Densities of the Eurasian Three- toed Woodpecker Picoides tridactylus calculated from sap row surveys are on par with estimates from fixed route bird censusing

94 Citation: Ferry B, Ekenstedt J & Green M. 2021. Densities of the Eurasian Three-toed Woodpecker Picoides tridactylus calculated from sap row surveys are on par with estimates from fixed route bird censusing. Ornis Svecica 31: 94–106. https://doi.org/10.34080/os.v31.22416. Copyright: © 2021 the author(s). This is an open access article distributed under the CC BY 4.0 license, which allows unrestricted use and redistribution, provided that the original author(s) and source are credited. R E S E A RCH PA PE R


Introduction
A global key question is how we can manage land use so that it contributes to the well-being of humans, without jeopardizing the rest of life on Earth. Monitoring the population development of endangered bird species is an important part of that work. In Sweden, territory mapping was used as the main method for bird surveys in the past. This method is based on detailed, repeated surveys of birds in specific habitats, and mostly in relatively small areas (Enemar 1959).
National population sizes for different species can then be calculated from recorded bird densities and the total area of the specific habitat in the whole country. Since 1996, the most important tool for following bird population trends in Sweden is the line transects of the Swedish Bird Survey's fixed routes (Green et al. 2020). The fixed routes are surveyed once a year in early summer, mainly in June, and cover the whole of Sweden in a systematic way.
The advantage of the fixed routes is that they, by the systematic approach, represent all larger-ranging habitats in more or less the same proportion as these cover Sweden. Hence, results from the fixed routes are more representative of the whole country than territory mapping in smaller, non-systematically selected areas. One disadvantage, if one wants to know true population sizes, is that line transects do not result in true densities of birds in the same way as territory mapping does. Line transect data can, however, be used for calculating densities, and overall population size, if assumptions of, or data on, detectability is available for the species and habitats in question. Another disadvantage if one wants to calculate densities and total population sizes based on line transect data collected on the fixed routes is that certain species-such as woodpeckers-are more active at other times of the year or the day than when the fixed routes are surveyed. This means that for the birds that are actually present, the proportion that are seen or heard is relatively low for such species. For following population trends, this does not affect the results in any way as long as the proportion of recorded birds in relation to the ones present remains at the same level. For calculating true densities and overall population sizes, however, this may be a problem. The Eurasian Three-toed Woodpecker Picoides tridactylus (henceforth referred to as the Three-toed Woodpecker) is such a species.
The Three-toed Woodpecker prefers mature, often conifer-dominated forests with large amounts of dead or dying trees (Hogstad 1970, Cramp 1985, Virkkala 1991, Stenberg 1996, Hagemeier & Blair 1997, Fayt 1999, Pechacek & Kristin 2004. The presence of the Threetoed Woodpecker is valuable for forestry as it can play a role in regulating the populations of bark beetles (Coleoptera: Curculionidae, Scolytinae), and especially the European spruce bark beetle lps typographus, in coniferous forest landscapes (Fayt et al. 2004).
However, the species is considered difficult to survey (Winkler et al. 1995, Winkler & Christie 2002. It is easily overlooked, partly because established pairs are very quiet and withdrawn, and it is very difficult to find unless birds are drumming or calling (Amcoff & Eriksson 1996). In addition, some Three-toed Woodpeckers only occasionally respond to playback of recorded drumming. In many cases, the presence of Three-toed Woodpeckers in an area is not detected through observations of the bird itself, but of the tracks and signs of it.
The species has a typical way of pecking rows of small holes around tree trunks, so-called sap rows, to utilize the sap flowing from the holes. The month before nesting starts, Three-toed Woodpeckers can use up to 33 % of its foraging time to drink sap (Pakkala et al. 2018), and sap rows on Norway spruces Picea abies are in Sweden considered to be a very good indication of nesting pairs nearby (Artdatabanken, 2019). Other woodpeckers too, especially the Great Spotted Woodpecker Dendrocopos major, make sap rows and they also do so at least occasionally on silver birches Betula pendula in Finland (Pakkala et al. 2018). We have not found any accounts of whether Great Spotted Woodpeckers also make sap rows on birches Betula sp. in northern Sweden.
Here we use systematically recorded observations of sap rows to calculate population densities of Threetoed Woodpeckers in inland areas, below the montane forest, in Västerbotten County, northern Sweden. We do so by combining the recorded proportion of fixed routes with fresh sap rows with the average home range and the proportion of active territories per year derived from the literature. This way, we can calculate densities (pairs per km2) of Three-toed Woodpeckers without observing a single bird. For comparison, we also calculate densities based on bird observations from fixed routes (2010-2019) and correction factors for ish Bird Survey, to record the presence of trees with sap rows. There are 716 fixed routes systematically distributed over Sweden, the distance between the routes is 25 km in both north-south and west-east direction. Each route consists of eight line transects of 1 km along a 2 × 2 km square, and eight point counts of 5 minutes at each full km where all seen or heard birds and larger mammals are counted. Lines and points are two separate samples of the same routes. The route starts and ends in the southwest corner and is surveyed clockwise. The routes are surveyed once a year during a three-week period adapted to local conditions, when the activity of most breeding bird species is expected to be as high as possible. As explained above this is not the case for all species, but for most of them. In the boreal part of northern Sweden, the routes are surveyed detectability based on Finnish line transects. We do this separately for three different sections of Västerbotten County: the montane forests, the inland, and the province of Västerbotten (which makes up the eastern part of Västerbotten County) closer to the coast. Finally, we use the calculated densities to estimate total population size of Three-toed Woodpeckers in the three different parts of Västerbotten County. By summing up these three figures we eventually present an overall population size for the whole county.

FIXED ROUTES AND SAP ROW DATA COLLECTION
We selected a cluster of 14 fixed routes from the Swed-FIGURE 1. Inset: Overview, our study area within the red box (around 65°09'N, 17°13'E). Main image: Detailed map with the three different parts of Västerbotten County used in our analysis; the montane forests, the inland (below the mountain area), and the province of Västerbotten (coastal area). Points (blue and red) show the locations of fixed routes with any proportion of forest cover (n = 89). Red points are routes surveyed for sap rows (n = 14).
in June (Green et al. 2020). The selected routes are all located in the boreal coniferous forest belt in northern Sweden (see Figure 1). Active forestry has been conducted in the area since the mid-1800s, at first through selective cutting of the largest trees, later gradually more by clear-cutting (Lundmark et al. 2013).
We looked for sap rows on the first, southwesternmost, kilometer of the route whenever possible (10 of 14 routes). If the terrain did not allow the observer to cover the first kilometer of the route, we chose the second kilometer, and so on. Along the lines, the observer walked slowly and looked for sap rows within about 10 m in both directions, which means that we covered a corridor of 20 x 1000 m. When the full kilometer was done, the observer turned around and went back the same way. Doing this, we searched for sap rows in both directions, and covered an area of about 2 ha per route (20 × 1,000 m) in about 1.5-2 hours (1 km back and forth). The observer only used eyesight, no equipment such as binoculars or ladders, in order to detect sap rows. We examined the tree trunks up to about 8 meters height. All field work was carried out in the autumn of 2020 by the same person (BF). Data was recorded in a smart phone application in ArcGIS Collector. We used a map application where the routes were displayed, and with which it was possible to collect geodata in the field. When a tree with sap holes was found, the following information was recorded: tree species, estimated age of sap rows, and number of sap rows. In addition, the position of the tree was recorded with the GPS of the phone. The age of the sap rows was categorized in three classes: < 3 years, 3-7 years and > 7 years old.

AGE OF SAP ROWS
Estimating the exact age of sap rows can be challenging. On Norway spruce, the bark is usually first "peeled off " first, and then the holes are pecked. This creates a ginger area on both sides of the sap row ( Figure 2a). Sap flowing from the holes and the ginger area around the row is a sign of fresh sap rows (Figure 2b). After a couple of years, the ginger color fades and becomes quite bright. Old sap rows that have been used for a long period become partly overgrown (Figure 2d), previous studies have shown that about 20 % of new sap holes are made within the old holes (Pakkala et al. 2018). It is, however, difficult to know when they were used. If there is sap flowing in the holes, they are really fresh; if the sap is yellow or white but not flowing anymore (Figure 2e), they are still quite fresh; otherwise, such sap rows are probably older than seven years. There are examples of trees that have been used as sap trees by Three-toed Woodpeckers for more than 100 years (Ruge 1968).
On birches, the sap holes grow as the tree grows. Rows made in the same season as the survey look like narrow cuts, reminiscent of a small cut made with a knife (Figure 2c). Sap holes made a year or two ago are still quite small, but easier to see than fresh rows. The category 3-7 years is usually the easiest to see, as they are clearly visible against the white bark and the width of single holes is about 3-5 mm ( Figure 2c). Older holes are usually wider than 5 mm. It is easier to detect sap rows on younger trees with smooth bark compared to older trees with a more complex surface structure. We saw most of the sap rows at a height of 2-8 meters, but on a large, felled birch, we found several sap rows high up in the crown and reaching out on the thick branches. Such sap holes are normally not detectable from the ground. Very fresh holes and holes on older trees located higher up in the crown are the most difficult to detect.
Marks on Scots pines are similar to those on Norway spruces but are in our opinion more difficult to detect, as the color where the bark has been peeled off does not differ much from the rest of the tree. In uncertain cases, we have, regardless of tree species, erred on the side of the older category.

HOME RANGE
Reported home ranges and / or territory sizes of Threetoed Woodpeckers vary from a few tens to several hundred hectares (Glutz von Blotzheim & Bauer 1994, Pechacek 2004. The home range probably varies over the year with the smallest size during the breeding season, especially during the period with young in the nest, and the largest size during the winter. The sap rows are mainly made during the period just before nesting (Pakkala et al. 2018) but can be made from early spring to late summer (Turcek 1954, Bailey 2008. For that reason, we want to estimate the home range during the pre-nesting period of the year. As there is a shortage of studies on home range size for Threetoed Woodpeckers in Scandinavia, we used a study from Germany that assessed home ranges/territory sizes for Three-toed Woodpeckers using radio-tagged birds (Pechacek 2004). During the pre-nesting season, the home range varied between 42.6 and 381.7 ha. We chose to use the average of these two extremes, 212 ha, as the average home range during the season in focus here. The German study included 15 radio-tagged birds and the average recorded home range in that study was 115 ha. However, home ranges are probably generally larger in northern Sweden compared to the German Alps due to lower biological productivity, more intensively managed forests, and lower volume of dead wood in northern Sweden (Schmitz et al. 2015, Nilsson et al. 2020. The Swedish woodpeckers probably need a larger area to find the food they need. In order not

years). (c) A downy birch
Betula pubescens with sap holes from the current season (small, "sharp" cuts) and sap holes that are about 5 years old (larger, darker, "squares"). (d) Partially overgrown sap rows on Scots pine Pinus sylvestris. This pine has been used for a long period of time; a closer look is needed to determine the age of the latest visit to these sap rows. (e) A close-up of the same pine reveals that old sap holes have been reopened, a bark flake has also been peeled off recently. The sap/resin is visible in the holes, but is not flowing, indicating it does not have to be from current season, but definitely not from very long ago (< 3 years).

PROPORTION OF ACTIVE TERRITORIES PER YEAR
To determine the proportion of active territories in any given year of the territories found over a longer time period, we looked at data from two areas that have been carefully surveyed over several years. In a 340 km2 (270 km2 forest land) area in southern Finland, 195 territories were found between 1987and 2000(Pakkala et al. 2002. The proportion of active territories in a single year was on average 40 % in this area. In a 100 km2 (70 km2 forest land) area around Forsmark, Uppland, in central Sweden, eight territories were found between 2015 and 2019, with an average of 48 % active territories in a single year (own unpublished data). The area in Finland has relatively high densities (0.29 pairs per km2) of Three-toed Woodpeckers, while the Forsmark area is located at the southern limit of the breeding distribution and holds about 0.04 pairs per km2. We chose to use the lower of these values in our calculations, 40 %, as this was based on a larger data set from areas more similar to northern Sweden than the Forsmark data. Note however, the similarity between the two studies in the recorded proportions.

SAP ROW BASED DENSITY CALCULATIONS
We calculated densities (pairs per km2) based on sap row data as follows. Presence of fresh sap rows (< 3 years old) on a route were regarded as one active territory (one pair) during any of the last three years. If the pair had been present during every year this would together with the estimated home range of 212 ha (2.12 km2) give a density of 1/2.12 = 0.47 pairs per km2.
With on average 40 % of existing territories being active in a single year this gives an annual density of 0.47 × 0.40 = 0.19 pairs per km2. Should average proportion of occupied territories be higher than 40 %, then resulting annual densities will be higher as well.
To get overall densities we then multiplied the annual density with the found proportion of routes with fresh sap rows (< 3 years old). This means that if sap rows were found on all routes, the resulting density is 0.19 pairs per km2. If sap rows were found on any other proportion of surveyed routes, the resulting density will be lower than this.
Since Three-toed Woodpeckers are well-known for making sap rows on conifers (Ruge 1968, Glutz von Blotzheim & Bauer 1980, Hess 1983, Cramp 1985, Pakkala et al. 2018 while there are few sources on sap rows made by Three-toed Woodpeckers on deciduous trees, we chose to calculate densities based on sap rows found on conifers only (as a minimum value) and on all trees (as a maximum value).

OBSERVATION DATA
It is possible to calculate densities from line transect (fixed route) data if one knows something about detectability and can correct the observed number of birds because not all birds present will be observed at any survey. In Finland, such correction factors have been developed. Within the Finnish Bird Survey, birds along line transects are recorded within and outside of a 50 m wide belt ("the main belt", 25 m on each side of the line). From data collected in this way it is possible to calculate lateral detectability. Lateral detectability is used for adjusting for the fact that an observer detects a higher proportion of the present birds close to the observer (line) than farther away from the line. Correction factors for lateral detectability were first published by Järvinen & Väisänen (1983), and the factors were updated by Lehikoinen (2014). We have used the updated factors here.
However, in order to get closer to real densities one must also correct for not all birds present being observed even very close to the observer, so-called basic detectability. We employed correction factors for basic detectability that were calculated by comparing territory mapping data with line transect data from the same areas in Finland (Rajasärkkä 2010). In Finland, birds counted along the line transect are recorded in "pair equivalents". Thus, a singing male is counted as one pair, as is a seen female (even if no male is seen or heard), while one male and one female at the same place is counted as one pair, and two males and one female are counted as two pairs, and so on. We have chosen to be somewhat more restrictive. In cases where more than one woodpecker was observed during a single km along the fixed routes, we have counted this as a maximum of one pair anyway, even if, for example, three individual birds were recorded. We did this to avoid double-counting of birds and in the end to avoid overestimation of densities and overall population size.
We calculated densities of Three-toed Woodpeckers at three different geographical scales based on observations from the fixed routes: 1) All routes within Åsele and Lycksele lappmarker (Sorsele, Malå, Storuman, Lycksele, Vilhelmina, and Åsele municipalities) below the montane forest. This corresponds to the area where we actively surveyed sap rows. We used a larger area than our actual study area in order to get a larger data set on bird observations (n = 37). 2) All routes containing montane forest, i.e. both coniferous forest and montane birch forest (n = 22). Here we used the border for montane forest as classified by the Swedish Forest Agency (Skogsstyrelsen 2020). 3) All routes in the province of Västerbotten, representing routes relatively close (within 125 km) to the coast (n = 30).
We calculated densities for all three regions based on data from the period 2010-2019 in order both to be representative of the present situation and to be able to include enough observations. Density was calculated following Svensson (2016) as average number of recorded pairs per survey kilometer × lateral detectability [7.03] × basal detectability [1.5] = pairs per km2.

CALCULATIONS OF POPULATION SIZE
Since the fixed routes are systematically spaced all over the country, including Västerbotten County, they also cover all major habitats in a proportional way. We used this relationship to calculate the forest areas for different parts of Västerbotten County. The total area of forest land in the county is 40,680 km2 (Nilsson et al. 2020). With 22 fixed routes in the montane forest this means that almost 25 % (24.7 %) of the forest in Västerbotten County is montane forest, corresponding to an area of 10 048 km2. The equivalent forested areas in Åsele and Lycksele lappmarker below the montane forest (41.6 % of the total forest area in the county) was 16,923 km2 and in the province of Västerbotten (33.7 % of the total forest area in the county) 13,709 km2.
We then calculated the population size of Threetoed Woodpeckers for these separate parts of Västerbotten County by multiplying the observed and calculated densities (pairs per km2) with the above-mentioned areas of forest land. We calculated population size for Åsele and Lycksele lappmarker below the montane forest both based on densities from our sap row observations and on densities based on bird observations. From the montane forest and the province of Västerbotten (coastal area) we calculated population size from densities based on bird observations. Overall population size for the whole of Västerbotten County was finally calculated by summing the observation-based population estimates for the three separate parts of the county.

Results
We found a total of 98 trees with fresh sap rows made during the last three years. Fresh sap rows on conifers were found on 21 % of the routes, and fresh sap rows on birches were found on all routes (100 %). On average, we found 7 (1-21) sap trees with fresh sap rows per route (Table 1) Average number of trees with fresh sap rows per route and observed range Genomsnittligt antal träd med färska savrader per rutt och observerat intervall 0.36 (0-2) 6.6 (1-21) 7.0 (1-21)

DENSITIES OF THREE-TOED WOODPECKERS
If we assume that Three-toed Woodpeckers only make sap rows on conifers, the calculated density in our study area is 0.04 pairs per km2. If we assume that Three-toed Woodpeckers also make the sap rows on birches, the resulting density is 0.19 pairs per km2. When we used observations of birds on the fixed routes and applied the Finnish correction factors for lateral and basal detectability, the density in Åsele and Lycksele lappmarker below the montane forest (corresponding to our study area) was 0.13 pairs per km2. The area covered by montane forest held much higher densities of Three-toed Woodpeckers, 0.38 pairs per km2. In the area closer to the coast, in the province of Västerbotten, the calculated density based on observations on the fixed routes was 0.14 pairs per km2 (Table 2).

IN DIFFERENT PARTS OF VÄSTERBOTTEN COUNTY
The population size for Åsele and Lycksle lappmarker below the montane forest, calculated from densities of sap rows, was 700-3,200 pairs, depending on whether we use densities based on sap rows on conifers only (lower limit) or on all trees (upper limit).
The population size calculated from observations of birds on the fixed routes in the same region was 2,200 pairs. Based on the high densities, the montane forests seemingly hold the highest number of Three-toed Woodpeckers in the county: observations on the fixed routes result in a population size of about 3,800 pairs. Finally, we calculated that almost 1,900 pairs could occur in the province of Västerbotten (coastal area). Altogether, this gives an total population size of 7,900 pairs for the whole county, based on observations on the fixed routes (Table 2).

Discussion
The densities we calculated based on sap rows corresponded relatively well with the densities based on observed birds on the fixed routes in Åsele and Lycksele lappmarker below the montane forests. The densities based only on sap rows on coniferous trees were, however, markedly lower than those based on observed birds or those based on sap rows on all trees. It is well-known that Three-toed Woodpeckers make sap rows on conifers (Ruge 1968, Glutz von Blotzheim & Bauer 1980, Hess 1983, Cramp 1985, Pakkala et al. 2018), but we have only found one publication showing intensive use of birches for sap drinking (Bailey 2008). This study regards the North American sister species P. dorsalis, previously considered conspecific with the Eurasian species P. tridactylus. We have made own observations from the montane birch forest at Ammarnäs in Västerbotten County of Three-toed Woodpeckers making sap rows on birches, but no direct observations of the behavior in other areas. Sap rows on birches found by us so far do, however, correspond very well with the distribution of Three-toed Woodpeckers (own unpublished data), but there are of course still some uncertainties about whether all sap rows found on birches in our study area were made by this species alone.
Other woodpeckers, especially the Great Spotted Woodpecker, are also known to make sap rows, and they also do so at least occasionally on birches in Finland (Pakkala et al. 2018). In Sweden, however, we have not found any reports of sap rows on birches made by Great Spotted Woodpeckers. We have also not observed the specifically longer distances between sap holes made by Great Spotted Woodpeckers, proposed as a species characteristic by Pakkala et al. (2018). More detailed studies are needed to find out more about the making of sap rows on different tree species in northern Sweden. The similarities between densities based on sap rows and observed birds do, however, indicate that at least in the northern parts of Sweden, Three-toed Woodpeckers may primarily be responsible for the sap rows also on birches.
Presuming that most or all of the sap rows we found were made by Three-toed Woodpeckers, this also means that looking for sap rows is a more efficient way of surveying the presence of Three-toed Woodpeckers than looking for the birds themselves. We noticed that Threetoed Woodpeckers have only been observed on two of the 14 routes we searched for sap rows between 2010 and 2019, while we found fresh sap rows on all routes. It is well known that Three-toed Woodpeckers are hard to find, even in an occupied breeding territory. Therefore, using observations of sap rows could be an easier way of collecting data on the occurrence of the species. It also has the advantage that it could be done at all times of the year and also without real knowledge about the sounds made by the birds. The method could for example be used by conservationists or forest owner who want to know if Three-toed Woodpeckers occur in specific areas.
The densities based on bird observation data that we found in different parts of the county below the montane forests were relatively similar, indicating rather low but evenly distributed numbers of Threetoed Woodpeckers through a large part of the county. We could, for example, not see any signs of successively increasing densities with increasing distance from the coast. Densities in the province of Västerbotten relatively close to the coast were very similar to those further inland in Åsele and Lycksele lappmarker below the montane forest. This indicates that in the parts of the county with long-term active forestry, densities over larger areas are more or less the same. At the same time, recorded densities based on fixed route observations in the montane forests were clearly higher. The latter finding was expected, as the mountain region holds large continuous areas of intact old forest areas (Svensson et al. 2018), much less affected by forestry. Densities in the montane forests were about three times higher than those in other parts of Västerbotten County.
Another recent analysis (Svensson 2016) also used the Finnish correction factors and suggested that earlier calculation methods show reliable results for common species (> 100,000 pairs in the whole of Sweden) but usually underestimate less common species (< 100,000 pairs in Sweden), such as Three-toed Woodpeckers. Svensson (2016) suggested that the size of the Swedish population of Three-toed Woodpeckers could be five times larger than the one reported in Ottosson et al. (2012). Our calculations also result in higher estimates for Västerbotten County and the province of Västerbotten than those in Ottosson et al. (2012). Even though we used a more conservative approach when interpreting the number of observed pair equivalents per km on the fixed routes, our results indicate a regional population size that could be about three to five (2.6-4.9) times larger than the earlier estimate for the county and almost four (3.7) times larger for the province. For Åsele and Lycksele lappmarker, including the montane forest, our estimate is 3.5 times higher than the estimate by Ottosson et al. (2012).
The new, partly higher estimates presented here should not be interpreted as a recent increase in population size of Three-toed Woodpeckers in Västerbotten County. The difference between ours and earlier estimates instead depends on different ways of estimating the population size. It remains to be shown which estimate is closest to the true number, but we argue that the methods and calculations that we have used here are more likely to reflect reality than the partly simpler calculations made by Ottosson et al. (2012).
Interestingly, our estimates of regional population size are very close to what can be calculated as the possible carrying capacity for the county, based on the Three-toed Woodpecker's requirements for dead wood. An analysis of data from both Sweden and Switzerland concluded that the threshold value for a 95-percent probability of Three-toed Woodpecker occurrence is 15 m3 of standing dead wood per ha within an area of 1 km2 (Bütler et al. 2004). Using this threshold together with data on general levels of standing dead wood in Västerbotten County (approximately 11,000,000 m3; Skogsdata, 2020) indicate a regional population capacity of 7,300 of Three-toed Woodpecker pairs.
We would like to encourage more detailed studies on both the occurrence of sap rows and of the woodpeckers themselves. This can be done, for example, through more surveys for sap rows also in parts of the county closer to the coast, as well as in forests in the mountain area. We believe that home range sizes in the mountain area are smaller compared to our study area, which is more affected by forestry. In order to find out more about woodpecker presence at a general level, the fixed routes could be used for more detailed investigations during the time of year when drumming activity is high. Using playback of species-specific drumming sounds in April, for example, could yield interesting data to be used in comparison with data from the ordinary surveys of the routes as well as findings of sap rows.
In conclusion, we find that systematic observations of sap rows can be used for calculating credible densities of Three-toed Woodpeckers. The fact that we found sap rows on all surveyed routes indicates that the Three-toed Woodpecker is well spread, albeit sparsely distributed, throughout the study area. Our density calculations based on sap rows are in good agreement with other methods, such as those based on observations of birds. The advantages of surveying sap rows instead of actual Three-toed Woodpeckers are simplicity and speed. Our method can be used by basically anyone, at any time of the year, and quickly provides densities of Three-toed Woodpeckers in any geographical area. The method can thus be valuable for both conservation and forestry.