Understanding the historical trajectories of ecosystems is of high importance to conservation and restoration science. Furthermore, conservation and restoration need to be considered within their socioecological context. Many ecosystems thought to be pristine from a perspective of Western conservation are being understood to be human modified systems. Understanding historical and ancient modifications of ecosystems are paramount when asking what we are restoring to and restoring for.
Pre-european modification of the Hawaiian landscape has been of substantial interest within the fields of archeology and anthropology as the islands serve as a model system for understanding socioecosystems and their development. Such fields have utilized physical remains to understand the distribution of pre-European modification. Further extrapolation has been conducted by relating types of modification to landscape parameters. The resulting models have been shown to be successful, with high levels of accuracy and validation. However, we postulate that significant modification of forests occurred that are not associated with any physical infrastructure and therefore are largely excluded from our perception of pre-European Hawai'i. We utilize remnant trees to understand the extent of the pre-European modification of forests for arboriculture.
Several prior models have examined the extent of pre-European modification of the landscape. Ladefoged et al (2009) first used soil fertility, topography, and the gravitational feed of water to model the extent of lo'i (flooded, terraced) and mālo'o (intensive rainfed) agriculture.
Building upon this, Kurashima et al (2019) added the cultivation of rejuvenated soils on colluvial slopes.
Finally, Gon et al (2018) utilized a broader suite of archaeological features to estimate the extent of non-agricultural modifications.
Our knowledge of traditional cultivation extents suggested that many areas were significantly underrepresented in terms of arboricultural developments. For instance, in both Ladefoged et al (2009) and Kurashima et al (2019) the entirety of Hāmākua and Puna are devoid of agriculture, despite ethnographic sources making it clear that vast arboricultural systems existed in these areas.
Initial analysis focused on the Hāmākua region, where we know extensive and diverse arboricultural forms existed from ethnographic sources and documentation (such as the Land Commission Awards from the original privatization of land in the mid-19th century).
Two species of interest were mapped - kukui (Aleurites moluccanus) and breadfruit (Artocarpus altilis). Along the "Hāmākua coast" between the Wailuku and Waipi'o rivers, over 26,000 individual kukui trees were hand mapped from remote imagery; 500 random trees were ground truthed. Utilizing drone video imagery 300 m surveys were conducted of every river and rivulet; these surveys were visually analyzed for the location of breadfruit trees. Such trees were considered to be remnant or descendant from the ancient agroforestry systems.
Of immediate interest was that breadfruit occurred extensively in the south but dropped off the landscape in the north (see Figure above). We do not think this is a feature of historical landscape modification, because Polynesian crops can be found to generally persist in the small valleys throughout the region. Using this division defined by the occurrence of breadfruit, we examined the distribution of kukui on the landscape in terms of landscape properties. In particular, we saw that kukui extends to a higher elevation in the northern areas (without breadfruit).
And that important relationships between soil fertility and kukui distribution appear to exist. In the south, a floor to soil fertility appears to exist with no kukui being grown beyond a certain point in soil fertility. Furthermore, the point at which breadfruit drops off the landscape corresponds to a point at which soil fertility shifts dramatically.
From this initial work we made three conclusions informed by our ethnographic knowledge of the landscape and previous work done on Hawaiian agriculture: (1) that there is a minimum soil fertility that supported the arboricultural modifications of pre-European Hawai'i, (2) that the more fertile northern region largely created a novel, kukui forest that powered a unique shifting cultivation system known as the pākukui, and (3) that in the less fertile, southern region a more permanent, diverse form of arboriculture was developed.
Following the pilot mapping project in Hamakua we aimed to scale the concepts learned to the entire state. Utilizing the environmental parameters that described the distribution of agroforestry in Hāmākua (rainfall, temperature, and soil fertility), a habitat model for traditional arboriculture was developed.
Remote sensing of kukui was conducted using a maximum likelihood distribution regression using World View 3 data, followed by manual cleaning on the data. The result was a statewide map of kukui trees. Maui is provided here as an example.
The extent of novel kukui groves in the state was used as a validation metric for traditional arboriculture.
While refinement of the final model is ongoing, the current distribution patterns of modeled agroforestry extent and kukui distribution aligns fairly well. Areas with high levels of kukui but not modeled to be agroforestry were primarily colluvial slope agriculture (as modeled by Kurashima et al., 2019). For areas that were modeled agroforestry but do not contain kukui are primarily non-dissected landscapes, where remnant kukui may not have been able to persist due to the widespread use of heavy machinery.
Building upon the previous models, models of seasonal planted areas based on a work done in a separate partnership (Kagawa-Viviani et al., 2018), and utilizing results presented here we created a revised look at pre-contact agriculture in Hawai'i.
This model, we argue, is much more complete than previous models, indicated most all the areas cultivated by Native Hawaiian agriculturalists. In total, our model predicts that approximately 15% of the islands' landmass was converted over to management for agricultural resources, compared to previous estimates that only include about 3% of the total land area (Kurashima et al., 2019). This is a significant shift in the view of ancient Hawai'i and the impact to native ecosystems. While we maintain that Hawaiians were excellent stewards, it is clear that considerable amount of ecosystem engineering was conducted by Hawaiians prior to European settlement.
The Indigenous Cropping Systems Laboratory is run by Dr. Noa Lincoln at the University of Hawai'i at Mānoa (http://www2.hawaii.edu/~nlincoln/lab.html). The interdisciplinary lab engages in biogeochemical, geographical, and ethnographical research methods to study traditional crops and agricultural systems. Of high interest is (1) how agriculture evolves over time within the biogeochemical context of a landscape, (2) how nutrients were sustained for long-term production, and (3) the role that indigenous agriculture can play in sustainable food systems today. Current projects include Agricultural Evolution of Tau, Samoa; Colluvial Slope Agriculture in Punalu'u, Hawai'i; Nutrient Cycling in Traditional Breadfruit Agroforestry; Breadfruit Diseases and Varietal Resistance; Ancient Moisture Management and Impacts on Soil Nitrogen Dynamics; and Modelling and Mapping Traditional Food Crops.