Part 5

Did the Roe Deer Expand its Distribution?

If a northward expansion occurred, we should be able to observe a clear latitudinal shift accompanying the threefold rise in harvest densities. To explore this, we used a logistic decline model relating harvest density to midpoint latitude by county from 1966 to 2023. Figure 2.11 shows average harvest density as a function of latitude. The relationship is clearly logistic, with high densities in the south, a relatively sharp transition zone, and very low densities toward the northern limit.

Here, D(lat) is the expected harvest density at a given latitude. The parameter Dmax represents the maximum harvest density under southern conditions. The parameter L₅₀ denotes the latitude at which harvest density has declined to 50% of its maximum, and thus marks the midpoint of the northward distribution. The parameter k controls the steepness of the decline, indicating how rapidly harvest density decreases with latitude.

Let us examine how L₅₀ changes over time (Fig. 2.12). The graph shows the movement of the estimated L₅₀, where the dotted horizontal lines mark the upper and lower limits. The dashed red line marks the peak harvest year in 1993. The result contradicts the common view that the increase in harvest density was caused by a northward expansion. Although harvest levels increased somewhat above L₅₀, the magnitude is small.

The total sum of maximum densities equals 412.4. Of this, 397.2 occurs below L₅₀, whereas only 15.2 occurs above L₅₀. This corresponds to an odds ratio of approximately 26:1 in favor of harvest occurring south of L₅₀. Expressed as percentages, 96.3% of the total harvest density is located south of L₅₀, while only 3.7% occurs north of this midpoint. These numbers strongly weaken the argument that the surge in harvest densities was caused by a northward expansion. Instead, the increase occurred predominantly within the established southern range. This pattern is more consistent with improved foraging conditions, potentially facilitated by vegetation succession and moose-mediated transformation of browse availability, rather than a latitudinal shift in distribution.

The reasons for movements of L₅₀

Roe deer densities rise slowly from very low harvest densities, while L₅₀ moved towards its northernmost latitude limit. Remember that L₅₀ only covers 50% of the possible range of latitudes. This period spans 22 years (1966–1988). After a four-year increase in L₅₀, the roe deer entered a period of fairly stable L₅₀, suggesting that the roe deer reached its possible latitudinal expansion.

Between 1988–1990, L₅₀ relocated southward rapidly by about 94 km. This short time span covers when the roe deer population was at maximum reproductive capacity (K/2). This dip in latitude depends on substantially better foraging conditions, where counties south of L₅₀ responded more quickly, while more northern counties lagged behind. It took another two years for the northern counties to respond to foraging conditions, moving L₅₀ back northward by 42 km.

After peak harvest density (1993), the population entered an extremely rapid decline. This decline, however, was not proportional along the latitude gradient. The fast depletion of forage abundance was likely fairly similar, but intermediate and northern counties were hit harder than can be explained by rapidly reduced carrying capacity, again resulting in a drastic southward shift in L₅₀. The disproportional decline near and above L₅₀ can rather be explained by the presence of European lynx, a highly specialized roe deer predator.

Between 1997–2007, L₅₀ moved northward again as the disproportional decline along the latitude gradient became balanced. The population density is now approaching actual carrying capacity. At the same time, a gradual increase in vegetation season length improves conditions across the gradient. This reduces the gap between southern and intermediate/northern counties and moves L₅₀ back northward.

After 2007, the estimated L₅₀ reached its all-time lowest value within three years. At this time, the roe deer population densities appear to be in equilibrium with actual carrying capacity from 2010 onward.

From 2010 onward, L₅₀ rose northward again with some fluctuations. These swings are most likely related to lynx predation. From 2015 onward, the trend becomes more stable, and L₅₀ seems to become stable at 59.75° N. Compare this with the maximum estimated limit at 60.6° N. The long-term stability of L₅₀ around 60° N suggests that this latitude represents a genuine ecological boundary for roe deer distribution. The extent to which lynx predation contributes to the shorter-term fluctuations around this boundary is, however, difficult to assess.

The movements of L₅₀ in relation to harvest densities

To assess whether the temporal dynamics of L₅₀ can be linked to changes in population size, we examined its relationship with roe deer harvest density (Fig. 2.13). If harvest density corresponds to population size relative to what the habitat can support (e.g. carrying capacity), changes in harvest density should be reflected in L₅₀.

The line in the plot suggests that this relationship shifts as densities change. At lower densities, L₅₀ increases, indicating a delayed response in northern latitudes relative to southern latitudes. As population densities become closer to what the habitat can support, L₅₀ stabilizes. However, during peak densities, southern latitudes respond earlier and with a larger change, leading to a southward shift in L₅₀. Northern latitudes are also likely to follow this trend, but later.

This comparison allows us to move beyond purely temporal patterns and evaluate whether variation in harvest density may help explain how L₅₀ varies along the latitudinal gradient (Fig. 2.13).

Den snabba ökningen från 1966 fram till den högsta skattade nivån för L₅₀ tyder på att de nordligare länen reagerade senare på förbättrade betesförhållanden än de sydligare. När samtliga områden närmar sig sina respektive bärförmågor, det vill säga när populationstätheterna blir proportionerliga mot den lokala födotillgången, förblir L₅₀ relativt stabil fram till dess att rådjurets avskjutningstäthet börjar öka kraftigt.

När denna ökning tar fart reagerar de sydligare länen tidigare och kraftigare på de förbättrade betesförhållandena, vilket tillfälligt bryter det proportionella sambandet i populationstäthet längs latitudgradienten. Områden vid lägre latituder når därmed jämvikt med den lokala bärförmågan tidigare, vilket leder till att L₅₀ förskjuts söderut. De nordligare länen följer därefter samma utveckling, men når aldrig riktigt samma nivå.

Under senare faser påverkades rörelserna för mittpunkten L₅₀ sannolikt även av en ökande lodjurspredation, särskilt inom området kring L₅₀. Lodjursstammens expansion från slutet av 1990-talet till mitten av 2000-talet kan lokalt ha bidragit till att hålla tillbaka rådjurstätheterna och därmed påverkat de observerade svängningarna i L₅₀. Det övergripande mönstret i Figur 2.13 tyder dock på att tillgången på foder fortfarande är den viktigaste förklaringen.

The importance of mixed foraging

An additional factor is the proportion of pasture by county (Fig. 2.14). Here, ‘pasture’ refers to all grass-dominated agricultural land, including both grazed land and fields used for fodder production (e.g. ley), as well as semi-natural grazing areas. Harvest density increases rapidly with increasing proportions of pasture land, approaching a maximum effect on roe deer harvest densities. Areas with low pasture proportions are dominated by production forests, whereas areas with more pasture support much higher roe deer densities. This points to landscape productivity, rather than geographic expansion, as the main factor behind the rise in roe deer harvest densities. The effect of harsher winter conditions at higher latitudes is another factor that reduces harvest densities.

Pasture land availability determines how much high-quality forage exists in the landscape. Counties with more pasture and agricultural mosaics produce more herbs and nutritious vegetation that roe deer can use. Latitude influences winter severity. Moving northward, winters become colder, snow periods are longer, and access to vegetation becomes increasingly restricted. Latitude also correlates with vegetation productivity and quality. Growing seasons shorten, plant growth slows, and the proportion of nutrient-rich herbs and crops decreases, while forest dominance increases.

Together, these three mechanisms create a gradient where roe deer populations perform best in the south, but north of L₅₀, performance gradually declines. The northward gradient clearly shows a decline in proportions of pasture with increasing latitude (Fig. 2.11). Though an important effect to consider, declining fractions of pasture are not the only factor that hampers roe deer density. Winter severity correlates with increasing latitude, and so do the length and quality of the vegetation season. As pasture proportions are relatively stable over time, these alone cannot explain the national boom in roe deer harvest density in 1992.

Figur 2.15 The plot shows the relation between the proportion of pasture and latitudinal midpoints by counties. The general trend is that pasture proportions decline with the northward gradient. The farther north, the smaller the pasture proportions. The two fitted model lines describe the general decline. The quadratic polynomial provides the best statistical description of the observed data but tends to underestimate pasture proportions in the northernmost latitudes. In contrast, the exponential decay model represents a more ecologically realistic asymptotic decline but overestimates pasture proportions in northern Sweden. However, remember that harsher winter conditions follow with increasing latitudes.

Together, Figures 2.14 and 2.15 help explain why L₅₀ stays within a limited latitudinal range (Fig. 2.12). Pasture availability sets how dense roe deer populations can become, and since pasture declines toward higher latitudes, that ceiling drops as well. At the same time, harsher winter conditions and shorter growing seasons further limit roe deer in the north. This creates a gradient where high densities are mainly reached in the south, keeping L₅₀ within a bounded interval rather than allowing a northward expansion.

Distribution of county-wise peaks in roe deer harvesting

The latitudinal distribution of peak harvest densities is concentrated south of L₅₀, and the strong relationship with pasture proportions therefore provides limited support for a northward expansion of roe deer distribution. Roe deer harvest densities converge with pasture coverage, as pasture supports higher densities (Fig. 2.16). The proportion of pasture land decreases with increasing latitude (Fig. 2.15). In addition, climatic and seasonal conditions become less favorable toward higher latitudes.

Figure 2.16 The graph shows the timing of peak harvest densities for counties near the L₅₀ latitude. The dotted line indicates the peak year around 1992. If a northward expansion had occurred, later peak years would be expected in counties located north of L₅₀. Instead, peak densities are closely synchronized around 1992, suggesting that improved forage availability was the main driver of the increase in roe deer harvests.

The observed boom in harvest density peaks is best explained by modern forestry practices, where forage conditions improved following clear-cutting and subsequent vegetation regeneration. Additional contributing factors include favorable climatic conditions and an already well-established southern distribution. If a northward expansion had occurred, we would expect delayed peaks in harvest densities in counties located north of L₅₀. This is not the case — instead, peak densities occur in the same year (Fig. 2.17).