**4. Discussion**

#### *4.1. Data Challenges and Benefits of Performing an Industrial Ecology Holarchic Analysis*

The results of analyses presented in this study illustrate how a holarchic investigation of material flows could benefit researchers in at least two ways. First, characterizing materials flows for multiple holarchic levels could provide frames of reference for understanding a holon's relative contributions to material flows at higher- and lower-level holons. For example, we found that Hilo and Kona together consumed 61% of all fossil fuels consumed on the Big Island, and the Big Island's fossil fuel consumption accounted for 6.5% of all fossil fuels consumed in the State of Hawaii. It follows that Hilo and Kona consumed approximately 4% of all fossil fuels consumed in the state. While appearing to be simple, these comparisons spanning more than two system levels will be difficult to make with conventional MFAs, which conceptualize material flows at a single system level. A holarchic MFA would allow for analysis similar to examining a contribution tree for environmental impacts in life cycle assessment [61]. Extending this analysis to higher-level holons like Hawaii's global supply system for fossil fuels could also illuminate the dependence between holarchic levels and associated political and economic factors, as in the analysis of petroleum imports in Section 3.5 above. The results from holarchic MFAs could also help identify opportunities for recovery of material outputs from a holon by examining flows for higher-level holons and evaluating whether the systems have any displaceable inputs.

The second benefit of a holarchic analysis is that it may be useful to quantify material flows for systems that are difficult to characterize because of data scarcity or complicated system boundaries. Material data often involve some level of aggregation or assumptions. There is also the additional

complexity introduced when dealing with system boundaries that span numerous administrative units. These issues may be especially important for systems that are understudied, such as rural economies. As an example of how the holarchy framework could address this issue, our Big Island results illustrated how the material flows for rural areas could be quantified by subtracting the flows for urban areas from those of the higher-level holon. The quantity for biomass extraction in the more rural areas of the Big Island can be derived from material accounts for Hilo, Kona, and the overall Big Island. Data had been derived for the next higher-level holon in our example, and material flows for urban areas were less complicated to quantify than other flows on the Big Island. These forms of analyses could, of course, also be applied to material flows in non-island systems.

Such multi-level analyses of material flows may be di fficult to achieve in practice, however. While a holarchic analysis could help fill some data gaps, the gap being evaluated may be too wide to allow for any approximation. In the case of waste accounts in the U.S., most counties and states do not report material recovery and make comprehensive accounting of recycling almost impossible. More potential di fficulties in conducting holarchic MFAs arise from the classification categories used to compile material accounts. Even if data are available, the same materials may be aggregated and reported under di fferent material categories based on the classification system used by the entity responsible for accounting. Again, such an issue is highly relevant for waste accounts in the U.S. This may not be a problem for materials for which reporting is required and standardized by regulating institutions, however. One potential issue about material categories, not specific to the context of a holarchic analysis, is that the classification categories may simply be inadequate for certain analyses. In our compiled dataset for Hawaii, more than half of the exported biomass is categorized as "other" or "food products not elsewhere classified" under the Eurostat methodology. Such rough aggregations make granular comparisons of flows di fficult. These challenges demonstrated the need for a better, standardized material accounting system.

#### *4.2. The Archipelago Over Time (Temporal Holarchic Analysis)*

Our analysis for holon level 5 illustrated how a holarchic analysis could help understand global material flow relationships with appropriate time-series data. In some cases, holons or holachic systems can be studied over a longer time period, enabled by historic data to start and then by continuing data acquisition and analysis. Figure 9 presents historical exports from the Hawaiian Islands (holon level 4) to the rest of the world compiled mainly from annual reports prepared by the firm Peirce and Brewer and the Honolulu Collector General's O ffice (working paper by Chertow and Paul [26]). The material accounts reflected in the figure indicate the emergence and decline of several industries in the Hawaiian Kingdom.

The material flows for the Hawaiian Islands were mostly self-contained until foreign commercial ships began to arrive after Captain Cook's "discovery" of the islands in 1778. In the following centuries, the islands' socio-economic metabolism went through a series of dramatic transformations, shaped mostly by global economic factors. For example, the whaling trade served as the backbone of Hawaii's economy during the mid-nineteenth century, but gradually declined as the whale populations diminished, the American Civil War reduced the whaling fleet, and whale oil faced competition from kerosene and the growing petroleum industry [62,63]. From the 1850s, investors turned to exporting agricultural products, especially sugar, an activity that was increasingly profitable owing to tari ff arrangements and increasing demand from fast-growing populations in the American West [63,64].

Sugar exports were also captured in our MFA for the Hawaiian Islands. The export quantities, however, hardly resemble what was once the Hawaiian Islands' largest industry, as agriculture became cost-ine ffective and Hawaii transitioned to a service-based economy. More recently, the last sugar plantation in the State of Hawaii, located on Maui, ceased operations in December 2016. Maui's land use, local economy, and material footprint may since have changed significantly. Analyzing material flows for a single system level certainly has its value. However, it is also important to think about the interconnections with higher- and lower-level holons, and ideally, over time, to understand fully which factors shape material flows at each holarchic level.

**Figure 9.** Exports from the Hawaiian Archipelago, 1805–1900. Data from working paper by Chertow and Paul [26].

#### *4.3. Comparing Holons at the Same Holarchic Level*

While this study principally focuses on comparing relationships across higher- and lower-level holons, useful information is also available by comparing the metabolism of holons at the same holarchic level (as in the comparison of Hilo and Kona in Section 3.2 above) in the student work compiled and presented by Kanaoka [55] (Figure 10).

**Figure 10.** (**a**) Direct material input and downstream material flows for five Hawaiian Islands in 2007. Left: total flows for islands, right: flows per capita de facto. (**b**) Direct material input and domestic material consumption by material type for five Hawaiian Islands in 2007. (**c**) Waste generation and reuse per capita de facto for five Hawaiian islands for various years: Oahu (2005), Maui (2006), Kauai (2005), the Island of Hawaii (2007), and Molokai (2006). Results for Lanai are not shown due to data gaps. Abbreviations: DMI: Direct material input; DMC: Domestic material consumption; EXP: Exports; Gen.: Generation. Adapted from Kanaoka [55].

As with the case of fossil fuel consumption in a previous section, Oahu accounted for the majority of the direct material input for the State of Hawaii in 2007, followed by the Big Island, then Maui (Figure 10a). For most islands, non-metallic minerals represented the largest ratio of domestically consumed materials (Figure 10b), most of which were used in construction. The only exception was Oahu, which imported and refined crude oil for use on all of the islands. Of the non-metallic-mineral inputs to each of the islands, 84–96% by mass was extracted on-island rather than imported.

The comparison of islands also showed that material flows per capita for islands had a strong negative correlation with population density (Figure 10a). This trend was in line with findings from Eurostat's MFA of 28 EU countries, which indicated that countries with higher population densities consumed fewer materials per capita [65]. For this reason, it was observed that the two disparately sized islands, the Big Island and Molokai, had similar material flows per capita that were significantly larger than the remaining islands. These results may illustrate the increasing effect of urbanization on resource efficiency from a materials perspective.

The quantities of waste generation and reuse per capita appeared to be comparable across the islands, but their compositions did not (Figure 10c). We found that the contrast in waste composition was mainly because of differences in the waste accounting methods employed by each county in Hawaii. There is no national standard for waste accounting in the U.S., so counties categorize the same materials under different waste categories, even within a single state. Similarly, some of the waste classification categories used by countries were not compatible with the Eurostat methodology. This issue illustrates some of the practical challenges for conducting a holarchic analysis.

#### *4.4. Related Analyses in Industrial Ecology*

The application of the holarchy framework to industrial ecology has been extremely limited, although there appears to be increased interest in recent years. One line of work develops minimization models for the environmental impacts of a product system-of-systems rather than merely individual product systems [66–68]. Another study proposed the holarchy framework for structuring supply chains equipped with smart technologies for improving their environmental sustainability [69]. Additionally, DeLaurentis and Ayyalasomayajula [70] explored how tools from industrial ecology could be integrated with system-of-systems models to better assess the environmental impacts of complex adaptive systems. Other applications of holarchy to material flows have mostly been qualitative, to conceptualize how socio-economic systems relate to the natural system (e.g., [71]) or how socio-economic systems include energy and material systems (e.g., [72]).

#### *4.5. Opportunities for Further Research*

The Hawaiian Islands holarchic MFA has provided a glimpse of the enhanced insight into social metabolism and material flows that the approach can potentially offer. To explore industrial ecology holarchy more generally, it will be necessary to recognize that two basic requirements exist for any material-related holon: a clearly defined spatial boundary and some way of monitoring or estimating flows across that boundary. New York City, much of which is connected by toll bridges, might form another example, with sections of Manhattan Island, the Borough of Manhattan, and New York City itself as candidate holons. More generally, some industrial ecology analyses have used multi-holonic data or data gaps to fill in missing information in systems that were not recognized as holarchic. A clear example is the zinc MFA of Meylan and Reck [73], which quantified holons at country, continent, and (though not explicitly computed) planetary levels. From that perspective, it is easy to imagine an extension of holarchic levels, this time downward, for example to Indonesia and its constituent islands, should appropriate data be available. Other examples doubtless can be devised by imaginative researchers, including those of non-island systems. It should be noted, however, that comparisons across holons without common governance structures, as seen in Hawaii, may be more difficult to achieve. Finally, as analyses of country-level imports and exports have provided much useful information on material flow and use, the extension of MFA into a variety of smaller-scale geographical

entities is likely to o ffer new insight as well. We encourage industrial ecologists to explore and profit from applying the holarchic concept to a variety of multi-holonic material stocks and flows around the world.

#### **Supplementary Materials:** Supplementary materials can be accessed at http://www.mdpi.com/2071-1050/12/8/3104/s1.

**Author Contributions:** All authors contributed equally to this work. M.R.C. and T.E.G. contributed to research design and provided supervision. T.E.G. conceptualized the study. M.R.C. organized and managed the project. Data acquisition, analysis, and interpretation were made by K.S.K. and J.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded over several years by the US National Science Foundation (Grant No. 0948781), the Hawaii Community Foundation, the Japan Foundation, the field trip fund of the Yale School of Forestry and Environmental Studies, and the Center for Industrial Ecology.

**Acknowledgments:** We are grateful to the many students cited in the reference section whose research provided the basis for much of this study. Former PhD student Matthew Eckelman played a formative role on the Oahu study, and former research assistant Frederick Reppun collated a grea<sup>t</sup> deal of data. We also thank the Kohala Center, especially Matt Hamabata and Betsy Cole for our cultural education, and the Hawaii Community Foundation for the original Oahu funding. We appreciate the support of the Yale School of Forestry and Environmental Studies and o ffer a grea<sup>t</sup> "mahalo" to all of our Hawaiian Island hosts.

**Conflicts of Interest:** The authors declare no conflict of interest.
