Choice of baseline affects historical population trends in hunted mammals of North America

Abstract

Establishing historical baselines of species' populations is important for contextualising present-day population trends, identifying significant anthropogenic threats, and preventing a cultural phenomenon known as ‘shifting baseline syndrome’. However, our knowledge of historical baselines is limited by a lack of direct observation data on species abundance pre-1970. We present historical data of species-specific fur harvests from the Canadian government and Hudson's Bay Company as a proxy for estimating species abundance over multiple centuries. Using stochastic stock reduction analysis originally developed for marine species, we model historical population trends for eight mammals, and assess population trends based on two different baseline years: 1850 and 1970. Results show that population declines are significantly greater when using an 1850 baseline, as opposed to a 1970 baseline, and for four species, the population trend shifted from a population increase to a decrease. Overall, the median population change of the eight species changed from a 15% decline for 1850, to a 4% increase for 1970. This study shows the utility of harvest data for deriving population baselines for hunted terrestrial mammals which can be used in addition to other historical data such as local ecological knowledge. Results highlight the need for developing historically relevant population baselines in order to track abundances over time in threatened species and common species alike, to better inform species conservation programs, wildlife management plans and biodiversity indicators.

Introduction

Species population declines and extinctions undermine the functioning and resilience of ecosystems on which humans and wildlife depend (Cardinale et al., 2012; Oliver et al., 2015). To monitor and respond to species losses, changes in population abundance are used as a sensitive metric of change (Collen et al., 2011; Shoemaker and Akçakaya, 2015) and have been incorporated into globally adopted biodiversity indicators such as the Living Planet Index, which tracks changes in vertebrate population abundance from 1970 (Collen et al., 2009). However, data on population abundance typically become scarcer beyond a few decades from the present, prior to the implementation of species monitoring programmes (Willis et al., 2005; Bonebrake et al., 2010).

Knowledge of historical populations acts as an antidote to ‘shifting baseline syndrome’; a phenomenon in which with each new human generation comes a lowered expectation of a species population norm (Pauly, 1995; Kahn and Friedman, 1995; Soga and Gaston, 2018). Historical population baselines have many practical policy implications, for example when defining population recovery and conservation legacy, deciding harvest quotas, and influencing the general public's perception of a species (Papworth et al., 2009; Davies et al., 2014; Roman et al., 2015; Akcakaya et al., 2018; see Fig. 1a). Additionally, estimating historical populations can help to differentiate between a population trend that is unidirectional or cyclical, such as the Atlantic Multidecadal Oscillation inducing bidirectional changes in fish abundance (Jackson et al., 2001; Willis et al., 2007; Sundby and Nakken, 2005; see Fig. 1b). Without long-term measurements, observers may misattribute downward phases of natural population cycles as human-caused population declines (Koslow and Couture, 2013). Finally, historical population data can help to identify historic drivers of population change (see Fig. 1c), which is important for quantifying the relative significance of each past and present threat in order to develop threat-specific management strategies and inform future scenario modelling (Baker and Clapham, 2004; Pinnegar and Engelhard, 2008).

Many techniques available to reconstruct historical population baselines emerged from the discipline of marine historical ecology (Lotze and Worm, 2009). Faced with the need to sustainably manage fish stocks, fisheries researchers have used recorded history (e.g. ‘local ecological knowledge’) (Sáenz-Arroyo et al., 2005; Turvey et al., 2010), archaeogenomic data (e.g. analysis of relative stable isotope concentrations) (Finney et al., 2002), and fish stock assessments from historical catch data (Myers and Worm, 2003; Baker and Clapham, 2004) to extrapolate population size over time and capture stock collapses that pre-date direct monitoring.

Recorded history has also provided us with historical population estimates for terrestrial species, although not as frequently as in the marine realm. These studies are extremely valuable in painting a picture of past population condition (Cole and Woinarski, 2000; Rowe and Terry, 2014), but with each historical data source comes its own unique set of limitations. For instance, museum and fossil records are often patchy and taxonomically biased, and local ecological knowledge generally only covers a couple of generations spanning <100 years (Miller, 2011). Here, we add to our growing knowledge on reconstructing population baselines by focussing on harvest data of terrestrial mammals as another data source which holds great potential in historical baseline reconstruction.

Reports from the Hudson's Bay Company (HBC), Canada, have been previously used to document lynx (Lynx canadensis) and muskrat (Ondatra zibethica) population cycles (Elton and Nicholson, 1942a; Elton and Nicholson, 1942b), predator-prey dynamics of lynx and snowshoe hare (Krebs et al., 1995), and the potential roles of climate, productivity and disease in these cycles (Gamarra and Solé, 2000; Yan et al., 2013; Row et al., 2014). Here, we (a) show the utility of these harvest data to reconstruct historic populations by applying a stochastic population model first developed for marine vertebrates (Christensen, 2006), (b) use these population reconstructions to demonstrate that baselines differ when using over 100 years of data compared to <50 years of data and (c) show that choice of different baseline years results in different interpretation of estimated population trends.