can have diverse water qualities: oligotrophic waters are blue, eutrophic are green, tannin-rich waters are brown, and high sediment content is murky. Notably, some of these traveling waters can have more than one classification of water quality, especially for eutrophic waters that can be green and murky or tannin-rich waters that have a large amount of sediment in the water column.

As an overall group, nonsupraglacial streams and creeks have a high productive nature and fast C cycling rates with relatively high donor control and extent of inputs.[142,143] These variables are subject to change at lower and higher elevations with varying surroundings. For example, higher elevation headwater streams may not experience the same diverse higher-plant inputs than at lower elevations and, therefore, may have a low productive nature and slower C cycling rates, describing the differences between their blue and brown water qualities (streams adjacent to forests, wetlands, etc.). Absorbance values also fluctuate with elevation changes, but overall, C in streams and creeks is dominated by terrestrial inputs (vegetation and soils), so their SUVA254 values are high. However, high SUVA254 values do not always correlate to high inputs of terrestrial material; microbial degradation products can also produce the same results. These examples coupled with FI and HIX and the ecosystem characteristics help us better understand where the C is coming from and what processes are occurring. Streams and creeks also had high HIX values and FI values centered around 1.5, which describe more humified compounds from internal degradation processes such as in situ microbial degradation and photochemical transformations with sunlight exposure, i.e., not in caves or with a thick canopy cover (Figure ).[144−149]

Figure 4

Terrestrial biosphere subdivision box plots and distributions of their dissolved organic matter calculated optical properties of (a) fluorescence index, (b) humification index (HIX), and (c) Specific UV Absorbance at λ = 254 nm (SUVA254). The number of data points is indicated next to each distribution (n = #). Data is presented for subdivisions in which peer-reviewed values were available.

Lakes and ponds are listed with oligotrophic and eutrophic divisions, like rivers. A higher productive nature with fast C cycling rates occurs in eutrophic lakes and ponds, whereas oligotrophic waters experience the opposite (Figure ). Donor control and extent are high, which play a role in the various DOM optical properties.[150−155] Like the open ocean, lakes and ponds may have considerably different DOM types with depth, mixing, and stratification, which also leads to a range of SUVA254, FI, and HIX values. Distributions of data were the least variable for SUVA254 and the most variable for FI. SUVA254 and HIX values were low, and FI values were mid to high, describing a dominance of microbial processes over terrestrial C inputs.

Lastly, lakes, ponds, rivers, canals, streams, and even creeks have their own benthic C reservoir buried in sediments. The productive nature in sediments is low, and C cycling rates are slow; and we acknowledge the dominance of microbial activity in these environments (in both oxic and anoxic zones) contributing to cycling. Donor control and extent are high as sediments accumulate sinking C over time. DOM optical properties vary as a function of microbial activities and processes,[156−158] which may explain the clustered data near FI = 1.5 and larger distributions at both low and high FI values. Water quality can range across all identifiers, depending on the amount of C deposited, resident microbial communities, and nutrient availability. Therefore, these ecosystems may contain a diverse array of simple and complex molecular moieties.

Like the marine biosphere, the terrestrial biosphere FI, HIX, and SUVA254 distributions across each subcategory are also quite unique (Figure a–c), and again, the data reflects sources and trends useful to help understand dominant C cycles in different continental compartments. The highest FI averages were reported for human manipulated soils, followed by lakes and ponds, wetlands, and the cryosphere, where microbial communities dominate C cycling processes. All other averages were near or below FI values of 1.5, which does not discount microbial influences but instead describes ecosystems with higher-plant inputs and plenty of them. HIX distributions were greatest for natural soils, temporary flow waters, and rivers, canals, and groundwaters; however, their averages were considerably low. Therefore, these ecosystems contain a great variety of humified and nonhumified DOM that is likely linked hydrologically throughout their watersheds. SUVA254 averages were highest for wetlands and streams, describing highly aromatic DOM molecular moieties. Soils and wetlands dominated the available data for this review, and interestingly, researchers studying temporary flow waters use all three optical property metrics to characterize DOM sources and chemical character, which is not always the case (Figure ).

Using Ecosystem Context to Properly Assess Optical Property Results

At the outset, when using absorbance and fluorescence spectroscopy for DOM characterization, notably there are many variables that shift the spectra and/or quench signal.[25,159−162] Therefore, the relationships used in optical measurements require an environmental normalization step or “ground-truthing” of individual sites and/or ecosystems to reduce misinterpretations.[36] This idea goes beyond basic correction factors of the raw absorbance or fluorescence data and should involve the effect of particles, pH, temperature, metals and other inorganic compounds, C concentration, oxygen availability, biological contributions, and so on. We stress the importance of these steps because interpreting DOM absorbance and fluorescence spectra to help understand the role of that C in ecosystem function cannot make sense without it. For example, reporting that higher plants and soils dominate DOM character from FI, HIX, and SUVA254 values is not logical for ecosystems that do not contain higher plants or soils or large aeolian deposits of those materials. Moreover, use of the soil community and organic chemistry terms “terrestrial-like”, “humic-like”, and “fulvic-like” can result in misinterpretations. Using these terms has generated chemical and ecological boundaries that do not exist in nature and/or inhibit creative thinking and understanding. Therefore, as an optical property researcher, be mindful of the fundamental principles governing the generation of data, the environment from which it was collected, and confirm that the appropriate optical property measurement is being applied to answer individual research questions.

Limitations of optical property measurements and surrogate uses for C concentrations have been discussed and reviewed,[32,36,163] yet challenges with C interpretation from these data sets continue to persist. In this section, we discuss why absorbance has worked so well, when is it better to use absorbance instead of fluorescence and vice versa, and is it wise to think about the selection of using one method over another or using one calculation and metric over another. The combination of using absorbance and fluorescence spectroscopy provides the most information to better understand DOM intrinsic properties, character, and the potential role of C in ecosystem function. However, different chemical perspectives may dictate when to use one application over another. For example, a photochemist may only need absorbance data to answer quantitative and qualitative C research questions, and a fluorochemist may only need fluorescence to answer qualitative C research questions. We submit that selecting one application over another can limit environmental C understanding. For those who benefit from the combination of both, when possible, we propose the term “photofluorochemists” and remind the community that fluorometry cannot exist without absorbance spectroscopy. Even at low C concentrations, where absorbance values are below a blank sample when measured in a 1 cm path length cuvette or sampling device, the term “photofluorochemists” can still apply to those using fluorescence spectroscopy to represent the chromophoric fraction of FDOM; there is no FDOM without CDOM.

Certainly, absorbance and SUVA254 work well as standalone measurements when there is high enough C concentrations or enough sample volumes. However, which technique is commonly measured when one or both variables are limited? The answer is researchers turn to fluorescence, FI, and HIX under those conditions as fluorescence is a more sensitive technique and can measure DOM quality in relatively simple terms for small sample volumes. Fluorescence spectroscopy offers the opportunity to do a great deal with low sample volumes. The issue with optically dilute samples in a 1 cm path length cuvette can usually be rectified with more sample volume when using absorbance spectroscopy; however, fluorescence spectroscopy may generate a wealth of data for the same samples with less volume (5 mL or less). For example, extreme environments, high-latitude ecosystems, and logistical constraints for conducting atmospheric, cryospheric, astrobiological, and other biogeochemical research may force DOM characterization research questions to be solely dependent on one optical property data set instead of two.[49,115,118]

What Do DOM Optical Properties Teach Us about Ecosystem Connectivity?

Determining the potential role of C in ecosystem function through results-style language from optical property measurements is only as precise as the context in which it is considered. Wetlands, streams, creeks, and human-manipulated soil have relatively distinct groupings for FI, HIX, and SUVA254, while coastal and natural soils showed more overlap for each index (Figures –4). However, when evaluating optical property metrics for these ecosystems at the biosphere level (i.e., atmospheric, marine, terrestrial), results showed no substantial differentiation (Figure ). This broad scale approach also captures various distributions (e.g., uni-, bi-, and trimodal), summarizing the biosphere subdivision meta-analysis results (Figures –4). DOM isolates such as Suwannee River Fulvic Acid (SRFA) and Pony Lake Fulvic Acid (PLFA) are often used as ecosystem endmembers of terrestrial and microbial material, inferring that all other results may fall within these boundaries. SRFA and PLFA DOM showed large differentiation for FI and HIX but were more similar for SUVA254. The idea of end members or “bookends” using these two reference standards only seems applicable when using FI. It also creates a linear perspective of DOM types and stifles a more accurate representation of DOM complexity. The HIX distributions were below PLFA and SRFA values (Figure b), and SUVA254 distributions extended well beyond (Figure c), insinuating a need to expand our idea of what constitutes an endmember and what “terrestrial” and “microbial” language means when using optical properties. These findings pose questions for optical property researchers to consider. Are new endmembers needed? Should endmembers like SRFA and PLFA be used at all?

Figure 5

Box plots and distributions of (a) fluorescence index (FI), (b) humification index (HIX), and (c) specific UV absorbance at λ = 254 nm (SUVA254) for dissolved organic matter (DOM) in each biosphere including atmospheric, marine, and terrestrial. The conceptual diagram in the upper right-hand corner depicts the environmental biogeochemical gradient and subsequent changes to FI, HIX, and SUVA254. DOM shifts to a lower FI, more humified (high HIX), and higher SUVA254 as it becomes more processed by various environmental aspects along the gradient. Typical endmember DOM isolates, Suwannee River I and Pony Lake Fulvic Acids (SRFA and PLFA), overlay each distribution plot.[9,23,159,164−167]

Pairing of FI and HIX values further demonstrated optical property overlap that we attribute to the interconnectedness of the biospheres (Figure ). At the biosphere scale, the atmospheric data clustered tightly within the marine and terrestrial data, describing potential sources of atmospheric C (Figure a). Within each biosphere, nearly all subdivisions shared similar characteristics at low HIX values across diverse FI (Figure b–d). Meta-analysis results showed that DOM chromophores and fluorophores from different ecosystems are more similar than previously thought. Substantial overlap in optical property metrics renders the results-style language, such as marine-like and terrestrial-like, meaningless. The overlap seen in optical property metrics also implies considerable interconnectedness among ecosystems and biospheres. Rather than considering these ecosystems independent of one another, researchers should consider ecosystems as a continuum.

Figure 6

Humification index plotted with fluorescence index for (a) each dissolved organic matter (DOM) biosphere and corresponding biosphere subdivisions: (b) atmospheric, (c) marine, and (d) terrestrial. Each polygon represents an area in which peer-reviewed, coupled FI and HIX values span. Biosphere categories and subcategories are referenced from Figure , and data is organized in Table S1.

Physical, chemical, and biological environmental variables and processes must be integrated into optical property result interpretation when evaluating DOM and ecosystem understanding; looking at a metric alone is not enough to capture the “full picture” of the DOM in an ecosystem. For example, similar FI values are found in some sediments as found in aerosols, clouds, dust, and fog. When considering these two groups of ecosystems, they both have a considerable amount of microbial and terrestrial influences. Their similar FI values should reflect similarities in the materials present in each ecosystem. However, we must not simply assume that groups of ecosystems have the same makeup by simply looking at a single metric. Rather, we must pair the data with context about what variables are present in the respective ecosystems and consider the source of the DOM being observed and how it potentially cycles through its ecosystem. Doing so creates a much clearer picture of the nature of the DOM across ecosystems and biospheres than looking at a number alone. Researchers will need to ask the following questions: do these optical property metrics answer the posed research question and do these results make sense in the context of this ecosystem and corresponding ecosystem cycles and processes.

Optical properties remain valuable tools for DOM characterization, but optical properties alone cannot be used to effectively inform ecosystem function. While these results demonstrate challenges in the field, they present an opportunity for researchers to innovate novel analytical techniques and metrics that can be used in conjunction with optical properties. Additional optical results from more diverse settings may aid in the differentiation among ecosystems and provide a better idea of how far the optical continuum extends. This approach would lead to a better understanding of end members and boundaries of optical property metrics used for DOM source classification assays. The C cycling and DOM optical property community would greatly benefit from a shift toward the vastly understudied subsect of DOM optical research and ecosystem connectivity and away from a binary/discrete perception of ecosystems. As Earth’s biospheres are all connected, so must be the optical properties of DOM.