Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Intertidal resource use over millennia enhances forest productivity


Human occupation is usually associated with degraded landscapes but 13,000 years of repeated occupation by British Columbia’s coastal First Nations has had the opposite effect, enhancing temperate rainforest productivity. This is particularly the case over the last 6,000 years when intensified intertidal shellfish usage resulted in the accumulation of substantial shell middens. We show that soils at habitation sites are higher in calcium and phosphorous. Both of these are limiting factors in coastal temperate rainforests. Western redcedar (Thuja plicata) trees growing on the middens were found to be taller, have higher wood calcium, greater radial growth and exhibit less top die-back. Coastal British Columbia is the first known example of long-term intertidal resource use enhancing forest productivity and we expect this pattern to occur at archaeological sites along coastlines globally.


Humans are usually associated with degraded landscapes and the ‘forest primeval’ is thought to be free of human influence. Removal of habitat, depletion of soil nutrients, forest productivity declines and reduction in biodiversity are commonly cited outcomes of modern human land-use patterns1,2. There is growing evidence, however, of long-term land and resource-use practices such as soil terracing3, fertilization4, forest management5,6 and controlled burning7, which provide alternative models to human presence resulting in environmental degradation. Here we show that millennia of resource use by coastal First Nations in British Columbia, Canada, contributes significant marine-derived nutrients and alters site conditions that enhance temperate rainforest productivity. We focus on western redcedar (Thuja plicata Donn ex D. Don), a species of cultural importance throughout the Pacific Northwest of North America8,9, and a valuable timber tree10.

Coastal First Nations of British Columbia have a long and vibrant history, with oral and archaeological data providing evidence of over 13,000 years of repeated occupation. Long-term records of occupation are found on the Central Coast of British Columbia, as sea level has been relatively static throughout the Holocene11,12. The term ‘habitation site’ refers to a place where people lived in the past resulting in the accumulation of a physical-material record following years of seasonal or multi-seasonal occupation. Most habitation sites are situated near the shoreline, where people lived in proximity to a diversity of terrestrial (for example, root gardens, managed plant resources and mammals) and especially marine (for example, clams and fish) resources13,14,15. Significantly, these resource-use patterns resulted in the importation of shell from the intertidal zone into the terrestrial environment. Typically, the major constituents of habitation site shell middens are shell, rock, bone, charcoal, plant remains (mainly rootlets), artefacts, archaeological features and organic soil.

Disposal and stockpiling of shell, as well as the cultural use of fire, altered the species composition of the forest and understory in and around habitation sites16,17. The legacy of human use is most visibly preserved in the accumulation of substantial shell middens, which in some cases exceed 5 m in depth18 and cover thousands of square meters of forest area as the shells are placed for terracing and drainage or are simply discarded as refuse. Less visible legacies at shell midden sites are the effects of ancient fires in the form of charcoal from repeatedly buried hearths19 and site-level burning before reoccupation20, which increases nutrient availability21 and shifts competition22. Although these sites continue to have important cultural value, many have not been occupied with the same intensity or regularity for at least the last 150–250 years, especially since smallpox epidemics in the nineteenth century23,24. However, the western redcedars that have subsequently grown on these shell middens have been intensively culturally modified for harvesting wood and bark. The vast majority of the study area has not been commercially logged or developed, helping to preserve shell middens, an abundance of culturally modified trees, and many stone fish traps and clam gardens in adjacent intertidal areas25,26.

The long-term human harvesting and deposition of shellfish and other animal remnants from the marine to near-shore environment represents a significant marine-derived nutrient input and modifier of soil pH. Some marine-derived nutrients associated with these sites, such as nitrogen from salmon carcasses moved inland by predators and scavengers such as bears and wolves, have also been demonstrated to be important to community and ecosystem structure27, although these nutrients are expected to deplete relatively quickly28. However, the slow release of calcium from degrading shells (CaCO3) is more persistent29. Calcium deficiencies are common in forested ecosystems and have been proposed as a significant contributor to top die-back in western redcedar30, although see ref. 31, a condition that results in the death of tree crowns and could represent a significant economic loss to forestry. Further, the addition of CaCO3 and charcoal from fire increases soil pH, which in turn increases the availability of macronutrients such as phosphorous32. Shell midden, as a bulk sedimentary matrix, changes the physical structure of the soil by improving drainage33 while the deposition of charcoal increases the porosity of highly weathered soils and also increases phosphorous34. Increased soil pH and drainage contribute to enhanced forest productivity.

Using a combination of airborne lidar and field-based ecological methods, we examine how the consequences of long-term site occupation, such as shell middens and fire, influence measures of forest productivity at our study site on the Central Coast (Fig. 1a,b). Forest productivity, expressed by forest canopy height, forest width, vegetation greenness and forest canopy cover (that is, productive forests extending further inland) and vegetation greenness at habitation sites was compared with forests along the entire coast contained within our study area using lidar extracted in 30 × 50 m cells set in from the coastal forest boundary by 10 m (Fig. 1c).

Figure 1: Study area map.

The study area (a) includes northwest Calvert Island and southwest Hecate Island on the Central Coast of British Columbia, Canada. (b) Map of the study area with all habitation sites (red) and control sites (blue). Fire surveys (orange-filled red circles, n=7) and detailed ecological analyses (white-filled red circles, n=6) were carried out on a subset of habitation sites. (c) Close-up map of habitation site (red circle) with approximate shell midden delineated by transparent red polygon. The blue circle shows the control site. Cells (50 × 30 m) used for extracting lidar data are shown in yellow. Data for base maps from used with permission under


Forest productivity is highest near and on human habitation sites

To explain and understand differences in forest productivity, we obtained eco-cultural parameters (shell midden depth, prevalence of fire, soil nutrients, wood nutrients and individual tree growth metrics) using a paired design with control sites located along the coast in areas with similar forest composition and high productivity, but lacking in shell middens (that is, these sites are assumed not to be habitation sites and archaeological surveys confirm this).

Measures of forest productivity were highest in close proximity to habitation sites and decreased within 200 m from the shell midden boundary. Trees growing on habitation sites were taller than those growing off of habitation sites (Welch’s two sample t-test: t=2.61, d.f.=12.27, P=0.022; Fig. 2a) and deeper shell middens exhibited a larger effect (Fig. 2b). This pattern was also observed in forest width (t=3.34, d.f.=12.36, P=0.006; Fig. 3) and vegetation greenness (t=3.12, P=0.009; Fig. 4), and, to a lesser extent, forest canopy cover (t=1.86, d.f.=12.58, P=0.087; Fig. 5). Only the amount of insolation was better than the distance from the shell midden boundary on habitation sites in explaining patterns of forest height across the coastal forests in our study site (28 and 24%, respectively) (Table 2). The most important factor explaining patterns of forest width was distance from shell midden boundary on habitation sites (38%) (Table 2). For vegetation greenness and forest canopy cover, distance from habitation site was similar to slope, aspect and elevation in explaining forest productivity patterns (20%) (Table 2).

Figure 2: Habitation site and shell midden depth influence on forest canopy height.

(a) Forest canopy height (in metres) by distance from the shell midden boundary (red) or distance from random locations (blue) shown with s.d. (red and blue ribbon) of bootstrapped boosted regression tree models, which are an additive regression model where terms are simple trees that are fitted in a forward, stagewise manner52. Mean canopy height across entire study area is shown by a dashed line (15.70 m). All distances are measured within the 30 m ribbon delineated from lidar data. (b) Box plot of difference in canopy height by site type and the depth of shell midden (deep (D)=dark green, medium depth (M)=green, shallow (S)=light green) with forest canopy height at habitation sites significantly greater than all non-habitation sites (P=0.02). Control sites are located along the coast in areas without shell midden where ‘All’ sites include the entire study area shown in yellow in Fig. 1b. Boxes represent first, second and third quartiles with whiskers extending to the highest and lowest value that is within 1.5 × interquartile range (IQR). Significantly higher values are denoted by ‘*’.

Figure 3: Habitation site and shell midden depth influence on forest width.

(a) Forest width (in metres) by distance from the shell midden boundary (red) or distance from random locations (blue) shown with s.d. of bootstrapped boosted regression tree models and mean forest width across entire study area shown by a dashed line (82.67 m). (b) Box plot of difference in canopy height by site type and size of the depth of shell midden (deep (D)=dark green, medium depth (M)=green, shallow (S)=light green) with forest width at habitation sites being significantly greater than all non-habitation sites (t=3.341, P=0.006). Control sites are located along the coast in areas without shell midden where ‘All’ sites include the entire study area shown in yellow in Fig. 1. Boxes represent first, second and third quartiles with whiskers extending to the highest and lowest value that is within 1.5 × IQR. Significantly higher values are denoted by ‘*’.

Figure 4: Habitation site and shell midden depth influence on vegetation greenness.

(a) Vegetation greenness by distance from the shell midden boundary (red) or distance from random locations (blue) shown with s.d. of bootstrapped boosted regression tree models and mean vegetation greenness across entire study area shown by a dashed line (0.117). (b) Box plot of difference in vegetation greenness by site type and the depth of shell midden (deep (D)=dark green, medium depth (M)=green, shallow (S)=light green) with vegetation greenness at habitation sites being significantly higher compared with all non-habitation sites (t=3.117, P=0.009). Control sites are located along the coast in areas without shell midden where ‘All’ sites include the entire study area shown in yellow in Fig. 1. Boxes represent first, second and third quartiles with whiskers extending to the highest and lowest value that is within 1.5 × IQR. Significantly higher values are denoted by ‘*’.

Figure 5: Habitation site and shell midden depth influence on forest canopy cover.

(a) Forest canopy cover (in percentage) by distance from the shell midden boundary (red) or distance from random locations (blue) shown with s.d. of bootstrapped boosted regression tree models and mean forest canopy cover across entire study area shown by a dashed line (73.01%). (b) Box plot of difference in forest canopy cover by site type and the depth of shell midden (deep (D)=dark green, medium depth (M)=green, shallow (S)=light green) with forest canopy cover at habitation sites being moderately higher compared with all non-habitation sites (t=1.857, P=0.087). Control sites are located along the coast in areas without shell midden where ‘All’ sites include the entire study area shown in yellow in Fig. 1. Boxes represent first, second and third quartiles with whiskers extending to the highest and lowest value that is within 1.5 × IQR.

Table 1 Description of habitation and control sites.
Table 2 Results from bootstrapped boosted tree regression models for measures of forest productivity.

Nutrient modifications enhance forest productivity

Soil from habitation sites can explain these striking differences in productivity, being significantly higher in calcium (nested analysis of variance: F=12.34, d.f.=1, P=0.001; Fig. 6a), phosphorous (F=7.01, d.f.=1, P=0.012) and pH (F=19.55, d.f.=1, P=0.001; Fig. 7a), and lower in potassium (F=5.03, d.f.=1, P=0.032; Fig. 7b). Nitrogen, sulphur and magnesium levels were similar (Fig. 7c–f). Western redcedar trees growing on habitation sites had significantly higher wood calcium levels (F=4.41, d.f.=1, P=0.037; Fig. 6b), greater radial growth (F=23.61, d.f.=1, P<0.001; Fig. 6c) and experienced less top die-back (χ2-test=14.19, d.f.=1, P<0.001; Fig. 6d). Lastly, above-ground evidence of fire was exclusively associated with habitation sites (Table 2) and this may also have contributed to forest productivity.

Figure 6: Soil calcium dynamics and ecological responses of western redcedar.

Levels of calcium in (a) soil and (b) western redcedar (T. plicata) wood, and with forest productivity metrics of (c) radial growth and (d) the proportion of western redcedar trees exhibiting top die-back (image showing redcedar with (i) and without (ii) top die-back). Red circles correspond to habitation sites with shell middens (n=6; see Table 1 for details). Blue circles correspond to control sites that were selected based on similar composition (dominance of western redcedar), high productivity and lacking a shell midden (hence was considered not to be a habitation site). Coloured circles are mean values and error bars represent standard error. Image credit: A. Trant.

Figure 7: Soil nutrients and properties at habitation and control sites.

Habitation sites had (a) significantly more basic soils (higher pH) (F=19.55, df=1, P<0.001), (b) significantly higher potassium (K) in control sites (F=5.03, df=1, P=0.032) and (c) significantly higher phosphorous (P) in habitation sites (F=7.01, df=1, P=0.012). There were no significant differences for (d) sulphur (S) (F=1.74, df=1, P=0.196), (e) nitrogen (N) (F=1.56, df=1, P=0.218) and (f) magnesium (Mg) (F=0.63, P=0.432). Coloured circles are mean values and error bars represent s.e. Significantly higher values are denoted by ‘*’.


Although pre-occupation site productivity is unknown, multiple lines of evidence demonstrate that productivity is enhanced at habitation sites, independent from any site selection effect that may exist.

First, from an ecological and archaeological perspective, the dominant tree species, western redcedar, did not appear in this region until at least 7,000–8,000 Cal BP35,36, with modern vegetation arising 4,000–2,000 Cal BP37, millennia after many of these sites were originally occupied (Table 1). Second, shell middens and historic fire dramatically change local site topography, nutrient dynamics, soil pH and site drainage. We are therefore confident that current measures of forest productivity are enhanced by these sites being occupied, regardless of site conditions before occupation.

Local patterns in forest productivity in these coastal regions have previously been shown to be primarily driven by insolation (aspect), water retention of soils and nutrient availability38. Our data demonstrate similar effects of these factors, but also reveal that distance from habitation sites is routinely ranked as one of the most important predictors of forest productivity (Table 2). Thus, shell midden habitation sites and the occurrence of historic fires transform the local site conditions through increased soil pH and macronutrient inputs39,40, better site drainage and more level ground, thus offering a mechanism by which forest productivity is enhanced in an otherwise nutrient-limited ecosystem. Shell middens have been shown to alter soil chemistry resulting in greater vegetation cover and higher species richness29 with qualitative reports of vegetation growing on shell middens being botanically different41, and greener and denser42. However, this is the first known documentation of changes to forest productivity resulting from long-term intertidal resource use patterns. Wider forests around habitation sites may be related to the nutrient and drainage benefits from shell middens. In addition, the cultural use of fire at habitation sites probably extended beyond the boundaries of the shell midden either intentionally or accidentally. Less pronounced effects of habitation sites on forest canopy cover might be influenced more by canopy competition for light than nutrients and site drainage.

We reveal a strong association between habitation sites and low-severity fire. The cool climate, high levels of annual precipitation (>4,000 mm) and low probability of summer lightning on Canada’s west coast render lightning-caused fires unlikely43. Historically, low-severity fires may have been an important contributor to increased forest productivity in our study area and may also be associated with active management practices such as land clearing, timber harvesting and promoting culturally important species such as berry-producing shrubs44,45.

Ecosystems with a history of extensive human use through commercial logging, development or other forms of contemporary resource extraction are often considered degraded and disturbed. Here we offer alternative consequences of extensive and long-term human management in coastal areas. Pockets of enhanced forest productivity are associated with increased phosphorous availability resulting from higher soil pH from the slow leaching of calcium from shell middens along with the nutrient amendments of past fires. Taken together, this leads to higher forest productivity than might otherwise be predicted for coastal soils with high nitrogen and low availability of other important nutrients such as calcium46. With a deep time perspective from 13,000 years of repeated occupation of the study area, it is clear that coastal First Nations people have developed practices that enhanced nutrient-limited ecosystems, making the environment that supported them even more productive. Middens containing large amounts of bivalves and univalves47 are common regionally48 and worldwide49, making the patterns we report generalizable to other coastal landscapes.


Study site

This research was conducted in the Hakai Lúxvbálís Conservancy on Calvert and Hecate Islands (N 51° 38′–W 128° 05′), located on the outer coast of central British Columbia (Fig. 1). The study site is situated in the very wet hypermaritime subzone (CWHvh2), a variant of the Coastal Western Hemlock biogeoclimatic classification50. The CWHvh2 receives some of the highest annual rainfall in North America (4,000 mm) and is characterized by a moderate year-round climate (average temperature 7 °C).

Forests in the nearshore environment are dominated by western redcedar (T. plicata Donn ex D. Don) and western hemlock (Tsuga heterophylla [Raf.] Sarg.), with lesser amounts of yellow-cedar (Cupressus nootkatensis [D.Don] Farjon and Harder) and Sitka spruce (Picea sitchensis [Bong.] Carr.). Productive forests growing in moderately drained soils are in contrast to the majority of forests in the study area, which exhibit stunted growth forms as a result of prolonged saturation of soils and acidic conditions associated with high rates of peat accumulation51. Compared with inland areas of the British Columbia coast, Calvert and Hecate Islands and are comprised of homogenous and nutrient-poor quartz granodiorite bedrock geology51.

We selected sites (n=15) in formerly occupied habitations that vary in usage type, intensity, year of establishment and year since last occupied (Table 1). All habitation sites in this study contain shell middens (>50 cm depth) and were categorized as shallow=50–99 cm, medium=100–249 cm and deep>250 cm. Field sites were compared with lidar analyses of the entire coastline contained within the study area (see below). For detailed ecological sampling and fire surveys, a subset of habitation sites (n=6 and n=7, respectively) were compared with control sites that were selected for having high productivity, similar forest structure and site attributes (bedrock and surficial till type and extent, aspect, slope and proximately to shore) but lacked any above or below ground evidence of shell midden (Fig. 1b) and thus are considered not to be habitation sites.


Shell midden sites were identified using map information from the British Columbia Site Inventory Database where possible. Field-based methods were also used to identify previously unrecorded shell middens. Field crews employed different types of subsurface testing methods to identify the extents of these deposits including the use of augers, probes, cores, shovel tests and excavation units. Charcoal and plant macrofossil samples from these tests were selected for radiocarbon dating and sent to the W.M. Keck Carbon Cycle Accelerator Mass Spectrometry Lab at the University of California, Irvine, for analysis.

Remote sensing data

Discrete return airborne scanning lidar was acquired on 15 August 2012 by Terra Remote Sensing Inc. (Sidney, BC, Canada). Lidar was collected from 1,150 m AGL at 100 kHz with a maximum scan angle of 26°. The resulting point data have an average point density of 2 points per m2 with an average vertical accuracy of 15 cm. A digital terrain surface was generated from classified ground returns using triangular irregular network interpolation and rasterized at a spatial resolution of 1.0 m. The terrain model was used to derive elevation, slope, aspect and upstream area layers, as well as to normalize non-ground lidar returns to height above ground surface. Vegetation greenness as expressed in the Normalized Difference Vegetation Index was calculated from SPOT6 images acquired on 10 August 2014 and corrected to surface reflectance.

Boosted regression tree modelling

To evaluate the effect of proximity to habitation sites on vegetation structure compared with other landscape factors we fit a Boosted Regression Tree model for four metrics of forest productivity acquired through remote sensing: forest canopy height, forest width, vegetation greenness, represented by Normalized Difference Vegetation Index, and forest canopy cover. Boosted Regression Tree modelling is flexible, as it accepts both continuous and classified variables and also allows for nonlinear relations52. In addition to the terrain covariates, we also included surface material and coastal morphology and exposure from the British Columbia shorezone map53. The study area has a strong gradient of forest structure perpendicular to the shoreline with productive forests on the coastline and mostly marginally productive vegetation and bog areas inland. To account for this gradient and coastal edge effects, we extracted all vegetation data and the covariates from cells that extended 30 m inland and ran 50 m along the shoreline. These cells were set in 10 m from the coastal forest boundary given that vegetation structure is highly variable due to exposure and edge effects. All model data were averaged to 50 m wide sections of this coastal buffer on the study area (n=600). Models were built using Boosted Regression Trees from the ‘gbm’ package in R statistical software54, following the methods described in Friedman55,56. We determined optimal fitting parameters in an initial exploration using tenfold cross-validation at a tree complexity of 4, minimum number of samples per node at 5 and a learning rate of 0.01. The number of trees was optimized for each individual model and was between 700 and 1,400. The only model settings and covariates that varied were distance to habitation and control sites, and we focus on the effect of these parameters. Model performance is indicated by the per cent deviance explained and we consider the effect of individual covariates as their relative influence (as per cent of total deviance explained). Models were bootstrapped (100 runs using 80% of data per run) to generate the s.d. that appears on mean values by distance from habitation site and random location. Differences in forest canopy height, forest width, forest canopy cover and vegetation greenness between habitation sites and non-habitation sites were assessed using a Welch’s two-sample t-test.

Fire surveys

We ran belt transects to examine the presence of fire in forests surrounding habitation and control sites (n=7). The location of transects was randomly assigned proportional to the size of the nearshore site area and transects 6 m wide and 30 m long (n=3) were completed perpendicular to the shoreline at each site. Trees with a diameter at breast height >15 cm diameter were sampled and tree height, health (top die-back, rot) and decay class were recorded. We also recorded the presence of culturally modified trees, char and fire scars.

Nutrients and tree radial growth

At each habitation and control site (n=6), we sampled 9–14 western redcedar trees and using a 5.15 mm increment borer, extracted 2 cores at breast height (1.4 m) for determining radial growth and wood nutrient analyses. For each tree, we recorded physical attributes of diameter at breast height, height and health (for example, top die-back). Increment samples for radial growth analyses were air-dried, mounted and sanded to a high polish. All western redcedar samples were measured and counted using a Velmex sliding stage micrometre (precision 0.001 mm). For wood nutrient analyses, increment samples were cut into decadal units using a scalpel under a dissection scope and stored in paper coin envelopes. Three samples from each tree were analysed for macro and micronutrients for the period 1905–1914, 1955–1964 and 2005–2014, representing evenly spaced decadal sections across the longest shared record.

At all habitation and control sites, we collected soil samples using an AMS soil auger with a diameter of 5.7 cm. Soil samples were taken at 30 cm below litter layer where we were assumed to be in the active rooting layer and when shell middens were present, samples were taken from soils developed on top of shell middens. Large roots and plant material were removed from the samples. Samples were stored in plastic bags and refrigerated until analysed. All soil and wood samples were analysed for nutrient content by the North Road Laboratory facility of the Research, Innovation and Knowledge Management Branch of the British Columbia Ministry of Forest, Lands and Natural Resources.

Differences in nutrients and radial growth were determined using a nested analysis of variance with sites being nested within site type (habitation or control). Differences in top die-back between site type were determined using a χ2-test to account for binary data. To be consistent between approaches, five trees were excluded from analyses, because they were >40 m from the shoreline and hence were outside of the area included in lidar data.

Data availability

The data that support the findings of this study are available from the corresponding author upon request.

Additional information

How to cite this article: Trant, A. J. et al. Intertidal resource use over millennia enhances forest productivity. Nat. Commun. 7:12491 doi: 10.1038/ncomms12491 (2016).


  1. 1

    Foley, J. A. et al. Global consequences of land use. Science 309, 570–574 (2005).

    CAS  ADS  Article  Google Scholar 

  2. 2

    Hoekstra, J. M., Boucher, T. M., Ricketts, T. H. & Roberts, C. Confronting a biome crisis: global disparities of habitat loss and protection. Ecol. Lett. 8, 23–29 (2005).

    Article  Google Scholar 

  3. 3

    Engdawork, A. & Bork, H.-R. Long-term indigenous soil conservation technology in the Chencha Area, Southern Ethiopia: origin, characteristics, and sustainability. Ambio 43, 932–942 (2014).

    Article  Google Scholar 

  4. 4

    De Smidt, J. Heathland vegetation in the Netherlands. Phytocoenologia 4, 258–316 (1977).

    Google Scholar 

  5. 5

    Posey, D. A. Indigenous management of tropical forest ecosystems: the case of the Kayapo Indians of the Brazilian Amazon. Agroforestry Syst. 3, 139–158 (1985).

    Article  Google Scholar 

  6. 6

    Oyuela-Caycedo, A. The forest as a fragmented archaeological artifact. The Archaeology of Anthropogenic Environments, Occasional Paper 37, 75–94 (2010).

    Google Scholar 

  7. 7

    Deur, D. & Turner, N. Keeping it Living: traditions of Plant Use and Cultivation on the Northwest Coast of North America Univ. Washington Press (2005).

  8. 8

    Hebda, R. J. & Mathewes, R. W. Holocene history of cedar and native Indian cultures of the North American Pacific Coast. Science 225, 711–713 (1984).

    CAS  ADS  Article  Google Scholar 

  9. 9

    Garibaldi, A. & Turner, N. Cultural keystone species: implications for ecological conservation and restoration. Ecol. Soc. 9, 1 (2004).

    Article  Google Scholar 

  10. 10

    Summary of cedar management considerations for coastal British Columbia. 1–13 (West Coast Region, Ministry of Natural Resource Operations and Forest Analysis and Inventory Branch, Ministry of Forests, Mines and Lands, Victoria, 2011).

  11. 11

    McLaren, D. et al. A post-glacial sea level hinge on the central Pacific coast of Canada. Quat. Sci. Rev. 97, 148–169 (2014).

    ADS  Article  Google Scholar 

  12. 12

    McLaren, D., Rahemtulla, F., Gitla, (White, E.) & Fedje, D. Prerogatives, sea level, and the strength of persistent places: archaeological evidence for long-term occupation of the Central Coast of British Columbia. BC Stud 187, 155–193 (2015).

    Google Scholar 

  13. 13

    Cannon, A., Burchell, M. & Bathurst, R. in Early Human Impact on Megamolluscs Vol. 1865, 7–22Archaeo Press Oxford (2008).

    Google Scholar 

  14. 14

    Lepofsky, D. et al. Ancient shellfish mariculture on the Northwest Coast of North America. Am. Antiquity 80, 236–259 (2015).

    Article  Google Scholar 

  15. 15

    Deur, D., Dick, A., Recalma-Clutesi, K. & Turner, N. J. Kwakwaka’wakw ‘Clam Gardens’. Hum. Ecol. 43, 201–212 (2015).

    Article  Google Scholar 

  16. 16

    Sawbridge, D. F. & Bell, M. A. M. Vegetation and soils of shell middens on the coast of British Columbia. Ecology 53, 840–849 (2001).

    Article  Google Scholar 

  17. 17

    Vanderplank, S. E., Mata, S. & Ezcurra, E. Biodiversity and archeological conservation connected: aragonite shell middens increase plant diversity. BioScience 64, 202–209 (2014).

    Article  Google Scholar 

  18. 18

    Rahemtulla, F. Archaeological Investigations of the Site EjTa-0004 located on Calvert Island, in the Hakai Luxvbalis Protected Area. Central Coast of BC (Permit 2012-0155) BC Archaeology Branch (2014).

  19. 19

    Lyons, N. & Orchard, T. J. Sourcing Archaeobotanical remains: taphonomic insights from a midden analysis on Haida Gwaii, British Columbia. Can. J. Archaeol. 28–54 (2007).

  20. 20

    Conover, K. J. Archeological Sampling at Namu: a Problem in Settlement Reconstruction UMI Dissertation Services (1973).

  21. 21

    Bowman, D. M. et al. The human dimension of fire regimes on Earth. J. Biogeogr. 38, 2223–2236 (2011).

    Article  Google Scholar 

  22. 22

    Ryan, K. C., Knapp, E. E. & Varner, J. M. Prescribed fire in North American forests and woodlands: history, current practice, and challenges. Front. Ecol. Environ. 11, e15–e24 (2013).

    Article  Google Scholar 

  23. 23

    Boyd, R. Smallpox in the Pacific Northwest: the first epidemics. B C Stud. 101, 5–40 (1994).

    Google Scholar 

  24. 24

    Olson, R. L. Notes on the Bella Bella Kwakiutl Vol. 14, (Univ. California Press (1955).

  25. 25

    White, E. Heiltsuk Stone Fish Traps on the Central Coast of British Columbia 75–90Univ. Alaska Press (2011)

  26. 26

    Maxwell, J., McLaren, D., Eldridge, M., Chattan, R. & Hoffmann, T. Heiltsuk Traditional Territory Archaeological Resource Overview (Culture Library of the Ministry of Small Business, Tourism, and Culture, Victoria, BC, (1997).

  27. 27

    Hocking, M. D. & Reynolds, J. D. Impacts of salmon on riparian plant diversity. Science 331, 1609–1612 (2011).

    CAS  ADS  Article  Google Scholar 

  28. 28

    Johnson, D. W. Nitrogen retention in forest soils. J. Environ. Qual. 21, 1–12 (1992).

    Article  Google Scholar 

  29. 29

    Cook-Patton, S. C., Weller, D., Rick, T. C. & Parker, J. D. Ancient experiments: forest biodiversity and soil nutrients enhanced by Native American middens. Landscape Ecol. 29, 979–987 (2014).

    Article  Google Scholar 

  30. 30

    Egan, B. The Ecology of the Coastal Western Hemlock Zone. 1–6 (Ministry of Forests, British Columbia, 1999).

  31. 31

    Radwan, M. & Harrington, C. A. Foliar chemical concentrations, growth, and site productivity relations in western red cedar. Can. J. Forest Res. 16, 1069–1075 (1986).

    CAS  Article  Google Scholar 

  32. 32

    Demeyer, A., Nkana, J. V. & Verloo, M. Characteristics of wood ash and influence on soil properties and nutrient uptake: an overview. Bioresource Technol. 77, 287–295 (2001).

    CAS  Article  Google Scholar 

  33. 33

    Ceci, L. Shell midden deposits as coastal resources. World Archaeol. 16, 62–74 (1984).

    Article  Google Scholar 

  34. 34

    Glaser, B., Lehmann, J. & Zech, W. Ameliorating physical and chemical properties of highly weathered soils in the tropics with charcoal–a review. Biol. Fertil. Soils 35, 219–230 (2002).

    CAS  Article  Google Scholar 

  35. 35

    Galloway, J. M., Doherty, C. T., Patterson, R. T. & Roe, H. M. Postglacial vegetation and climate dynamics in the Seymour-Belize Inlet Complex, central coastal British Columbia, Canada: palynological evidence from Tiny Lake. J. Quat. Sci. 24, 322–335 (2009).

    Article  Google Scholar 

  36. 36

    Eamer, J. L. Late Quaternary vegetation dynamics on Calvert Island, British Columbia 1–23Hakai Institute (2016) at

  37. 37

    Hebda, R. J. British Columbia vegetation and climate history with focus on 6 ka BP. Géogr. Phys. Quat. 49, 55–79 (1995).

    MathSciNet  Google Scholar 

  38. 38

    Banner, A., LePage, P., Moran, J. & de Groot, A. The HyP3 Project: pattern, process, and productivity in hypermaritime forests of coastal British Columbia–a synthesis of 7-year results. BC Min. For. Res. Br. Vic. BC Spec. Rep. 10, 142 (2005).

    Google Scholar 

  39. 39

    Lombardo, U. et al. Early and middle Holocene hunter-gatherer occupations in Western Amazonia: the hidden shell middens. PLoS ONE 8.8, e72746 (2013).

    ADS  Article  Google Scholar 

  40. 40

    Knicker, H. Pyrogenic organic matter in soil: Its origin and occurrence, its chemistry and survival in soil environments. Quat. Int. 243, 251–263 (2011).

    Article  Google Scholar 

  41. 41

    Hrdlicka, A. Man and plants in Alaska. Science 86, 559–560 (1937).

    ADS  Article  Google Scholar 

  42. 42

    Meigs, P. Vegetation on shell mounds, Lower California. Science 87, 346–346 (1938).

    ADS  Article  Google Scholar 

  43. 43

    Burrows, W. R. et al. Lightning occurrence patterns over Canada and adjacent United States from lightning detection network observations. Atmos. Ocean 40, 59–80 (2002).

    Article  Google Scholar 

  44. 44

    Turner, N. J. in Indians, Fire and the Land in the Pacific Northwest 185–218Oregon State Univ. Press (1999).

  45. 45

    Hoffman, K. M., Gavin, D. G., Lertzman, K. P., Smith, D. J. & Starzomski, B. M. 13,000 years of fire history derived from soil charcoal radiocarbon dates in a British Columbia coastal temperate rainforest. Ecosphere (In press).

  46. 46

    Perakis, S. S. et al. Forest calcium depletion and biotic retention along a soil nitrogen gradient. Ecol. Appl. 23, 1947–1961 (2013).

    Article  Google Scholar 

  47. 47

    Cannon, A. in Shell Energy: Mollusc Shells as Coastal Resources eds Bailey G. N., Hardy K., Camara A. 21–34Oxbow Books (2013).

  48. 48

    McMillan, A. D. & McKechnie, I. Investigating indigenous adaptations to British Columbia’s exposed outer coast: Introduction to these outer shores. B C Stud. 187, 3–20 (2015).

    Google Scholar 

  49. 49

    Bailey, G. N., Hardy, K. & Camara, A. Shell Energy: mollusc Shells as Coastal Resources Oxbow Books (2013).

  50. 50

    Meidinger, D. & Pojar, J. Ecosystems of British Columbia. Special Report Series Number 6 Research Branch, Ministry of Forests, Victoria (1991).

  51. 51

    Banner, A., LePage, P., Moran, J. & de Groot, A. The HyP 3 Project: pattern, process, and productivity in hypermaritime forests of coastal British Columbia–a synthesis of 7-year results. BC Min. For. Res. Br. Vic. BC Spec. Rep. 10, 142 (2005).

    Google Scholar 

  52. 52

    Elith, J., Leathwick, J. R. & Hastie, T. A working guide to boosted regression trees. J. Anim. Ecol. 77, 802–813 (2008).

    CAS  Article  Google Scholar 

  53. 53

    Howes, D., Harper, J. & Owens, E. Physical shore-zone mapping system for British Columbia. Report prepared by Environmental Emergency Services, Ministry of Environment (Victoria, BC), Coastal and Ocean Resources Inc. (Sidney, BC), and Owens Coastal Consultants (Bainbridge, WA) (1994).

  54. 54

    R Development Core Team (2015) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL

  55. 55

    Friedman, J. H. Stochastic gradient boosting. Comput. Stat. Data Anal. 38, 367–378 (2002).

    MathSciNet  Article  Google Scholar 

  56. 56

    Friedman, J. H. Greedy function approximation: a gradient boosting machine. Ann. Stat. 29, 1189–1232 (2001).

    MathSciNet  Article  Google Scholar 

Download references


This research was supported by funding to B.M.S from the Hakai Institute, an NSERC Discovery Grant (386594-2010), the Canada Foundation for Innovation Leader’s Opportunity Fund (26102), the British Columbia Knowledge Development Fund (804609) and the Ian McTaggart Cowan Professorship at the University of Victoria. We thank E. Peterson, C. Munck, B.C. Parks, J. Bennett, C. Darimont, C. Dawson, J. Fisher, O. Fitzpatrick, K. Fretwell, S. Friesen, B. Hawkins, W. Housty, P. Johnson, K. Jordan, A. MacKinnon, J. Maxwell, W. McInnes, I. McKechnie, M. Oelbermann, F. Rahemtulla, J. Reynolds, N. Shackelford, J. Silberg, D. Smith, J. Stafford, N. Turner, E. Urquhart, J. Vickers, E. White and S. Wickham.

Author information




A.J.T. and B.M.S. originally conceived the study, with W.N. and K.M.H. adding additional methods. A.J.T., K.M.H., B.M.S., D.M. and D.L.M. conducted the fieldwork. A.J.T. and W.N. performed the data analysis aided by B.M.S., K.M.H. and T.A.N. A.J.T., B.M.S., K.M.H. and W.N. drafted the manuscript. All authors discussed the results and commented on the manuscript.

Corresponding author

Correspondence to Andrew J. Trant.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Supplementary information

Rights and permissions

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Trant, A., Nijland, W., Hoffman, K. et al. Intertidal resource use over millennia enhances forest productivity. Nat Commun 7, 12491 (2016).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing