Introduction

Over the past century the range of many species has changed, often attributed to climate change and land cover modification (Thomas 2010, Laliberte and Ripple 2004, Walther et al 2002). A species can respond to environmental changes by exploiting resources at the extremities of its niche breadth (Sexton et al. 2017), by phenotypic plasticity (Valladares et al. 2014, Nicotra et al. 2010), or by adaptation (Williams et al. 2008). However, the rate at which current conditions are changing might make adaptation impossible for many species because the process of natural selection is too slow (Davis and Shaw 2001). Consequently, species will have to track their bioclimatic niche (Visser 2008) the ability to do so will be an essential characteristic influencing future biodiversity (Bell and Gonzalez 2011, Schloss et al. 2012, Travis et al. 2013).
Terrestrial species that have ranges near the poles will be limited in their ability to track climate because they are limited by the availability of space to move to higher latitudes (Kerr and Packer 1998). Therefore, many unique cold adapted species will eventually perish unless they somehow adapt to much warmer conditions and to new biotic interactions. Understanding how and why the warmer range edge of cold adapted species has been changing would help us in making better informed conservation decisions, since anthropogenic change is not slowing down.
The Canada lynx (Lynx canadensis ) is an iconic carnivore that largely resides in the boreal forest of North America and its northern range has some expansion potential into northern taiga landscapes but is generally bounded by tundra and the Arctic Ocean (Poole 2003). The lynx is a habitat specialist because it almost exclusively preys on snowshoe hares in the boreal forest (O’Donoghue et al. 1998). Consequently, its population dynamics are highly coupled to the 8-11 year population cycle of the snowshoe hare, mirroring it with a 1-2 year delay (Poole 2003). Since presettlement times the Canada lynx range has contracted by 40% (Laliberte and Ripple 2004). However, most of this range reduction took place prior to the 20th century and was attributed to unregulated harvest and habitat loss due to land clearing during European colonization (de Vos 1952, de Vos 1964, Hovinget al. 2003, McKelvey 2000, Poole 2003).
Canada lynx are predominantly found in areas where snowshoe hare density is above 0.5 per hectare (Zahratka and Shenk 2008, Hodges et al.2009, Berg et al. 2012, Ivan et al. 2014). In the southern periphery of the lynx range, hare population densities have declined compared to historic levels (Aubry et al. 2000, Hodges 2000, Murray 2000) and this most likely accounts for the contraction of the lynx from its historic range (Poole 2003). Following years with high peak hare abundance, Canada lynx appear to migrate from range core to range periphery as a result of density dependent dispersal (McKelvey et al. 2000). Such dispersal pulses might lead to higher occupancy of the southern range periphery (McKelvey et al 2000, Murray et al.2008). Consequently, lower peaks in hare abundance might decrease the likelihood of dispersal of lynx into the southern periphery (Poole 2003). Southern dispersal might also be limited in some locations by connectivity with the range core (Ruggiero et al. 2000, Buskirk 2000, Walpole et al. 2012).
The warming climate might indirectly impact the lynx through its main food source the snowshoe hare. The timing between molt and season change for the snowshoe hare is important in decreasing predation rates (Zimova et al. 2016). A changing snow regime could increase snowshoe hare predation rates by increasing the rate of mismatch between snowshoe hare molt and season change. Increased predation rates might also reduce the amplitude of the hare cycle (Krebs 2010).
Climate change will also open formerly inhospitable habitat to new species in the lynx range. Bobcats (Lynx rufus ) and coyotes (Canis latrans ) have smaller feet than Canada lynx, consequently they cannot support as much weight as lynx in deep snow without sinking (Parker et al. 1984). This might be one factor that has hindered the bobcat from invading Canada lynx territory in the past (Marston 1942, McCord 1974, Murray et al. 2008, Parker et al.1984). However, since the climate is warming, and snow depths across the southern periphery of the lynx range are shallower, southern competitors might be less hindered by snow, increasing their competitive potential (Buskirk et al. 2000, Ruediger et al. 2000, Scullyet al. 2018). In fact, Parker et al. (1984) found that after several years of low snow the bobcat invaded the lowlands of Cape Breton while the Canada lynx left the area. Marrotte et al.(2020a) found that deep winter snow in the Great Lakes region limited bobcat expansion northward, suggesting that greater expansion will result from additional climate warming.
The lynx once occurred in 24 of the United States (McKelvey 2000), but currently only occurs in 7 (US Fish and Wildlife Service 2017). As a result, the lynx is designated as ‘threatened’ in the contiguous United States (US Fish and Wildlife Service 2000). Despite the range contraction, its protection status is being debated and it might be removed from the list of endangered species in the United States. In Canada, the lynx occupies 95% of its historic range (Poole 2003). However, it is designated as provincially endangered in Nova Scotia (Parker 2001), and New Brunswick (New Brunswick Endangered Species Regulation 2013) and was extirpated from Prince Edward Island (Poole 2003). Further analysis has demonstrated that the range of lynx in British Columbia has been stable since the 1930s (Gooliaff and Hodges 2018). In contrast, the lynx range in Ontario appears to have contracted northwards by 175 km from 1972 to 2010 (Koen et al. 2014).
We estimated the past extent of the Canada lynx southern range in Ontario, Canada using harvest records and then determined whether the spatial-temporal patterns could be attributed to snowshoe hare and boreal lynx population dynamics, connectivity, climate, land use and competition. We predicted that years with low Canada lynx abundance in the boreal forest were associated with a reduction in the extent of the southern lynx range. We also predicted that areas with high human disturbance, shallow snow, presence of competitors, and with low connectivity to boreal lynx populations were less likely to be part of the southern range.