Historical ecology of forest garden management in Laxyuubm Ts’msyen and beyond

ABSTRACT Cultural land-use is an important driver of ecosystem change, influencing the composition of species across landscapes and through time. Recent research in northwestern North America has shown that historical Indigenous land-use and forest management has resulted in relict forest gardens dominated by edible fruit, nut, and berry producing trees and shrubs – many of which continue to grow adjacent to archaeological village sites today. Our research combines archaeological and ecological methods to better understand the historical ecology of seven forest gardens in the Pacific Northwest. Vascular plant communities at all sites are evaluated for distinctiveness using ANOSIM, NMDS, and indicator species analyses. We identify 15 forest garden indicator species, all of which are culturally significant edible fruit or root-bearing species. We then present the results of an intensive historical-ecological study of one site in Laxyuubm Gitselasu (Ts’msyen). Paleoethnobotanical data, soil and tree ring analyses, and radiocarbon dates show that forest management in the Gitsaex forest garden of Gitselasu pre-dates settler colonialism and shows that people likely modified soils and otherwise enhanced their immediate environment to increase the productivity of desired plant species. Given the importance of Indigenous peoples’ role in sustaining forested foodsheds, there is an ongoing and urgent need to support their revitalization and management and better understand the integrated cultural practices and ecological processes that result in these vast cultural landscapes.


Introduction
Land-use scientists increasingly recognize that ecological and anthropogenic forces have long interacted in complex ways, forming many of the landscapes we observe today (Hammett 1992;Munoz et al. 2014;Pavlik et al. 2021), including some of the most desirable ecosystems for human use and occupation (McCune et al. 2013;Coughlan et al. 2018).Under certain circumstances, varied Indigenous land-use and management of forests have resulted in the expansion of healthy (e.g.biologically diverse) ecosystems (Levis et al. 2017;Odonne et al. 2019).In fact, such cases demonstrate that the perceived 'wild' or 'untouched' forests in many parts of the world are direct reflections of past human activity (Fairhead and Leach 1995;Casas et al. 2007;Ross 2011;Warren and Crawford 2016) and that many biodiverse systems require continued and active management to persist.For example, globally, over one-third of biologically diverse and intact forests with the highest conservation values have been formed and maintained by Indigenous peoples, in some cases for millennia (Ford and Nigh 2009;Garnett et al. 2018;Fa et al. 2020).Given the importance of Indigenous peoples' role in sustaining forested landscapes and the realities of colonial legacies, ever-encroaching industrial development, and climate change impacts on terrestrial ecosystems, there is an urgent and ongoing need to understand these integrated cultural practices and ecological processes (Grenz 2020;Fletcher et al. 2021; M'sɨtno 'Kmaq et al. 2021;Sultana 2022).
Globally, historical and ancient forest management has been well-documented using various archaeological, ethnographic, and paleoecological methods -this is especially true in Neotropical and Afrotropical regions (Alcorn 1984;Fairhead and Leach 1995;Ford and Nigh 2009;Ekblom et al. 2011;Neves and Heckenberger 2019).However, we know less about the long-term development, effects, and relationships that have formed between people and forests in temperate regions, for example, in the Pacific Northwest of North America (for the Northeast see Black et al. 2006;Abrams and Nowacki 2008; see also MacDougall 2003).Managed landscapes near archaeological village sites on the Pacific Northwest Coast, such as clam gardens, intertidal marsh gardens, and Wapato (Sagittaria latifolia Willd.)gardens have been the highlight of recent ethnoecological and archaeological literature (Deur and Turner 2005;Lepofsky et al. 2015;Deur et al. 2015;Hoffmann et al. 2016;Toniello et al. 2019).Much less attention has been given to forested ecosystems in the region, where people lived in relatively dense populations for thousands of years and are known to maintain and manage a variety of wild plant species (Turner 2014).
Here, we present historical-ecological research on Indigenous forest gardens -forests managed by Indigenous peoples in the past that continue to grow adjacent to archaeological villages in British Columbia (BC, Canada).Untold centuries, perhaps millennia, of Indigenous forest management has resulted in broadleaf edible fruit/nut trees and shrubs which contribute to heterogeneous landscapes in otherwise conifer-dominant woodlands of the Pacific Northwest.Forest gardens in the Pacific Northwest have only recently been recognized and characterized as significant landscapes in scientific literature but have long been recognized in Indigenous communities (McDonald 2005;Armstrong et al. 2021Armstrong et al. , 2022;;Vanier 2022).To date, our surveys show that they grow along the coastline or in river valleys within a few hundred kilometres of the coast and are located exclusively in association with archaeological village sites.Previous research highlights how, compared to naturally occurring conifer forests, forest gardens are more biologically diverse, more functionally diverse (contributing to overall ecosystem health and resilience), and have a higher proportion of animal dispersed/pollinated species, therefore providing important habitat and resources for wildlife (Armstrong et al. 2021).While we know about the species composition of some forest gardens, we know little about their size, age, structure, variability, and their legacies and occurrence throughout the region.The lack of data is unsurprising given that Indigenous land-use practices have long been overlooked or de-emphasized in archaeology, plant community ecology, and climate change research (Goring et al. 2009;Lepofsky and Armstrong 2018).
The reason the study of forest gardens has been downplayed in research and management contexts is threefold.First, archaeologists and historical ecologists are hampered by the scant material evidence left behind by ancestral gardeners.Indigenous management practices (e.g.burning, transplanting, clearing, etc.) often leave subtle signatures which tend to mimic many natural biophysical processes and can be challenging to document (Lepofsky and Lertzman 2008;Lepofsky and Armstrong 2018).Second, the industrialization of logging and other resource extraction practices has transformed countless ecosystems in Northwestern Canada, erasing many Indigenous peoples' living archives embedded in forested landscapes.For example, in our study region, logging, mining, and linear developments (highways, railways, pipelines) have led to extensive and compounding environmental impacts (slope failures, soil erosion, extreme hydrological discharges, and invasive species), often resulting in drastic changes to the geophysical and vegetal landscape (Gottesfeld and Rabnett 2008).A final challenge to overcome is that settler-colonial narratives about Indigenous peoples' land-use tend to either erase or minimize the legacies of people within their lived environments.For example, in global land-use literature, some scholars hold that Indigenous peoples had little to no effect on their lived landscapes (Oswald et al. 2020) or that they generally exhausted and degraded it (Krech 1999).Increasingly, and with thanks to Indigenous teachers and researchers (e.g.Geniusz 2009; Grenz 2020; see also Turner and Neis 2020), these previously attributed (and racist) tropes have been thoroughly debunked.Indigenous land-use is unquestionably pervasive, complex, and varied, and grounded in intricate frameworks of localized management, governance, and ideology -all of which change, rearrange, and are reinscribed through space and time (Anderson 1996(Anderson , 2014;;Menzies 2006).
Forest gardens have tremendous heritage, environmental, and relational value.Their recognition across British Columbia is timely, as industrial land-use, particularly logging and oil and gas development, has erased important traces of Indigenous land-use and as such, they have not been adequately documented and protected.Other concerns, including conifer encroachment, fire suppression, and conservation parks, which do not allow for Indigenous-led management, have also motivated communities in Ts'msyen, Gitxsan, and Coast Salish territories to study and promote forest garden management.To help achieve these goals of greater documentation and protection, the present project offers succinct and replicable methods for studying the history and ecology of Indigenous forest gardening and management.With this aim, a historical-ecological approach and assessment of seven forest gardens and adjacent archaeological sites was undertaken in northwestern and southwestern BC between 2016-2021.More intensive historical-ecological fieldwork and analyses were undertaken at one forest garden in Ts'msyen territory to determine baseline ecosystem metrics (including spatial extent, soil properties, species composition and structure) and ascertain archaeological contexts (through mapping and paleoethnobotanical analyses).The new appreciation for, and re-discovery of, managed forest gardens means that we are only beginning to tackle experimental methodologies and the logistics of 'how to study' these unique culturalecological systems in relevant and respectful ways.

Study sites and field surveys
In total, seven archaeological village sites (Figure 1) with potential forest gardens were selected for survey based on: (1) the prior knowledge of colleagues who learned about plant stewardship legacies at particular sites from their teachers, extended families, and other community members; (2) observable archaeological features (e.g.cultural depressions, village sites); (3) the recent occupation of the village site within the last several hundred years; and (4) relatively low levels of local forest disturbance which could otherwise obscure signatures of pre-colonial management.
To identify forest gardens and their edges (i.e.size and extent), we first reviewed historical air photos and satellite imagery.Forest gardens are composed of broadleaf species that are distinct from naturally occurring conifer forests in the region and since forest gardens are sometimes visible in aerial imagery, Bing Maps and Google Earth can be useful for distinguishing potential canopy boundaries.Pedestrian and mapping surveys were then carried out using GPS tracks, walking the potential perimeter of each forest garden to distinguish them from surrounding conifer forests (hereafter, periphery forest).Archaeological features, such as house platforms and cultural depressions within or adjacent to forest gardens, were mapped.In some cases, previous archaeological research was collated and used for spatial reference (McMurdo 1975;Coupland 1988;Ritchie and Lepofsky 2020).
At each of the seven archaeological village sites, we conducted botanical inventories within the forest gardens and adjacent periphery forests (periphery forest controls were used to characterize the ambient environment).Plants were inventoried in 5 × 5 m plots, selected by composite (grab) sampling across the site.The number of plots per site was based on a stratified (proportional to site size) sampling strategy resulting in an average of seven plots per site, for a total of 79 plots across all sites.Tree, shrub, and herbaceous species were recorded in each plot.Tree species were counted and shrub and herbaceous species were tallied based on a percent coverage of the plot.Abundance data for each plot were converted to presence-absence data for analysis.Presence-absence data have the potential to better represent some plant communities than abundance data, especially when organisms included in analyses have dramatically different life strategies such as forbs and trees, which may otherwise be heavily biased toward patterns of forb diversity due to their tendency to occur in relatively greater numbers (Hirst and Jackson 2007;Bastow and Podani 2012).
We expect that tree, shrub, and herbaceous species composition will differ significantly between forest gardens and periphery forests, but that forest garden species composition across all sites will be significantly similar.However, given that our study area covers over five degrees of latitude and encompasses some 1000 square kilometres, we anticipate that latitudinal variations and general floristic trends will be apparent in both forest garden and periphery forest vegetation inventories (Meidinger and Pojar 1991).
To test whether the co-occurrence of species across all plots was non-random in terms of plot location, we used the multipart function in the R (R Core Team 2013) package indicspecies ver.1.7.9 (De Caceres et al. 2016) with 9999 permutations.To visualize similarities and differences in the vegetation present in putative forest gardens and peripheral forests, we carried out a nonmetric multidimensional scaling (NMDS) analysis using the R package vegan ver.2.5-7 (Oksanen et al. 2020) with the Jaccard dissimilarity index.The results were illustrated using the R package ggplot2 ver.3.3.3(Wickham 2011).Finally, to test for statistical differences across all forest garden sites (and their corresponding control plots) we conducted an analysis of similarities (ANOSIM) with the same version of the R package vegan, the Jaccard dissimilarity index, and 9999 permutations.

Gitsaex forest garden (Laxyuubm Gitselasu, Ts'msyen)
To better characterize the history and ecology of forest gardens, an in-depth investigation was undertaken at the Gitselasu village of Gitsaex between 2017-2021.The Gitsaex village is one of five archaeological village sites in the Kitselas Canyon on the Skeena River in northwestern BC.The site itself was believed to be relatively recently occupied (ca.250 years BP), but the Canyon spans over 6000 years of Gitselasu occupation (Inglis and MacDonald 1979;Coupland 1988).Our research at Gitsaex was solicited on behalf of the Kitselas government; many families continue to live and make their livelihoods around the Canyon today and there is a desire to better understand and restore these edible forests.Gitsaex is the largest archaeological village in the Canyon and consists of 17 (~10 m × 12 m) house platforms arranged in two rows on a terraformed (human modified) terrace above the Skeena River.Ethnographic and ethnohistoric research compiled by one of the authors, in collaboration with Kitselas Lands and Resources Department, was evaluated for specific mentions of Gitselasu use and management of forest garden species (Armstrong 2022).
The Gitsaex landscape is characterized by (1) a terrace of clustered house platforms with extensive midden deposits whose edges are being eroded against the Skeena River, which transitions to, (2) a second, higher terrace with scattered cultural depressions, radiating out towards (3) a periphery conifer forest (Figure 2).The boundaries between these sites are indistinct and blend into one another along ecotonal gradients.Over time, forest garden species have likely encroached onto the village house platforms, and coniferous trees are evidently encroaching into the forest garden community.
We expect that conifer intrusion into forest gardens is relatively recent.To test this, tree cores were taken to create a relative chronology for conifer establishment.Tree cores were sampled at breast height with an increment borer, or in the case of Pacific crabapple, cookie cross-sections were taken 10-20 cm from ground level.Pith-rot is common in Pacific crabapples and their trunks are often recumbent, making good core samples difficult to obtain.Cores and cookies were dried, mounted, bisected, sanded, and polished, and rings were counted to date the trees and establish a relative chronology of establishment.
Excavation units were dug to characterize the subsurface soil structure of the Gitsaex landscape.Excavation units (0.5 × 0.5 m) were dug within the village (n = 3), forest garden (n = 5) and conifer forest periphery (n = 3) for a total of 11 units.Soil profiles were measured, drawn, photographed, and described.Cultural materials consisting mainly of fire-altered rock and chipped or ground stone artifacts were counted and photographed.We expect soil formation processes to differ between the immediate village, the human modified forests (forest gardens), and naturally occurring forests (periphery forests).
Paleoethnobotanical specimens were collected from excavation units representing four distinct contexts: (1) a cultural depression within the village (between two house platforms), (2) two cultural depressions within the forest garden, (3) a forest garden sample with no cultural feature, and (4) a conifer forest periphery sample with no cultural feature.We anticipate that cultural features in the forest garden and near house platforms will exhibit plant macrofossils reflecting past human plantuse, especially those species used for preserving and cooking.Specifically, we hypothesize that plant remains in cultural features will include at least a subset of forest garden indicator species, and that these specimens will represent a relatively significant time-depth (i.e.specimens will reflect pre-colonial plant-use).
Bulk sediment samples for extracting plant remains ranged in volume from 0.35-2.0litres.To sort plant remains and other materials, we used a standard flotation method (Pearsall 2016); sediment was measured then mixed with water in a bucket and light fraction plant remains floated to the surface.Heavier fractions (pebbles and other materials) sank to the bottom of the bucket.The light fraction was poured into 1.0 mm and 0.425 mm screens and the heavy fraction into the 1.0 mm screen.The sediments from Gitsaex (except the conifer forest periphery sample) were lumped with silts and clays and required de-flocculation.To gently disperse the sediments and release plant remains, samples were soaked in a 10% solution of baking soda for periods of time ranging from 20 minutes to 48 hours prior to flotation.Two samples had to be partially water-screened due to the high organic matter.Light fractions consisted of modern and charred botanical remains and some microfauna.Heavy fractions were sparse and consisted mostly of pebbles.
Standard paleoethnobotanical techniques were used in the sorting and identification of all plant remains (Pearsall 2016).Samples were sorted into their constituent parts under a dissecting microscope (7-30 × resolution), identified using a comparative collection and both published and online sources (Cappers et al. 2006;Klinkenberg 2020; USDA 2022).Plant remains were then quantified.Since uncharred plant parts do not preserve well in the low pH and active microbial soils characteristic of the region, we assumed that uncharred remains were recently deposited and therefore only charred remains were included in our analysis.Three paleoethnobotanical specimens (chosen because they were forest garden indicator species) were sent to the André E. Lalonde AMS Radiocarbon Laboratory (Ottawa, Canada) for radiocarbon dating.While we do not expect that these radiocarbon dates will directly date forest garden sites per se, we do expect the dates to demonstrate the presence of forest garden species prior to the onset of colonization when forest gardening and local fruit production was common.

Forest garden community composition
A total of 155 species were recorded across all seven sites.In comparison with periphery forests, forest gardens have a much more diverse assembly of tree, shrub, and herbaceous layers.Vegetation surveys confirmed our preliminary observations, that forest garden species consisted predominantly of broadleaf edible fruit/nut trees and shrubs, with an understory of herbaceous species including a suite of ground level root foods.Fifteen species are statistically significant (p < 0.05) indicators of forest garden communities, all of which are (or once were) important edible and medicinal nut, root, and fleshy-fruited species (Table 1).As expected, conifer forests surrounding forest gardens consisted of old growth or second growth conifers with little to no understory or shrub layer.
The NMDS and ANOSIM analyses indicated clear differentiation between forest garden and periphery forest botanical plots (Figure 3).The NMDS analysis converged with a stress value of 0.146 in two dimensions.With the exception of a single plot at the Tsleil-Waututh village of Shxwpópélem and one plot in the Gitsm'geelm village of Dałk Gyilakyaw, there is no overlap of the plot types (forest garden vs. periphery forest plots) on the NMDS1 axis.The analysis of similarities (ANOSIM) tests supported the distinction between forest garden plots and conifer forest periphery plots across all sites (p < 0.0001).Our comparison of forest garden and periphery forest plots on a site-by-site basis provides strong evidence for the   difference between the two types of forest [Gitselasu 1; Gitsaex (p = 0.0076), Gitselasu 2; Wiln'aba (p = 0.0019), Tsleil-Waututh (p = 0.0152), Gitsm'geelm (p = 0.0028), Sts'ailes (p = 0.0038), Luutkudziiwus (p = 0.0042), and Gitlut'saaw (p = 0.0022); Appendix S1].The NMDS and ANOSIM analyses also suggested some differentiation between northern and southern forest garden sites.The NMDS analysis coarsely separated sites latitudinally, with the southern sites (Sts'ailes and Tsleil-Waututh) toward the bottom of NMDS2 and the northern sites (Gitselasu, Gitsgm'geelm, Luutkudziiwus) generally toward the top of NMDS2.The northern sites also tended to have greater distinction between forest garden and periphery forest plots than the southern sites, with all R values over 0.60 and reaching 1.0 at Gitsaex, and the two southern sites having values that were lower, though they were still highly significant (0.48 and 0.49).

Relative dendrochronology
While dendrochronological analyses can provide absolute dates (in calendar years), they also provide comparative dates that can be useful in revealing recent landscape history, such as the relative timing of the establishment of different tree communities.At Gitsaex, 5 Pacific crabapple trees (3 cookies and 2 cores) and 5 beaked hazelnut shoots (2 cookies and 3 cores) were sampled within the forest garden.The largest conifer trees in or at the edge of the forest gardens were cored (8 hemlock and 5 balsam fir).A total of 31 tree cores and cookies were sampled; 3 cores did not reach the pith and were omitted, making a total of 29 samples, representing 22 individual trees (Figure 4).
Dendrochronological analysis demonstrated that conifer trees living around the forest garden edge were established beginning 81 years ago, whereas the oldest living Pacific crabapple tree in the forest garden was 108 years old.This indicates that some forest garden species pre-date conifer tree establishment at the intersection of both ecosystems and that those conifer trees are likely encroaching on the forest garden ecosystem.This interpretation is supported by observations of conifer saplings growing on the edge of the forest gardens at all sites.Our dendrochronological data cannot confirm the long-term presence of hazelnuts in forest gardens as their average age is only 46 years old.However, this relatively recent age is likely attributed to the life history of hazelnuts.In particular, beaked hazelnut reproduces clonally from a single long-living root crown but shoots typically do not live past 50-60 years (Kurmis and Sucoff 1989).Thus, while the genotype or genetic individual may be older, above-ground stems that were available for dendrochronological sampling were relatively young.Paleoethnobotanical evidence (described below) confirms a significantly longer time depth for hazelnut at Gitsaex.

Soil and paleoethnobotanical analyses
We expected that the soil structure in excavation units would reflect different types of land-use and site history.In the peripheral conifer forest, all three excavation units exhibited intact forest floor structure, with a predominantly matted organic horizon and sharp transition to mineral soil (Figure 5).In contrast, the three units excavated directly in the Gitsaex village (within ~20 m of house platforms and in one cultural depression) reflected an intensively occupied landscape.These sediments consisted of thick (~80 cm) midden deposits, composed of shell free, charcoal-rich matrices, with cultural material including fire-altered rock, lithic tools, and flakes.Forest garden species, including beaked hazelnut, red elderberry, and highbush cranberry currently grow directly on the house platform surfaces upon which plank houses once stood.Their presence on the former house sites indicates relatively recent succession, where pioneer forest garden species were established after people left the village within the last 110 years or more.
Five excavation units located within the forest garden exhibited neither natural edaphic structure nor the cultural midden deposits that were found adjacent to house platforms.All forest garden units were located on a level terrace above the Skeena River, with similar elevation and aspect to the conifer forests; no units were in areas exposed to periodic flooding.Apparent in all units was a mixing of organic material with mineral soil -characteristic of a mull humus form typical of mixed deciduous forest stands.However, the degree of Ah formation was largely out of character for local humus forms in the absence of humans (or European earthworms) (Figure 5).The thickness and darkness of the Ah layers, and evidence of mixing, indicates that people likely modified these soils.Periphery forest soils characteristic of the region are deficient in the types of organic content required for many fruit and nut species to thrive.In some units, relatively thin (2-5 cm) lenses of midden-like material, characteristic of a paleosol or other anthropogenic influence, were apparent.In two excavation units, ground stone tools were uncovered, indicating that food processing may have occurred within the forest garden landscape.Taken together, anomalies in soil formation processes within the forest garden are clearly distinct from both large midden deposits in the immediate village and soil formation processes in naturally occurring forests.
The recovery of plant remains from paleoethnobotanical samples confirmed that people were processing, cooking, and preserving plant species associated with forest gardens.A total of 19 plant taxa from 10 botanical families, encompassing 4 coniferous and 15 deciduous taxa, were identified in 6.65 litres of sediment from all excavation units.From this inventory, seven forest garden indicator species from the vegetation inventories were identified in the paleoethnobotanical record of Gitsaex (Table 1).Based on our experience with paleoethnobotanical samples collected elsewhere in the region (Lepofsky and Lyons 2013), the present samples from both the village and the forest garden cultural depressions were, for their size, very rich in paleoethnobotanical remains, including both high-value plant foods (beaked hazelnut, Pacific crabapple, red elderberry, and various species of Rubus) and weedy species deposited by seed rain.Few plant remains (only some charred conifer needles) were recovered from both the nonfeature excavation units in the forest garden and the periphery forest.The lack of plant remains outside of cultural contexts supports our expectations, given that uncharred plant macrofossils do not typically preserve in conifer-dominant forests of the Pacific Northwest (Lepofsky and Lyons 2013).
We expected paleoethnobotanical remains to reflect a significant time-depth (at least precolonial) at the site, corresponding with the age of the village, or even pre-dating it.Radiocarbon dating of three plant specimens from the house contexts confirmed our predictions -a charred Pacific crabapple seed and a charred hazelnut shell fragment were dated to 407 ± 26 cal.BP and 480 ± 25 cal.BP, respectively.A fragment of ancient fruit leather (a common form of fruit preservation in the region) from the village cultural feature was dated to 292 ± 28 cal.BP.While these results do not indicate that either species was necessarily growing at the site 300-500 years ago, since seeds and shells could have been transported to the site from distant locations, it does indicate both species were being used well before the onset of settler invasions in the early 19th century.Furthermore, these dates suggest an earlier occupation of Gitsaex, which was previously thought to have been around 250 years ago (Berthiaume 1999).

Discussion
Contemporary forest structure and composition can reflect peoples' ancient cultivation and management strategies even centuries after those practices have ceased.Historical-ecological evidence of such legacies were identified at all seven sites in Coast Salish, Ts'msyen, and Gitxsan territories.While some anthropogenic ecosystems appear 'wild' or ephemeral, historical-ecological methods can be effectively used to identify anthropogenic ecosystems such as forest gardens in the Pacific Northwest.

Ecological patterns and paleoethnobotany
Botanical analyses presented here strengthen previous findings (e.g.Armstrong et al. 2021Armstrong et al. , 2022;;Vanier 2022), showing that forest garden ecosystems are structurally distinct from ambient or dominant ecosystem types for the region (e.g.coastal and interior cedar-hemlock forests).The current study builds on earlier research to include more study sites and introduces NMDS and ANOSIM analyses, both of which strongly support the unique composition of forest gardens as distinct from surrounding forest controls.Although some species represent statistically significant forest garden indicators (hazelnut, Pacific crabapple), more generally, the combined forest garden data shown on the NMDS analysis vary along a latitudinal gradient, suggesting that the composition of forest garden species can be variable, and appears to be influenced by local vegetation patterns, geophysical settings, and/or cultural preferences.This latter point is important for researchers seeking to include forest gardens in their field surveys since universal 'species lists' alone will likely not result in conclusive identifications.A more holistic and concerted approach that considers archaeological and ethnographic data, and engagement with (and consent from) communities whose ancestors created and maintained these spaces, is vital.Due to ecological and cultural connectivity in the Pacific Northwest, we expect that future research will reveal the presence of forest gardens in more regions throughout the province of BC and north and south into the United States (Alaska, Washington, Oregon, and northern California).
Data from the Gitsaex village in Laxyuubm Gitselasu show that the landscape is composed of dynamic forest ecosystems, with patchworks of fruit/nut/berry productivity, and mosaics of localized disturbances and regenerating vegetation (forest gardens, conifer forests, conifer-broadleaf ecotones).The encroachment of conifers into the forest gardens likely reflects that Gitsaex forest gardens have not been tended, cleared, and/or burned in recent generations -all of which are important management practices that help to maintain these ecosystemsand that secondary conifer succession is occurring under highly localized conditions.Remarkably, forest gardens persist after over a century of management and research has indicated that these complex systems may be able to withstand a certain level of conifer encroachment (Armstrong et al. 2021).
A major challenge for historical ecologists is the parsing out of anthropogenic and biophysical factors that might influence forest species composition and vigour.Here, we used standard historical-ecological methods, including botanical inventories and soil analyses (see also Lepofsky et al. 2003;Isendahl and Stump 2019;Odonne et al. 2019) to distinguish human signatures from naturally occurring forest formation processes in Gitsaex.These results show a correspondence between plant remains excavated from cultural features within the village site proper (adjacent to house platforms) with forest garden species composition, indicating that people were intensively using and processing plant foods that were growing nearby.While paleoethnobotanical analyses are useful for understanding a wide range of plantuses in cultural contexts, they are less helpful in reconstructing ancient environments due to the poor preservation of uncharred plant remains outside of cultural feature contexts (conifer forests or forest garden).Our research confirms this, suggesting that pollen and/or phytolith analysis may be a productive avenue for future research in reconstructing forest garden environmental history.

Anthropogenic soils
Results from Gitsaex indicate that a continuum of heavily altered middens (around house platforms) graded to altered (mixed) soils in the forest gardens, and then to less visible human signatures in periphery forest soils.Certainly, people used conifer species extensively, as indicated by the abundance of culturally modified trees in the Kitselas Canyon landscape, spanning at least 800 years (Armstrong 2022).The lack of large stumps (e.g. from logging) and presence of charcoal in periphery forests may indicate consistent clearing (e.g. for fuel, enhanced viewsheds) over relatively large time spans.However, the intensity of impacts to the vegetation, or intensity of use (as indicated by soil modifications alone), decreases as one moves farther away from the village.
The creation and modification of forest garden soils is significant given that previous studies in the Pacific Northwest noting the concentration of plant foods growing at archaeological sites only focus on plants growing directly on shell-bearing middens (Trant et al. 2016;Fisher et al. 2019).These previous studies provide an opening for the interpretation that succession of edible plant species occurs only after people leave (or are forcibly removed) from their villages.The aggregation of forest garden species outside of the immediate village context (i.e.not directly on middens), has not previously been studied.After people left or were forcibly removed from Gitsaex, the open canopy and nutrient-rich middens in and around the house platforms would have been primed for succession and expansion of some forest garden species.Indeed, at Gitsaex, forest garden species thrive directly on house platforms, which certainly would not have been the case at the height of occupation.What our data demonstrate, however, is that forest gardens thrived outside of the immediate village area, reflecting intensive forest management that influenced plant species and soils across a much larger landscape area than the 'village boundaries' typically ascribed by archaeologists.
The Gitsaex forest garden likely resulted from generations of clearing around the village and encouraging, and likely transplanting, fertilizing, and controlled burning of, edible plant species.Regular harvesting of the forest garden, and the importance of the plant foods for Gitselasu people, is implicated by the paleoethnobotanical remains recovered from cooking pits within the forest garden and settlement area.The relatively large (~40 cm) dark earth deposits in the forest gardens are characteristic of a mull humus form, with evidence of anthropogenic mixing between organic matter and mineral soil.These soils are unlike anything encountered in the ambient, surrounding conifer forest and future research may indicate the extent of anthropogenic activity on soil formation processes.

Policy and management
To date, traditionally managed ecosystems (e.g.orchards, berry patches) are not recognized by archaeological practitioners in British Columbia.Although provincial heritage management laws intend to protect archaeological heritage, in practice they have failed to protect the types of heritage that many Indigenous communities deem important; especially living heritage like forest gardens (Lepofsky et al. 2020).There are underlying issues with heritage management in British Columbia, stemming from the deeply rooted colonial history of discipline, systemic racism, power imbalances in decision-making, and for-profit models that underpin most practices (Schaepe 2018;Lyons et al. 2022).While these are bigger issues not directly addressed here, we argue that expanding the scope (literally the boundaries) of the archaeological landscape may be one avenue for a more just approach to archaeological management.Cultural landscapes do not start and stop at the walls of a longhouse; people are sustained by and thrive in the entirety of the landscape.The close association of forest gardens with archaeological villages, their unique soil signatures, and composition of species are a starting point for researchers and managers attempting to identify these historical and ancient ecosystems and integrate them into their cultural heritage assessments and land-use planning practices.
Indigenous peoples' cultivated landscapes are also not recognized in environmental management contexts.Legislation focusing on rare or endangered plant species does not account for peoples' use and labour in managing plant communities which may be otherwise rare, hard to get to, or are in a legacy state from the cessation of historical and ancient management practices.Forest garden research recognizes the crucial role of Indigenous peoples in forest management and the legacies that these practices can contribute to contemporary forest conservation and land-use decision-making.This research is situated within the context of present-day Gitselasu forest management strategies and community desires to revitalize these culturally relevant foodsheds.Ongoing and future forest garden research will be focused on reclamation and restoration as all partnering communities have underscored the importance of forest gardens in their food sovereignty initiatives.The recognition of forest gardens is one avenue for promoting traditional environmental management practices in the face of industry and colonial governments that tend to limit ongoing Indigenous land management.

Conclusion
This research challenges long held assumptions about Indigenous land-use and resource management in the Pacific Northwest, which depicts Indigenous peoples as principally relying on marine and riverine resources (e.g. for shellfish, salmon, or herring), only intermittently 'gathering' edible plants from the bounty of naturally productive forests.This entrenched perspective has oriented archaeological research for decades, so that we are only now compiling the data required to learn about the diversity, nuance, and complexity of ancient human-forest interactions, and implications relating to land-use, food revitalization efforts, and climate change.This can be a frustrating exercise for many people who have always known the extent to which their landscapes were managed.While this research has implications for communities leading conservation and management efforts which seek to better integrate biodiversity values with traditional food management strategies (e.g.Joseph and Turner 2020;Dick et al. 2022), it also comes with important lessons for settler scholars.
Forest gardens may be a new term in the Pacific Northwest, but it is not a new concept.Since the onset of settler colonialism, people have managed, maintained, and cared for their lands, learning valuable lessons along the way (Wright 1962).These lessons have often been ignored, overlooked, or appropriated by settlers, scholars, and government (Geniusz 2009;Grenz 2020), only recently now being held up and acknowledged.This form of colonial gaslighting is common, and as researchers, consultants, and others increasingly look to Indigenous peoples to save the planet from wicked problems like biodiversity loss and climate change (e.g.Wildcat 2010), it is important to recognize and reflect on the colonial legacies and practices that have exacerbated these problems.

Figure 1 .
Figure 1.Forest gardens identified to date and included in this study (British Columbia, Canada).

Figure 2 .
Figure 2. Profile representation of Gitsaex landscape.Botanical inventories, excavation units, and paleoethnobotanical sampling was conducted within the 3 contexts of the site (periphery forest, forest garden, house platforms).
x a Indicator species (p < 0.05) for all forest garden and conifer forest periphery sites.b Presence/absence of plant macrofossils at Gitsaex.c Village Feature (VF), Forest Garden Feature (FGF), Forest Garden Non-Feature (FGN), Periphery Conifer Forest (PCF).d Ethnobotanical significance of plant foods based on Gitselasu-specific indices(Armstrong 2022).X = abundant use; x = moderate use.

Figure 3 .
Figure 3. Non-metric dimensional scaling (NMDS) of forest garden and periphery (control) plots across seven sites.

Figure 4 .
Figure 4. Age of the 21 individual tree and shrub species dated by dendrochronology within forest gardens and conifer forest periphery.The top of each bar represents the first year of growth.

Figure 5 .
Figure 5. Plan view of Gitsaex village.Note the continuum between (A) totally altered sediments (dark-black midden and firealtered rock beside platforms), to (B) moderately altered sediments in the forest garden, to (C) natural soil formation processes in the conifer periphery forest.

Table 1 .
Forest garden and periphery forest species list: indicator species for all sites, ubiquity of paleoethnobotanical remains at Gitsaex, and ethnobotanical significance at Gitsaex.