Predation of boreal owl nests by pine martens in the boreal forest does not vary as predicted by the alternative prey hypothesis

The alternative prey hypothesis (APH) states that temporally synchronous population fluctuations of microtine rodents and other small herbivores are caused by generalist predators that show functional and numerical responses to the abundance of microtines. This would lead to an increased predation of alternative prey in the low phase of the microtine population fluctuations. One candidate for such a predator is the tree-climbing pine marten (Martes martes), which includes bird eggs in its diet, among them eggs of the cavity-nesting boreal owl (Aegolius funereus). I used long-term data to test whether pine marten predation of boreal owl eggs in nest boxes varied as predicted by the APH. The probability of predation of owl nests situated < 45 km from a site where microtines were trapped in spring during four decades increased with microtine trapping index, which is opposite to the prediction from the APH. As the data set was limited to one nest per box, I extended it spatially and temporally using the clutch size of each boreal owl nest as a proxy for the actual microtine abundance at the site. The probability of nest predation increased with clutch size. However, the effects of microtine index and owl clutch size became non-significant when I controlled for habitat, and in particular cavity age, which had an overriding effect. The increase in predation probability with cavity age suggests that the long-term spatial memory of pine marten is an important factor in the pattern of its nest predation in tree cavities. Supplementary Information The online version contains supplementary material available at 10.1007/s00442-022-05149-0.


Table S1
Number of boreal owl nests distributed on number of nest boxes used. _____________________________________________________________________

All nests
Nests situated < 45 km from trapping site (1970 -2018) in trapping years (1977-78 and 1981-2018 (total and predated) in the whole study area and in the area < 45 km from the microtine rodent trapping site, when only one nest per nest box is included. CavAge is the mean number of nesting seasons since the nest box was installed. DistEdge is the mean distance from the actual nest box to the edge between old forest and clear-cuts and other open habitats, with negative values for boxes in forest and positive for boxes in open habitats. MicIndex is microtine rodent trapping index in spring. For nest boxes with more than one nest during the study, only the first nest is included, separately for the whole study area during the whole study period, and for the area <45 km from the microtine rodent trapping site during the years when microtine rodents were trapped. Nest situated < 45 km from the microtine trapping site in 1970-76 and 1979-80

Table S4
Parameter estimates for the probability that the predator of a predated boreal owl nest was scored as pine marten rather than as an unidentified predator as an effect of a fixed variable, corrected for the random effect of year. a Fixed effect is the microtine rodent trapping index in spring (60 nests, 25 years). b Fixed effect is the change in microtine rodent abundance from the previous spring to the current spring (51 nests, 23 years). c Fixed effect is the size of the actual boreal owl clutch predated or the size of the clutch of the nearest neighbor, taken as a proxy for the microtine rodent abundance (

Table S5
Model selection based on Akaike´s information criterion (AIC) to determine which variables influenced the probability of predation of a boreal owl nest situated < 45 km from the microtine rodent trapping site in a year when microtine rodents were trapped (n = 217), with year as random effect (n = 35). MicRod (range 0.10 -8.04) denotes the microtine rodent spring trapping index (see text for details). Cav (range 1 -15) denotes the time since the actual nest box was installed (see text for details). Dist (range -100 -100) denotes the shortest distance from the nest box to the nearest edge between forest and an open habitat (see text for details). Models are sorted by their degree of support from the data, as indicated by the AICc weight (w  Table S6 Model selection based on Akaike´s information criterion for small sample size (AICc) to determine which variables influenced the probability of predation of a boreal owl nest (n = 217) situated < 45 km from the microtine rodent trapping site in a year when microtine rodents were trapped (1977-78 and 1981-2018), with year as random effect (n = 35). BankVole (range 0.00 -3.10) denotes the bank vole spring trapping index (see text for details). Cav (range 1 -15) denotes the time since the actual nest box was installed, with 1 denoting the first nesting season the box was available (see text). Dist (range -100 -100) denotes the shortest distance from the nest box to the nearest edge between forest and an open habitat, with zero for boxes situated at the edge between forest and open habitat, a negative value for boxes situated within a forest stand, and a positive value for boxes situated in a tree in a clear-cut or a bog (see text for details). Models are sorted by their degree of support from the data, as indicated by the AICc weight (w). Models with DAICc ≤ 2.0 from the best model are shown in bold. ______________________________________________________________________

Variables
LogLikelihood df AICc DAICc AICc w ______________________________________________________________________  Table S7 Model selection based on Akaike´s information criterion for small sample size (AICc) to determine which variables influenced the probability of predation of a boreal owl nest (n = 217) situated < 45 km from the microtine rodent trapping site in a year when microtine rodents were trapped (1977-78 and 1981-2018), with year as random effect (n = 35). Micr (range 0.00 -2.77) denotes the Microtus vole spring trapping index (see text for details). Cav (range 1 -15) denotes the time since the actual nest box was installed, with 1 denoting the first nesting season the box was available (see text). Dist (range -100 -100) denotes the shortest distance from the nest box to the nearest edge between forest and an open habitat, with zero for boxes situated at the edge between forest and open habitat, a negative value for boxes situated within a forest stand, and a positive value for boxes situated in a tree in a clear-cut or a bog (see text for details). Models are sorted by their degree of support from the data, as indicated by the AICc weight (w). Models with DAICc ≤ 2.0 from the best model are shown in bold.

Variables
LogLikelihood df AICc DAICc AICc w ______________________________________________________________________  Table S8 Model selection based on Akaike´s information criterion for small sample size (AICc) to determine which variables influenced the probability of predation of a boreal owl nest (n = 217) situated < 45 km from the microtine rodent trapping site in a year when microtine rodents were trapped (1977-78 and 1981-2018), with year as random effect (n = 35). WL (range 0.00 -2.86) denotes the wood lemming spring trapping index (see text for details). Cav (range 1 -15) denotes the time since the actual nest box was installed, with 1 denoting the first nesting season the box was available (see text). Dist (range -100 -100) denotes the shortest distance from the nest box to the nearest edge between forest and an open habitat, with zero for boxes situated at the edge between forest and open habitat, a negative value for boxes situated within a forest stand, and a positive value for boxes situated in a tree in a clear-cut or a bog (see text for details). Models are sorted by their degree of support from the data, as indicated by the AICc weight (w  Tables S6-S8 for the probability of predation of a boreal owl nest situated < 45 km from the microtine rodent trapping site in a year when microtine rodents were trapped (n = 217), corrected for the random effect of year (n = 35). a Bank vole (Table S6). b Microtus voles (Table S7). c Wood lemming (Table S8)

Table S10
Model selection based on Akaike´s information criterion (AIC) to determine which variables influenced the probability of predation of a boreal owl nest situated < 45 km from the microtine rodent trapping site in a year when microtine rodents had also been trapped in the previous year (n = 187), with year (n = 33) as random variable. MicrotineCh (range -3.53 -5.03) denotes the change in the microtine rodent trapping index from the previous spring to the current spring (see text for details). Cav (range 1 -15) denotes the time since the actual nest box was installed (see text for details). Dist (range -100 -100) denotes the shortest distance from the nest box to the nearest edge between forest and an open habitat (see text for details). In all models Year in included as a random variable. Models are sorted by their degree of support from the data, as indicated by the AICc weight (w  Table S11 Model selection based on Akaike´s information criterion for small sample size (AICc) to determine which variables influenced the probability of predation of a boreal owl nest (n = 187) situated < 45 km from the microtine rodent trapping site in a year when microtine rodents had been trapped the previous year (1978 and 1982-2018), with year (n = 33) as random variable. BankVoleCh (range -1.63 -2.64) denotes the change in the bank vole trapping index from the previous spring to the current spring (see text for details). Cav (range 1 -15) denotes the time since the actual nest box was installed, with 1 denoting the first nesting season the box was available (see text). Dist (range -100 -100) denotes the shortest distance from the nest box to the nearest edge between forest and an open habitat, with zero for boxes situated at the edge between forest and open habitat, a negative value for boxes situated within a forest stand, and a positive value for boxes situated in a tree in a clear-cut or a bog (see text for details). In all models Year in included as a random variable. Models are sorted by their degree of support from the data, as indicated by the AICc weight (w). Models with DAICc ≤ 2.0 from the best model are shown in bold. ______________________________________________________________________

Variables
LogLikelihood df AICc DAICc AICc w ______________________________________________________________________  Table S12 Model selection based on Akaike´s information criterion for small sample size (AICc) to determine which variables influenced the probability of predation of a boreal owl nest (n = 187) situated < 45 km from the microtine rodent trapping site in a year when microtine rodents had been trapped the previous year (1978 and 1982-2018), with year (n = 33) as random variable. MicVolesCh (range -1.81 -1.74) denotes the change in the Microtus vole trapping index from the previous spring to the current spring (see text for details). Cav (range 1 -15) denotes the time since the actual nest box was installed, with 1 denoting the first nesting season the box was available (see text). Dist (range -100 -100) denotes the shortest distance from the nest box to the nearest edge between forest and an open habitat, with zero for boxes situated at the edge between forest and open habitat, a negative value for boxes situated within a forest stand, and a positive value for boxes situated in a tree in a clear-cut or a bog (see text for details). In all models Year in included as a random variable. Models are sorted by their degree of support from the data, as indicated by the AICc weight (w). Models with DAICc ≤ 2.0 from the best model are shown in bold. ______________________________________________________________________

Variables
LogLikelihood df AICc DAICc AICc w ______________________________________________________________________  Table S13 Model selection based on Akaike´s information criterion for small sample size (AICc) to determine which variables influenced the probability of predation of a boreal owl nest (n = 187) situated < 45 km from the microtine rodent trapping site in a year when microtine rodents had been trapped the previous year (1978 and 1982-2018), with year (n = 33) as random variable. WlCh (range -0.45 -2.68) denotes the change in the wood lemming trapping index from the previous spring to the current spring (see text for details). Cav (range 1 -15) denotes the time since the actual nest box was installed, with 1 denoting the first nesting season the box was available (see text). Dist (range -100 -100) denotes the shortest distance from the nest box to the nearest edge between forest and an open habitat, with zero for boxes situated at the edge between forest and open habitat, a negative value for boxes situated within a forest stand, and a positive value for boxes situated in a tree in a clear-cut or a bog (see text for details). In all models Year in included as a random variable. Models are sorted by their degree of support from the data, as indicated by the AICc weight (w). Models with DAICc ≤ 2.0 from the best model are shown in bold. ______________________________________________________________________

Variables
LogLikelihood df AICc DAICc AICc w ______________________________________________________________________  Table S14 Parameter estimates from the highest-ranked models that included the microtine species in Tables S11-S13, and from the most parsimonious models that included the microtine species and the distance to forest edge in Tables S11-S13, for the probability of predation of a boreal owl nest (n = 187) situated < 45 km from the microtine rodent trapping site in a year when microtine rodents had been trapped the previous year, corrected for the random effect of year (n = 33). a The highest-ranked model that included bank vole (Table S11). b The most parsimonious model that included bank vole and distance to forest edge (Table S11). c The highest-ranked model that included Microtus voles (Table S12). d The most parsimonious model that included Microtus voles and distance to forest edge (Table S12). e The highest-ranked model that included wood lemming (Table S13). f The most parsimonious model that included wood lemming and distance to forest edge (

Table S15
Model selection based on Akaike´s information criterion for small sample size (AICc) to determine which variables influenced the probability of predation of a boreal owl nest (n = 214) situated < 45 km from the microtine rodent trapping site in a year when microtine rodents were trapped (1977-78 and 1981-2018), with year as random effect (n = 33). Clutch (range 3-10) denotes the clutch size of the actual nest or its nearest neighbor in the same year if the actual nest was predated before the first nest check (see text for details). Cav (range 1 -15) denotes the time since the actual nest box was installed, with 1 denoting the first nesting season the box was available (see text). Dist (range -100 -100) denotes the shortest distance from the nest box to the nearest edge between forest and an open habitat, with zero for boxes situated at the edge between forest and open habitat, a negative value for boxes situated within a forest stand, and a positive value for boxes situated in a tree in a clear-cut or a bog (see text for details). Models are sorted by their degree of support from the data, as indicated by the AICc weight (w). Models with DAICc ≤ 2.0 from the best model are shown in bold. ______________________________________________________________________

Variables
LogLikelihood df AICc DAICc AICc w ______________________________________________________________________  Table S15 for the probability of predation of a boreal owl nest (n = 214) situated < 45 km from the microtine rodent trapping site in the years when microtine rodent were trapped, corrected for the random effect of year (n = 33

Table S17
Model selection based on Akaike´s information criterion for small sample size (AICc) to determine which variables influenced the probability of predation of a boreal owl nest (n = 337), with year as random effect (n = 40). Clutch (range 2 -10) denotes the clutch size of the actual boreal owl nest, or the clutch size of the nearest recorded neigbor nest in the same year in case the actual nest was predated prior to the first nest check, and is a proxy for the microtine rodent abundance at the actual nest (see text for details). Cav (range 1 -15) denotes the time since the actual nest box was installed, with 1 denoting the first nesting season the box was available (see text). Dist (range -100 -100) denotes the shortest distance from the nest box to the nearest edge between forest and an open habitat, with zero for boxes situated at the edge between forest and open habitat, a negative value for boxes situated within a forest stand, and a positive value for boxes situated in a tree in a clear-cut or a bog (see text for details). Models are sorted by their degree of support from the data, as indicated by the AICc weight (w

Table S19
Model selection based on Akaike´s information criterion for small sample size (AICc) to determine which variables influenced the probability of predation of a boreal owl nest (n = 77) in a nest box the first season after it was installed (cavity age = 1) situated < 45 km from the microtine rodent trapping site in a year when microtine rodents were trapped (1977-78 and 1981-2018), with year as random effect (n = 23). MicRod (range 0.18 -4.02) denotes the microtine rodent spring trapping index (see text for details). Dist (range -100 -100) denotes the shortest distance from the nest box to the nearest edge between forest and an open habitat, with zero for boxes situated at the edge between forest and open habitat, a negative value for boxes situated within a forest stand, and a positive value for boxes situated in a tree in a clear-cut or on a bog were as (see text for details). Models are sorted by their degree of support from the data, as indicated by the AICc weight (w). Models with DAICc ≤ 2.0 from the best model are shown in bold. ______________________________________________________________________

Variables
LogLikelihood df AICc DAICc AICc w ______________________________________________________________________  Table S20 Model selection based on Akaike´s information criterion for small sample size (AICc) to determine which variables influenced the probability of predation of a boreal owl nest (n = 71) in a nest box the first season after it was installed (cavity age = 1) situated < 45 km from the microtine rodent trapping site in a year when microtine rodents had been trapped the previous year (1978 and 1982-2018), with year (n = 21) as random variable. MicrotineCh (range -2.07 -1.34) denotes the change in the microtine rodent trapping index from the previous spring to the current spring (see text for details). DistEdge (range -100 -100) denotes the shortest distance from the nest box to the nearest edge between forest and an open habitat, with zero for boxes situated at the edge between forest and open habitat, a negative value for boxes situated within a forest stand, and a positive value for boxes situated in a tree in a clear-cut or on a bog were as (see text for details). In all models Year in included as a random variable. Explanatory variables are standardized. Models are sorted by their degree of support from the data, as indicated by the AICc weight (w). Models with DAICc ≤ 2.0 from the best model are shown in bold. ______________________________________________________________________

Variables
LogLikelihood df AICc DAICc AICc w ______________________________________________________________________  Table S21 Model selection based on Akaike´s information criterion for small sample size (AICc) to determine which variables influenced the probability of predation of a boreal owl nest (n = 151) in a nest box the first season after it was installed (cavity age = 1), with year as random effect (n = 31). Clutch (range 2-8) denotes the clutch size of the actual nest or its nearest neighbor in the same year if the actual nest was predated before the first nest check (see text for details). Dist (range -100 -100) denotes the shortest distance from the nest box to the nearest edge between forest and an open habitat, with zero for boxes situated at the edge between forest and open habitat, a negative value for boxes situated within a forest stand, and a positive value for boxes situated in a tree in a clear-cut or a bog (see text for details). Models are sorted by their degree of support from the data, as indicated by the AICc weight (w). Models with DAICc ≤ 2.0 from the best model are shown in bold. ______________________________________________________________________

Variables
LogLikelihood df AICc DAICc AICc w ______________________________________________________________________  Table S22 Parameter estimates from the highest-ranked models that included microtine rodent abundance alone (a) or in combination with distance to forest edge (b) in Table  S19 for the probability of predation of a boreal owl nest (n = 77) in a nest box the first season after it was installed (cavity age = 1), corrected for the random effect of year (n = 23

Table S23
Parameter estimates from the models that included change in microtine rodent abundance alone (a) or in combination with distance to forest edge (b) in Table  S20 for the probability of predation of a boreal owl nest (n = 71) in a nest box the first season after it was installed (cavity age = 1), corrected for the random effect of year (n = 21

Table S24
Parameter estimates from the models that included boreal owl clutch size alone (a) or in combination with distance to forest edge (b) in Table S21 for the probability of predation of a boreal owl nest (n = 151) in a nest box the first season after it was installed (cavity age = 1), corrected for the random effect of year (n = 31