Calf/female ratio and population dynamics of wild forest reindeer in relation to wolf and moose abundances in a managed European ecosystem.

BACKGROUND: The alternative prey hypothesis describes the mechanism for apparent competition whereby the mortality of the secondary prey species increases (and population size decreases decreases) by the increased predation by the shared predator if the population size of the primary prey decreases. Apparent competition is a process where the abundance of two co-existing prey species are negatively associated because they share a mutual predator, which negatively affects the abundance of both prey Here, we examined whether alternative prey and/or apparent competition hypothesis can explain the population dynamics and reproductive output of the secondary prey, wild forest reindeer (Rangifer tarandus fennicus) in Finland, in a predator-prey community in which moose (Alces alces) is the primary prey and the wolf (Canis lupus) is the generalist predator.
METHODS: We examined a 22-year time series (1996-2017) to determine how the population size and the calf/female ratio of wild forest reindeer in Eastern Finland were related to the abundances of wolf and moose. Only moose population size was regulated by hunting. Summer predation of wolves on reindeer focuses on calves. We used least squares regression (GLS) models (for handling autocorrelated error structures and resulting pseudo-R2s) and generalized linear mixed (GLMs) models (for avoidance of negative predictions) to determine the relationships between abundances. We performed linear and general linear models for the calf/female ratio of reindeer. RESULTS AND SYNTHESIS: The trends in reindeer population size and moose abundance were almost identical: an increase during the first years and then a decrease until the last years of our study period. Wolf population size in turn did not show long-term trends. Change in reindeer population size between consecutive winters was related positively to the calf/female ratio. The calf/female ratio was negatively related to wolf population size, but the reindeer population size was related to the wolf population only when moose abundance was entered as another independent variable. The wolf population was not related to moose abundance even though it is likely to consist the majority of the prey biomass. Because reindeer and moose populations were positively associated, our results seemed to support the alternative prey hypothesis more than the apparent competition hypothesis. However, these two hypotheses are not mutually exclusive and the primary mechanism is difficult to distinguish as the system is heavily managed by moose hunting. The recovery of wild forest reindeer in eastern Finland probably requires ecosystem management involving both habitat restoration and control of species abundances.

Introduction

When sympatric prey populations share a common generalist predator their populations can be differentially related to the abundance of the predator. The two major theoretical concepts that predict the impact of the shared predator on its prey populations are the alternative prey hypothesis [1, 2] and the apparent competition hypothesis [3, 4]. The alternative prey hypothesis describes a mechanism where the predator functionally responds to the relative abundance of different prey species by shifting its diet between different prey species. The apparent competition hypothesis refers to the process where the predator population size positively responds to moose density that is likely to be the main determinant of prey biomass, and predicts that the consequent higher predator density increases more predation on the secondary prey.

Human-induced changes in the environment may differentially favor different prey species which may result in increase imbalances in predation on sympatric prey species. Habitat mediated apparent competition following long-term human-induced modifications is a complex task that has been rarely achieved for large mammals [5]. Well known example of such a suggested habitat change-mediated increase in predation rates is the declining population of threatened woodland caribou (Rangifer tarandus caribou) in North America [6-8]. Most woodland caribou populations are declining, and extirpation is ongoing [9-11]. The possible causes for the declines in caribou populations owing to predation are often based on the idea of apparent competition whereby anthropogenic changes in the environment favor other ungulate prey species, such as moose (Alces alces), which in turn boosts predator population growth and increase predation on caribou [5–9, 11, 12]. Calf recruitment rates in one woodland caribou population was negatively related to coyote (Canis latrans) abundance which was positively correlated with moose abundance [5].

The active control of species abundances might be a reasonable means to at least stop the decline in populations and provide more time for habitat restoration [8, 9, 13, 14]. Increasing the harvesting of predators’ primary prey could decrease predator population size and thereby predation on caribou [9]. Combinations of treatments encompassing reductions in both predators and overabundant prey have produced the highest population growth rates [9]. Predator control is an option that might, however, be difficult to justify because large carnivores are iconic animals that often exist as small, threatened populations and are thought to provide ecosystem services [15, 16]. Ecosystem management plans aimed at the recovery of boreal wild reindeer and caribou might require several concurrent actions to yield concrete results. Mature forests are key habitats for forest-dwelling Rangifer [12, 17–19], but habitat management alone might be insufficient because the restoration of key habitats may take too long to decrease the risk of extinction; thus, management plans relying only on habitat protection and restoration will likely fail [9].

In this study we examine the size of recently recolonized population of European wild forest reindeer (R. t. fennicus) in relation to adundances of grey wolf (Canis lupus) and wolves’ primary prey, the moose. Ecology of wild forest reindeer is largely similar to that of woodland caribou [17]. Our study area resembles North American wolf-caribou ecosystem outside protected areas where logging is extensive and moose the primary prey of wolves. Wild forest reindeer were once distributed across the boreal coniferous zone in Europe but now are present only patchily, with a total population of approximately 10,000 animals, of which approximately only 2,300 exist in two populations in Finland [18, 20]. We have documented that the calf/female ratio in this particular population has formerly been strongly related to wolf population size using 10 years shorter time period [21].

We considered the relationships between wild reindeer, wolf and moose in the light of hypotheses of alternative prey [1, 2, 22–24] and apparent competition [4, 8, 25]. We assumed that predation by wolves influences the survival of reindeer calves in summer [21, 26, 27]. Whenever the alternative prey hypothesis accounts more we predict positive correlations of reindeer population size and the calf/female ratio in reindeer with the abundance of moose.

If the apparent competition hypothesis is the main driver, the abundance of the wolf population should be related positively to moose density which would bring about a decrease in the calf/female ratio in reindeer and, consequently, a decline in the reindeer population size. In our study area the moose population size is regulated by recreational hunting to limit browsing damage caused by moose. This may reduce apparent competition between reindeer and moose.

Methods

Study area

The study area is a ca. 6,000 km2 area in east-central Finland (Fig 1). The region is dominated by highly managed productive boreal forests [28, 29]. Approximately 90% of the land area is covered by forests in which the main tree species are Scots pine Pinus sylvestris, Norway spruce Picea abies and two birch species (Betula pubescens and Betula pendula). The topography is flat, with the highest hills measuring 270 m above sea level. The terrain is characterized by lakes and peat bogs. Human settlements and high-traffic roads are scarce, but isolated houses and low-traffic roads are widespread in the study area. Other large carnivores that are known to kill wild reindeer are the brown bear (Ursus arctos), Eurasian lynx (Lynx lynx), and wolverine (Gulo gulo, A. Paasivaara unpublished data). The minimum brown bear population density in 2006 was estimated to be 16 bears/1000 km2. Only wolf and brown bears prey on moose.

Fig 1

Territory boundaries of GPS-collared wolves from winter 2010–2011, road network and the location the study area.

Wild forest reindeer populations rebounded our study area during the early 1960s after an absence lasting approximately 40 years [17]. The return of reindeer resulted from the expansion of reindeer populations from Russia. Wolves returned as permanent breeders in the mid-1990s [30]. Moose are the primary prey of gray wolves (Canis lupus) in European boreal forest ecosystems [29, 31, 32]. Moose population densities increased substantially during the 1970s largely due to widespread clearcutting, which created more habitats favored by moose [33]. In our study area, however, winter densities of the moose population were relatively low; in 2000–2015, the population densities of moose varied between 0.17 and 0.36 moose/km2 (Pusenius et al., unpublished data). The moose population size was limited by harvest to control browsing damage to forestry [34-36] and the number of traffic collisions [37]. Wild reindeer are protected from hunting. A few legal wolf removals have been conducted. Poaching has presumably had a large impact on the wolf population in eastern Finland [38].

Data

We examined a 22-year time period (1996–2017). The population size of wild reindeer was assessed through yearly or biyearly total population counts by helicopter during late winter when reindeer are gathered up in their winter ranges [20]. For years when reindeer were not counted (1997,2002,2004), the population size was estimated as the mean of the population size before and after these years. Wolf population monitoring was based on a combination of snow-tracking of collared animals by experts, voluntary observations and genetic analyses [29, 30, 39]. Snow tracking was conducted to assess pack sizes. Wolf population estimates are for early winter. The moose abundance index is based on records by hunters. Moose hunting clubs are operating within their own territory of approximately 5 000 ha. Each club records the number of moose observations per hunting day during the autumn hunting season (from the last Saturday of September to the end of December) and the resultant index provides the annual mean number of moose seen per hunting day for our study area. We used this index because moose density was not estimated for years 1996–1999.

Numerical calf/female ratios of wild reindeer were based on field observations made by professional field technicians during September-November after the season of highest calf mortality, the first 80 days after the birth of the calf [21]. When examining relationships without lags, we related the calf/female ratio in reindeer to the previous winter’s reindeer population size for density dependence, and to the wolf population estimate and moose abundance index for the same year. In models where the reindeer population size was an independent variable, we used the previous year’s wolf population and moose abundance as independent variables, the time difference being 3–4 months with wolf and 3–5 months with moose.

Statistical analysis

The data consisted of a time series of three wildlife populations in eastern Finland: population sizes for wild forest reindeer and wolves, and abundance index for moose. The aim of the study was to model the relationships between these populations without any lag effects between the abundances and with a one-year lag in the adundances. In addition, the relationship of the calf/female ratio to the wolf population was modeled with and without moose abundance as an independent variable. Interaction terms between wolf and moose abundances were tested to evaluate indication that wolf predation on reindeer might be related to moose abundance.

Strong correlations also existed between the abundances of reindeer and moose. Autocorrelations restricted the use of the modeling methods to only those methods where the autocorrelation could be estimated. The need to consider the autocorrelation was checked by using ACF plots and partial ACF plots, and by checking the residuals and how strongly they were correlated in time (ACF plots of the residuals). The Durbin-Watson test was used as an additional measure of autocorrelation of errors. The response variables were 1) reindeer (individuals), 2) wolves (individuals) and (3) the calf/female ratio in reindeer.

Population abundance models were performed using both generalized least squares (GLS) and generalized linear mixed (GLM) models. GLS could handle the auto-correlated error structures, and provide the reasonable pseudo-R2 -values [40, 41]. On the other hand, Poisson models could not give the negative predictions otherwise to the GLS-models. Therefore prediction plots illustrating the effects of the explanatory variables were computed using the GLM-models [42].

The response variables in the models for the wild reindeer population size were log-transformed to normalize the distribution. The model for the wolf population did not require any log-transformation of the response variables; the transformation was found to weaken the distribution of residuals. Generalized least squares models provided AICs (Akaike Information Criterion [43]) that was used for comparing models with different autoregressive orders.

The need for testing autocorrelations was based on ANOVA using maximum likelihood estimation for alternative models. The final model was computed using restricted maximum likelihood (REML). The goal was to use AR orders (NULL, AR1, AR2) that would minimize AIC. In addition, the possible need for moving average parameters (MAs) was tested, but no MAs increased the model fit.

Because the response variables of the wildlife populations represented count data, Poisson or negative binomial models could have beeen more appropriate to the responses. However, in R a generalized linear model (GLM, using R-function glm) without any random factor did not allow the use of autoregressive correlation structure for the error term, but the function glmmPQL did. The function glmmPQL with the Poisson family and an estimation the overdispersion were used in the “reference” (alternative) analysis for the GLS-models by building a “pseudo” random factor (one group). The variance of the “random effect” was computed (near zero) and the “residual” in the model output described the square-root of the dispersion parameter illustrating the overdispersion in the models. The results of these models are published in Table 2 in addition to the results of the GLS-models. The results with most of the models were close to the GLS results, and none of the interpretations of the results changed.

Table 2

Statistics for the independent variables and the adjusted r-squared values in three beta regression models evaluating the relationship between the calf/female ratio, the wolf and reindeer population, and the moose abundance in eastern Finland from 1996–2017.

Independent variables	Estimate	Standard error	z	P	Pseudo- R2
Wolf population	-0.034	0.006	-5.460	< 0.001	0.565
Wolf population	-0.031	0.006	-4.930	< 0 001
Reindeer population	3.075e-4	2.156e-4	1.426	0.154	0.606
Wolf population	-0.034	0.006	-5.452	< 0.001
Moose abundance	0.042	0.123	0.341	0.733	0.568

In some of the models, the temporal autocorrelation was so strong that the second-order autocorrelation was needed to obtain the sufficient standard errors and p values.The autocorrelated error (ε) can be described for the first-order auto-regressive process (AR(1)) as follows: where the random shocks ν are assumed to be Gaussian white noise and ∅ is the estimated first-order autoregressive coefficient between the two adjacent error terms.

The second-order autocorrelated error (AR(2)) can be described as follows: where ∅1 and ∅2 are the autoregressive coefficients for the first and second orders.

More information about the GLS regression and autocorrelated variance-covariance matrix was provided by Fox and Weisberg [40, 41]. The generalized linear Poisson models (estimated using PQL) were computed using the R statistical environment and the R package MASS [44]. These quasi-likelihood models did not provide AICs for model comparison. Generalized least squares (GLS) regression was computed using the R package nlme [45]. The beta-regression models with logit link function were computed and its function betareg [42]. Model predictions were computed using the R-package effects [46] The pseudo-R2 -values for the GLS models were computed using the R-package rcompanion. All the other computations were performed in the R statistical environment [47, R Core Team 2018].

The models for the calf/female ratio were performed using beta-regression which was the most reasonable choice to model the ratio. Betareg did not allow the use of autocorrelated error term in the model. However, the autocorrelation was not significant in the time series of the calf/female ratio (Durbin-Watson statistics = 1.754, p value = 0.203).

Results

Population trends

The wild forest reindeer population size increased from 1996 through 2001 and then rapidly decreased. The decrease became more moderate from 2008 onwards, but no signs of recovery existed (Fig 2). The trends in the moose abundance index were almost identical to the trends in reindeer population size but unlike the reindeer population size, they showed a decrease up to the end of the study period (Fig 2). Wolves returned by mid-1990, and their numbers increased rapidly from 1996–2001 but fluctuated from 2001 onwards (Fig 2). The reindeer population size and moose abundance index were more temporally autocorrelated than wolf population size (Fig 2).

Fig 2

Population sizes of wild forest reindeer and wolves, the abundance index of moose and calves/females ratio in reindeer with temporal autocorrelation functions in eastern Finland from 1996–2017.

Relationships between species abundances

Reindeer

Reindeer population size was not related to wolf abundance in models (GLSs or GLMs) where wolves were the only independent variable (Table 1), but models where wolf and moose populations were entered as independent variables, reindeer population size was negatively related to the wolf population size and positively related to the moose abundance (Fig 3). In models with a one-year lag, the reindeer population was not related to the wolf population alone but was related to both the wolf and moose populations in a model where both were entered as independent variables (Table 1). Akaike Information Criterion [39] could not be used for comparing models with one or two independent variables because of autoregression parameter in models with two independent variables and only pseudo R-squared figures could be calculated for the GLS models; however, based on the p values, the models with two independent variables appeared to fit better than the models with wolf population only (Table 1).

Table 1

Student t-values and probabilities for generalized least squares (GLS) models and alternative GLMM-models (in parenthesis) for wild forest reindeer population size, the moose abundance index and the wolf population in eastern Finland from 1996–2017.

Based on residual autocorrelation tests, AR correlation structures were used in models if the residuals were autocorrelated. Cox & Snell pseudo R2 were computed for the GLS-models.

Dependent variable	Independent variable(s)	t	P	R2
Reindeer without lag (llog-normal gls)	Wolf population	-0.466	0.647	0.010
		(-0.803)	(0.431)
Reindeer without lag	Wolf population	-3.584	0.002	0.878
		(-3.298)	(0.004)
	Moose abundance	10.934	<0.001
		(10.910)	(<0.001)
Reindeer, one-year lag	Wolf population	0.013	0.989	-0.001
		(0.742)	(0.467)
Reindeer, one-year lag	Wolf population	-3.863	0.001	0.782
		(-3.885)	(0.001)
	Moose abundance	6.703	<0.001
		(6.728)	(<0.001)
Wolf without lag	Moose abundance	0.744	0.466	0.003
		(0.584)	(0.566)
Wolf without lag	Reindeer population	-0.505	0.619	0.000
		(-0.774)	(0.448)
Wolf, one-year lag	Moose abundance	0.023	0.982	-0.011
		(0.063)	(0.950)
Wolf, one-year lag	Reindeer population	-0.279	0.783	-0.015
		(-0.475)	(0.640)

Fig 3

Relationship of the calf/female ratio of wild forest reindeer to the wolf population in eastern Finland from 1996–2017 in a model where the wolf population was the only independent variable.

Wolf

In a model without a lag, the wolf population size was positively related to moose abundance and negatively related to reindeer population size (Table 1, Fig 4). In a model where the wolf population was the dependent variable with a one-year lag, neither the moose nor reindeer population was significantly related to the wolf population (Table 1). Reindeer population size was not significantly related to two-way interaction term wolf population size*moose abundance index (p value > 0.10).

Fig 4

Relationships of wolf population (a) and moose abundance (b) to wild forest reindeer population in eastern Finland from 1996–2017 in models in which another sympatric ungulate was treated as another independent variable.

Calf/female ratios of wild forest reindeer

For reindeer, the yearly calf/female ratios from 1996–2001 were higher than those later in our study period (Fig 2). The annual growth rate of the wild forest reindeer population size (Y) was related to the calves/females ratio (X) in the linear model Y = -0.16 + 0.473 * X, t = 2.57, p = 0.020).

The beta-regression analyses showed that the calf/female ratio of reindeer was related to wolf population size in a highly significant fashion (Table 2, Fig 5). In a model where the population size of reindeer was entered as another independent variable for potential density dependence, the calf/female ratio was related to the wolf population size in a highly significant fashion, while no relationship between the ratio and reindeer population size existed (Table 2). Also in the model where moose abundance and wolf population size were entered as independent variables, the calf/female ratio was related only to wolf population size (Table 2). The calf/female ratio was not significantly related to the two-way interaction term wolf population size*moose (GLS; t = -0.697, p = 0.495, GLM; t = -0.856, p = 0.404).

Fig 5

Relationships of wild forest reindeer population (a) and moose abundance (b) to wolf population in eastern Finland from 1996–2017 in models in which another sympatric ungulate was treated as another independent variable.

Discussion

Our main results provide some support to the alternative prey hypothesis but less to the apparent competition hypothesis. Population size estimates of the wild forest reindeer and moose were generally positively associated and moose density was also positively associated with reindeer calf/female ratio. These results do not support apparent competition hypothesis. The simultaneous decrease of reindeer population size and moose abundance could be due to wolves’ dietary shift from moose to reindeer when moose abundance was decreasing. However, our results must be interpreted cautiously as we examined only a part of the factors affecting the relative abundance of these species and our predator-prey system is heavily affected by humans, both through habitat changes and hunting on moose. The absence of two-tailed interaction wolf and moose abundances on reindeer population size and the calf/female ratio might indicate that moose abundance did not play a significant role for calf/female ratios in reindeer. Vital rates may be related to diseases, parasites and weather conditions. We do not have data on diseases and parasites, but in 1996–2007 the calf/female ratio was not related to previous winters’ snow depth or the timing of snowmelt in spring [20].

We found that calf/female ratio in reindeer was negatively related to wolf population size and the return of wolves could be one of the reasons that turned the reindeer population size trend from increase to decrease. Indeed, our previous studies have shown that in summer the most common kill of wolves calf reindeer [24]. Therefore the strong negative relationship of calf/female ratio to the size of wolf population most likely is due calf predation by wolves. Other large carnivores predate also on calves. However, in unpublished predation study from two GPS-collared bears tracked in our study area (methods for wolves [22]) within one month after the calving season of wild forest reindeer, only moose calves were found on kill sites. No data on predation by lynx and wolverine are available.

Reindeer population size was in close positive correlation with moose abundance but, owing to the correlative nature of our analyses we cannot draw conclusions about the reasons for this relationship. The stabilization of the reindeer population at the same time the moose population decreased to the lowest level might be in line with the relationships of woodland caribou populations to experimental reductions in moose populations in a Canadian mountain ecosystem where the decrease of caribou levelled off after moose abundances had decreased as a result of reduced apparent competition between caribou and moose [8]. However, in our study area wolf abundance was not related to moose abundance which is assumed by apparent competition hypothesis. A rapid decrease in the population of a principal prey species may cause a decline in the population of a secondary prey because predators may first consume more secondary prey, as suggested in the alternative prey hypothesis for cyclic populations [48-52]. A gradual decline in the primary prey is supposedly less detrimental to the secondary prey [25]. The similarities in the population dynamics between reindeer and moose and the stabilization of the reindeer population when the decline in the moose population slowly continued in our study area fit this assumption. However, predator-mediated apparent competition between prey species sharing a common predator would be most obvious when the predator population is responding to prey biomass, which is suggested to be reflected in encounter rates between the predator and the secondary prey [8, 53–55]. This response often occurs with a 1-2-year lag [25]. Our analysis did not provide significant evidence for the predator population’s response to the moose abundance that is likely to be the primary determinant of prey biomass: the body mass of moose is about three times bigger than that of reindeer and densities in 2000–2015 were higher (0.17–0.36/km2, Pusenius unpublished data) than those of reindeer (0.13–0.28/km2).

The wolf population was remarkably labile in our study area. The wolf is officially a protected species in Finland. Some legal, lethal wolf removals occur for definite reasons, but all in all, known mortality is usually low and does not account for variation in population growth rates [38]. Population fluctuation was, instead, highly correlated with estimates of poaching based on the known and rumoured fates of GPS-collared wolves in eastern Finland [38].

The declines in the population of woodland caribou are largely connected to anthropogenic disturbances on interactions between caribou, predators and other prey species [5, 55]. Mumma et al. [56] reported that anthropogenic linear elements (roads and seismic lines) alone increased wolf predation on woodland caribou, but the authors did not find any relationships between anthropogenic elements, moose density and woodland caribou survival, although the predation risk was increased by caribou-moose co-occurrence. In our study area, disturbances are concentrated in a dense network of small roads constructed for forestry. These forest roads are preferred as travel routes by most wolf packs in eastern Finland [26]. Most of these roads were made before our study period, during 1970s and 1980s.

Population growth of North American woodland caribou was strongest where multiple recovery options (reductions of predators and overabundant prey, translocations, fenced refuges from predators) were applied simultaneously [9]. We assume that multiple actions should be included also in plans targeting the recovery of wild forest reindeer in Europe. A reintroduction into an almost predator-free region has resulted in a new population in central-Finland, but the growth rate of that particular population has been low [20], which emphasizes the need to protect habitats that are critical to boreal Rangifer [12, 57]. Caribou mortality is higher in disturbed than undisturbed landscapes [12, 55]. Restoration of key habitats would require such large changes in forest management strategies within such wide areas that such plans may remain unrealistic. The ecology of forest-dwelling Rangifer is relatively similar across the entire coniferous zone, and in both ecosystems of North America and Finland, these deer may periodically be dependent on arboreal lichens due to deep snow that precludes cratering for terrestrial lichens [17, 58], the latter of which is the main winter forage of wild forest reindeer in eastern Finland [19], and e.g., in montane ecosystems in Alberta [59]. The biomasses of arboreal lichens are much higher in old-growth forests than in managed second-growth forests [60, 61]. The biomasses of ground lichens are also highest in old-growth forests, but their relationship to stand age is not as clear as that of arboreal lichens [61, 62]. Sustainable recovery of wild forest reindeer, however, probably requires ecosystem management where one component is habitat management. In the human-modified forest landscape, active control of the species abundances appears to be necessary for the population recovery of Rangifer. The reduction in the abundance of primary prey might decrease the predation risk although this was not supported in our study. Potential plans for the removal of predators should take into account the viability of predator populations. For example, in Finland, the brown bear and lynx are not threatened species, unlike the wolf, which is highly endangered nationally [63]. To control predation by bears and lynx, regional license allocation for leisure hunting, which is the primary method for regulating bear and lynx abundances in Finland [21], could constantly inform the vulnerability of wild forest reindeer within the coming decades. However, the managing of wild forest reindeer population should preferably be cautious whenever the mechanism is not identified [64]. For example, habitat mediated apparent competition appears to decouple in northernmost ranges of woodland caribou where moose and wolf densities are low [64] likewise in our study area.

13 May 2021

PONE-D-21-02662

Predation by wolves on European wild reindeer in a managed boreal ecosystem

PLOS ONE

Dear Dr. Kojola,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

The submitted manuscript is important as it provides useful information on prey and predator interaction. However, there are critical issues highlighting by the reviewers which needs to be addressed in the revised version.

Please submit your revised manuscript by Jun 27 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Lalit Kumar Sharma, Ph.D

Academic Editor

PLOS ONE

Additional Editor Comments:

In addition to the comments of two reviewers I also feel that there are issues related to hypothesis tested to understand the interaction of wolf and reindeer. Both of the reviewers have significantly improved the manuscript in terms of language and readability. Hence, the authors should address the comments of both the author in the revised manuscript.

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For more information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially sensitive information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

3. We note that Figure 1 in your submission contain map images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

3.1.    You may seek permission from the original copyright holder of Figure(s) [#] to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

3.2.    If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

The following resources for replacing copyrighted map figures may be helpful:

USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/

The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/

Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html

NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/

Landsat: http://landsat.visibleearth.nasa.gov/

USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/#

Natural Earth (public domain): http://www.naturalearthdata.com/

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors present four time series 1996-2017 from eastern Finland: 1) forest reindeer abundance, 2) forest reindeer calf/female ratio, 3) moose density index and 4) wolf abundance. They aim at testing the alternative prey hypothesis and the theory of apparent competition, with the reindeer being the alternative prey and moose the main prey of wolves. Moose are regulated by humans, while reindeer and wolves are not (except of some legal culling and unquantified poaching of wolves). They do not present data on predation itself, and so they test for numerical responses of predator and prey. In many ways, the system is similar to that of the woodland caribou in N-America, where human landscape modification (logging, seismic lines) increased wolves’ main prey, and where research supports apparent competition between woodland caribou and other deer species. The same authors have published a study on the reindeer-moose-wolf-system in eastern Finland in 2009, but now they present ten more years of data.

General comments:

1) Although you merely present time series and test for correlations between them, you make very strong conclusions on causal relationships (apparent competition and alternative prey hypothesis moose-reindeer-wolf). As you mention in the discussion, there are many other factors that could contribute to the observed population dynamics in forest reindeer, such as bear and lynx population, gravel roads, lichen abundance, and there might be more (disease and parasites? Snow and icing conditions?). Still, you “only” present data on wolf and moose abundance as predictors of reindeer abundance and calf/female ratio. This might be because you do not have access to the other time series. Anyway, or even more importantly, you need to make this clear to the reader in the introduction already, that you only look into one of several limiting factors, and you need to reword your text throughout the manuscript to make it less “causal”.

2) Alternative prey hypothesis APH and apparent competition ACH: You want to test if any of these two hypotheses may be supported in your system. I propose you introduce the reader better to how these two processes may affect the reindeer population. The APH is in the first hand a functional response of the predator to changes in the availability of its main prey, i.e. switching to alternative prey when main prey abundance is low. The functional response of the predator will have an effect on the mortality rate of the alternative prey and therefore change population structure (in case of age-specific mortality) and finally population growth. You would therefore exoect a positive relationship between moose and reindeer densities. The ACH on the other side describes the numerical response of the predator to changes in main prey density, which then spills over to predation rates in the alternative prey. You would expect a negative relationship between reindeer and moose abundance. These two hypotheses are not mutually exclusive, so both may take place in the same system. As a result, the positive effect from APH and the negative effect from ACH would be dampened. It could be informative to show this graphically in the introduction, and this could lead you to make more clear predictions for the reindeer population and calf/female ratio in the end of the introduction.

3) Your analyses 1: it is difficult to follow your motivations for the different models. Probably the wolf is predating mostly on the juvenile segment of the reindeer, that is why you test the calf/female ratio ~ wolf abundance. Make this more clear in the introduction already.

4) Your analyses 2: You first tested reindeer abundance ~ moose abundance and found a strong positive relationship between the two, with and without lag. Still, you chose to include both reindeer and moose abundance as predictors when modelling a) the calf/female ratio and b) wolf abundance. The strong collinearity of reindeer and moose abundance can give faulty model results and should be avoided.

5) Your results and discussion: Your main results are 1) negative relationship for calf/female ~ wolf abundance, which might indicate calf predation by wolves; 2) a negative relationship for wolf and positive for moose for reindeer abundance ~ wolf abundance + moose abundance; 3) No relationship between wolf abundance ~ moose abundance. Although the latter two findings give some support to APH, but not to ACH, you still put a lot of emphasis on ACH in the discussion, and you even propose reduction of moose population size as a measure to decrease predation on reindeer. I think evidence for this statement is not present in your results.

6) The English needs to be improved. I have not focused on he language in my review, but can do so in a revised manuscript.

Specific comments:

L1: The title is misleading. You did not study predation. You assume that predation may be an underlying mechanism of the observed correlations between the time series, that’s all.

L10-13: rewrite this first paragraph of the abstract. Include the alternative prey hypothesis and theory of apparent competition, end up with your predictions – see general comments

L39-40: the message of this sentence is unclear as it is written know. You refer to the paper of Holt and Lawton (1994) and took some of their wording in the abstract into your sentence. They write “…Theory suggests victim-species coexistence depends on particular conditions…”. By using the same wording, the context went missing. I think you can delete this first sentence and rewrite the entire first paragraph of the introduction. Explain the two hypotheses that describe direct and indirect interactions of prey species with a shared predator.

L41: Apparent competition – is it a theory or a hypothesis? I see both is used in literature. I personally would prefer hypothesis, so it is on the same line as the alternative prey hypothesis.

L42: I would put APH before ACH, because APH describes a mechanism (prey switching), while ACH is more of a process. This would also change the order in the text of the following lines.

L68-73: It would make sense to first talk about predator control, so move the last sentence up, and so talk about the regulation of the main prey, and so the combination of both.

L86: The reproductive rate is so much more than calf/female ratio. Make it more specific and introduce the reader to the wolf being a calf specialist – if this is true.

L103: Any chance to incorporate bear data into your models? I think they might be an important predator on neonate calves. What about lynx?

L124: Why had collared animals to be snow-tracked? To find pack size? Clarify!

L126-127: Clarify!

L132: So the reindeer counts were a sample, not a full count? Or is the number of observations identical with abundance?

L133: Number of observations: Is this the number of females, the number of animals, or the number of days people have been out observing?

L138: I guess this needs to be dependent, not independent.

L160 onwards: GLM with proportional data, GLS and AR-structures are standard methods, and there is no need to show the model equations. However, the calf/female ratio is not really a proportion, it is a ratio. For this, beta-regression would be more adequate than a GLM with binomial family.

Results: I would move the abundance sub-chapter (relationships between species-abundances) up, so it comes right after the first paragraph on trends.

L243: Where do you present those pseudo R-squared?

L248: See general comment on collinear predictors, so I don’t think you can include both reindeer and moose. But anyhow: What do you define as significant? In Table 2, p-values of those two predictors are > 0.05.

L252: No, you did not study predation on boral wild reindeer! See general comment 1).

L255: …major reason…? Rather: …one of the reasons…

L261-264: But the wolf population did not really decrease after moose population decreased. Which brings to my mind that the ACH should test predator numerical response not only as a function of its principal prey, but rather of the total prey base, i.e. the sum of available prey biomass irrespective of species. You however use an index for moose abundance, so it will be difficult to estimate the sum of available biomass of reindeer and moose.

L288: I assume the gravel roads are a consequence of forestry. Clarify, maybe call them forest roads. Has their density changed throughout the years? Or has their use changed?

L292-293: This is a really strong statement that has no direct support from your data, see also general comments 1) and 5).

L314: Moose abundance was positively related with reindeer abundance, and I consider the model looking into calf/female ratio ~moose abundance + reindeer abundance as invalid due to collinearity of the predictors. So no support for a negative relationship between moose abundance and “predation risk”.

L525 and else in the text: You use abundance, population, population size and density (e.g. Figure 5b) intermittently. Be consistent.

Figure 1: Finland is not labelled, while all other countries are! The map allows for inclusion of more information, which could make the study easier to understand. You could for example add wolf territories, e.g. as overlapping territories from all years. And/or the network of forest roads. Maybe also the area of the other wild reindeer population. Anything to make it more informative!

Figure 2: It would make sense to include two similar graphs for calf/female ratio.

Figure 5: The y-axis goes below zero, which does not make sense for wolf abundance. This makes me think that you maybe rather use a Poisson-regression when comparing wolf abundance (a count) to moose abundance index (or reindeer abundance).

Reviewer #2: The authors have been studying wolf predation on wild forest reindeer in a system where an endangered predator (wolf) has a primary non-endangered prey species (moose) and secondary endangered prey species (wild forest reindeer). In the manuscript the authors have used correlative approach to investigate species abundance interactions on reindeer and wolf population sizes. In addition, they have also used the same approach to investigate calf/female ratio in winter herds of wild forest reindeer. In former studies this ratio has been shown to be related on wolf abundance and in this manuscript the authors have broadened their former study by analysing longer data period and adding the yearly abundance of moose into their models. Their main findings were that the calf/female ratio was negatively associated negatively to both wolf population size and moose abundance. Wild forest reindeer size was only dependent on wolf population size in model where the moose abundance was entered as another independent variable. There was not strong evidence on effect of moose abundance or reindeer population size on wolf population size.

The long term time series for wildlife species used in the study are imposing and the statistical analyses are rather comprehensive, even though there have been some problems in the analyses because of autocorrelations. The study system and the results are very interesting. However, the interpretation of the causal effects behind the species population size variation are difficult because of the correlative approach used in the study. The fact that moose population size is heavily regulated via hunting and wolf population is fluctuating mainly because of poaching makes interpretation of the results even more difficult. However, the observation that the predation by wolves on reindeer might be influenced by moose abundance could still have substantial management implications and the authors are discussing praiseworthy on possible management actions.

The paper is generally well written. However, especially in the abstract the authors should present their results more comprehensively to make the message of the paper clear (see below). And, please, do not use “population” as a substitute of “population size”.

Minor points:

Abstract:

Line 15: “…reproductive rate…”. What does this mean? Is it your calf/female ratio. Reproduction is generally measured by gross reproduction rates or net reproduction rates that generally indicate the ratio between the sizes of the daughter's and mother's generations and you are not really measuring them.

Line 15: “wild reindeer”. Sometimes wild reindeer and wild forest reindeer. Please, be consistent.

Line 20: “Reindeer and moose abundances were highly correlated…”. Is this already a result and should be in your Results section of your abstract. You are actually repeating this result also on line 24-25. “The trends in reindeer population size and moose abundance were almost identical”.

Line 27: “Change in reindeer population between consecutive winters”. Should not this be “Change in reindeer population *size* between consecutive winters”.

Line 28: “The calf/female ratio was closely related to wolf population size”. The calf/female ratio was closely related *negatively* to wolf population size.

Line 29: “the reindeer population was related to the wolf population”. I suppose that you mean “the reindeer population *size* was related *negatively” to the wolf population *size*. Other wise this text sounds funny; and tell also that the relationship is negative. …”

Line 30: “The wolf population was…”. Should be “The wolf population *size* was

Introduction: “

Lines 85-90: I suggest that the authors construct a table where they show what are the predictions of their two theoretical models and how their different results support or do not support the two hypotheses.

Data:

Lines 132-133: “the calf/female ratio was weighed against the annual number of observations”. In the analyses?

Lines 145-146: “The aim of the study was to model the relationships between these populations”. Or: “The aim of the study was to model the interactions in abundances of these populations” or did you really model the relationship between the populations?

Line 147-148: “In addition, the relationship of the calf/female ratio to the wolf population..” Or: “In addition, the relationship of the calf/female ratio to the wolf population *size*”.

Result:

Line 217: “Reindeer population and moose abundance.”Or “ Reindeer population *size* and moose abundance “

Line 221: “The annual growth rate of the wild forest reindeer population was related to the number of calves weighted by the number of females (the calves/females ratio) in the linear model Y = -0.16 + 0.473 * X, t = 2.57, p = 0.020).” Is the number of calves weighted by the number of females” really correct here. Thus what are your Y and X here. Please, clarify.

Line 225: “but the ratio was related to the wolf population” …. Should be: “but the ratio was related to the wolf population *size*”.

Line 227-228: “the calf/female ratio was related to the wolf population in a highly 28 significant fashion”. Should be: “the calf/female ratio was related to the wolf population *size*” in a highly significant fashion”.

Line 228-229: “while no relationship between the ratio and the reindeer population *size* existed”. However, there was significant association between reindeer population size and calf/female ratio in your model with three independent variables. Should you mention it, too?

Line 237-241: “but in a model where wolf and moose population *sizes* were entered as independent variables, reindeer population size was negatively related to the wolf population *size* and positively related to the moose population *size*. In models with a one-year lag, the reindeer population *size* was not related to the wolf population *size* alone but was related to both the wolf population *size* and moose*abundance* in a model where both were entered as independent variables”.

Lines 247-250: “Wolf. In a model without a lag, the wolf population size was positively related to moose abundance and negatively related to reindeer population size (Table 2, Fig. 5). In a model where 249 the wolf population size was the dependent variable with a one-year lag, neither the moose nor reindeer population *size* was significantly related to the wolf population (Table 2).

Discussion:

Lines 254-255: “We found that calf/female ratio in reindeer was negatively related to wolf population *size*…

Lines 260-261: “The stabilization of the reindeer population size at the same time the moose population *abundance* decreased…”

Figures:

Fig 4: “Relationship of the calf/female ratio of wild forest reindeer to the wolf population *size*in…”

Fig 5: “Relationships of reindeer population *size*(a) and moose abundance (b) to wolf population *size*…”

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

28 Sep 2021

Responses to referees’ comments:

REFEREE 1: General comments:

1) Although you merely present time series and test for correlations between them, you make very strong conclusions on causal relationships (apparent competition and alternative prey hypothesis moose-reindeer-wolf). As you mention in the discussion, there are many other factors that could contribute to the observed population dynamics in forest reindeer, such as bear and lynx population, gravel roads, lichen abundance, and there might be more (disease and parasites? Snow and icing conditions?). Still, you “only” present data on wolf and moose abundance as predictors of reindeer abundance and calf/female ratio. This might be because you do not have access to the other time series. Anyway, or even more importantly, you need to make this clear to the reader in the introduction already, that you only look into one of several limiting factors, and you need to reword your text throughout the manuscript to make it less “causal”.

RESPONSE: We have made extensive changes into Abstract, Introduction and Discussion. We reworded the also the title so that causal relationships between reindeer, moose and wolf are not assumed. We do not have data on diseases and parasites which is now notified in Discussion. We refer to one former study where no relationship of the calf/female ratio to snow depth and the timing of snowmelt was not detected in this reindeer population.

2) Alternative prey hypothesis APH and apparent competition ACH: You want to test if any of these two hypotheses may be supported in your system. I propose you introduce the reader better to how these two processes may affect the reindeer population. The APH is in the first hand a functional response of the predator to changes in the availability of its main prey, i.e. switching to alternative prey when main prey abundance is low. The functional response of the predator will have an effect on the mortality rate of the alternative prey and therefore change population structure (in case of age-specific mortality) and finally population growth. You would therefore expect a positive relationship between moose and reindeer densities. The ACH on the other side describes the numerical response of the predator to changes in main prey density, which then spills over to predation rates in the alternative prey. You would expect a negative relationship between reindeer and moose abundance. These two hypotheses are not mutually exclusive, so both may take place in the same system. As a result, the positive effect from APH and the negative effect from ACH would be dampened. It could be informative to show this graphically in the introduction, and this could lead you to make more clear predictions for the reindeer population and calf/female ratio in the end of the introduction.

RESPONSE: We clarified the predictions for our study population based on hypotheses of alternative prey and apparent competition (in this order as suggested by the referee).

3) Your analyses 1: it is difficult to follow your motivations for the different models. Probably the wolf is predating mostly on the juvenile segment of the reindeer, that is why you test the calf/female ratio ~ wolf abundance. Make this more clear in the introduction already.

RESPONSE: Based on earlier studies, summer predation by wolves on reindeer focuses on calves. These results have been now mentioned, reasoning the treatment of calf/female ratio as a dependent variable.

4) Your analyses 2: You first tested reindeer abundance ~ moose abundance and found a strong positive relationship between the two, with and without lag. Still, you chose to include both reindeer and moose abundance as predictors when modelling a) the calf/female ratio and b) wolf abundance. The strong collinearity of reindeer and moose abundance can give faulty model results and should be avoided.

RESPONSE:

We re-analyzed data using beta regression for the calf/female ratio and GLM models together with GLS models for analysis of relationships between populations. To avoid errors due to multicollinearity, reindeer and moose abundances were not entered into the same model as independent variables.

Below is the more detailed description about data re-analysis:

The multicollinearity of the predictors such as moose and reindeer abundances:

As referees pointed out, the wildlife populations correlate with each other, and multicollinearity is usually a problem in the interpretation of the results. However, in our case our models are just explanatory models although we illustrated the effects using the effects plots for the predictors, fixing the other predictor(s) at their mean values. The correlating predictors, being in the same model also “purified” each other’s effects. We wanted to see how they worked when they were added to the model one by one. They “purified” together the autocorrelated errors as well in some of the models. The models for the calf/female ratio by adding the variable one by one using GLM and binomial distribution:

1. Wolves only

Coefficients:

Estimate Std. Error t value Pr(>|t|)

(Intercept) 0.135406 0.198021 0.684 0.502

Wolves -0.037399 0.007529 -4.967 8.56e-05 ***

2. Wolves and moose

Coefficients:

Estimate Std. Error t value Pr(>|t|)

(Intercept) 0.23356 0.50458 0.463 0.649000

Wolves -0.03759 0.00779 -4.826 0.000136 ***

Moose -0.03326 0.15651 -0.213 0.834104

---

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for quasibinomial family taken to be 6.946016)

3. Wolves and reindeer

Coefficients:

Estimate Std. Error t value Pr(>|t|)

(Intercept) -0.3114357 0.4046989 -0.770 0.451549

Wolves -0.0331237 0.0081116 -4.084 0.000697 ***

Reindeer 0.0003449 0.0002736 1.260 0.223625

---

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for quasibinomial family taken to be 6.311177)

All three explanatory variables (wolves,reindeer,moose)

Coefficients:

Estimate Std. Error t value Pr(>|t|)

(Intercept) 0.4656050 0.3680501 1.265 0.222917

allwolves -0.0202632 0.0069474 -2.917 0.009617 **

REINDEER 0.0018684 0.0004542 4.114 0.000725 ***

moose.density -0.9279260 0.2464483 -3.765 0.001543 **

---

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for quasibinomial family taken to be 3.70951)

(Dispersion parameter for quasibinomial family taken to be 6.575754)

Thus, both populations, reindeer and moose are strongly correlated with each other. This can be seen also in Fig. 2. The significant negative coefficient of moose density is difficult to understand and interpret. Thus, the model 4 has been excluded.

Certain multicollinearity is evident in the models for wildlife populations, in general, and it might reveal some interesting features of the relations between the populations when the models were built by adding the variables one by one and testing also the interaction effects, as we have done in our modeling approach. We skipped the interaction effects in the final model versions because they were clearly non-significant.

GLS vs GLM models

In table 2 we presentd GLS models and log-transformations (if needed) for the response variables to obtain better residuals, although the data was count data. Poisson or negative binomial models would be a more reasonable alternative to the count data like this.

The reason for our selection in our manuscript was the fact that in R the pseudo-likelihood based function glm in MASS package could not allowed autoregressive structures for the error term, but gls function in nlme package could. Especially when a certain population was modeled using just one other population as an explanatory variable, the autoregressive correlation structure was highly needed (AR1 or AR2) for reasonable standard errors for the estimated coefficients and the corresponding p-values. However, Poisson models using the AR-structures could be computed using R function glmmPQL. The function needs the specification of a random part of the model. Our data is non-hierarchical, and the only way to use glmmPQL, that can be handle the AR-errors structures is to specify a “pseudo” random part that gives the variance very near to zero. In addition, the function glmmPQL estimates so called residual, corresponding to the square-root of dispersion parameter. The only problem using that function is slightly biased degrees of freedom, because of the estimated random effect. It could be corrected using e.g. Kenward-Rogers df corrections, but as far as we know, they are not available for almmPQL (also pseudo-likelihood method).

In R, there are some other packages for Poisson or negative binomial models and autoregressive error structures (e.g. glarma), but the predictions with confidence intervals are much harder to compute, because they can not utilize the smart R package effects in the computing of the predictions with their confidence intervals.

Our solution to the referee’s comment to replace GLS using GLM was to add the corresponding GLM-based t-values and corresponding p-values to table 2 using glmmPQL function and illustrate the effects in the figures using the GLM-approach. The results and interpretations did not change considerable, but no possible negative predictions were met otherwise than using the GLS. Interpretation of the results remained the same.

We also tested how reindeer population size and the calf/female ratio in reindeer was related to two-way interaction between wolf population size and moose abundance.

Binomial distribution in the models for calf / female ratio vs. beta distribution

A referee: ”… For this, beta-regression would be more adequate than a GLM with binomial family”

The calf / female ratio is merely a ratio than the proportion, as the referee mentioned. Beta distribution would be nice and alternative option to model the phenomenon.

In fact, our binomial model would be approximately expressed as the proportion of the females with a calf (success) / the females without a calf (failure). Using the proportion of females with calves / all females weighted by all females would give the same results.

The binomial model awaked the problem of overdispersion. This problem could be avoided by using the beta distribution, but also by estimating the dispersion parameter in the binomial model, as we did in our original model versions.

In R, there is beta distribution available in package betareg.

Although the results (their interpretations) were almost the same, as we illustrate later, we used the beta distribution in our models, because it is more appropriate to interpret the calves by females as a ratio, such as the referee suggested. A good think is that the R package betareg can today utilize the R function effects (in R package effects) to obtain the predicted values with the confidence intervals!

Binomial model

> mod<-glm(calves/females ~ wolves+reindeer+moose.density,weights=females,family="quasibinomial",data=d2)# Uusi malli uudella datalla 160519!!!

> summary(mod)

Call:

glm(formula = calves/females ~ allwolves + REINDEER + moose.density,

family = "quasibinomial", data = d2, weights = females)

Deviance Residuals:

Min 1Q Median 3Q Max

-3.4961 -1.0399 0.2541 0.7820 2.9779

Coefficients:

Estimate Std. Error t value Pr(>|t|)

(Intercept) 0.4656050 0.3680501 1.265 0.222917

allwolves -0.0202632 0.0069474 -2.917 0.009617 **

REINDEER 0.0018684 0.0004542 4.114 0.000725 ***

moose.density -0.9279260 0.2464483 -3.765 0.001543 **

---

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for quasibinomial family taken to be 3.70951)

Null deviance: 288.820 on 20 degrees of freedom

Residual deviance: 62.182 on 17 degrees of freedom

AIC: NA

Number of Fisher Scoring iterations: 4

Beta regression model with logit-link

# vasaos denotes the ratio of calves/females

> mod <- betareg(vasaos ~ allwolves+REINDEER+moose.density,link="logit", data = d2)

> summary(mod)

Call:

betareg(formula = vasaos ~ allwolves + REINDEER + moose.density, data = d2, link = "logit")

Standardized weighted residuals 2:

Min 1Q Median 3Q Max

-2.4928 -0.6743 0.1543 0.5653 2.5688

Coefficients (mean model with logit link):

Estimate Std. Error z value Pr(>|z|)

(Intercept) 0.290914 0.338926 0.858 0.390705

allwolves -0.022109 0.006349 -3.482 0.000497 ***

REINDEER 0.001475 0.000464 3.178 0.001481 **

moose.density -0.693933 0.253986 -2.732 0.006292 **

Phi coefficients (precision model with identity link):

Estimate Std. Error z value Pr(>|z|)

(phi) 72.21 22.15 3.26 0.00111 **

---

Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Type of estimator: ML (maximum likelihood)

Log-likelihood: 31.56 on 5 Df

Pseudo R-squared: 0.7031

Number of iterations: 46 (BFGS) + 4 (Fisher scoring)

Results are not completely the same, but the interpretations are. The example here was the model with the highly correlating population of moose density and reindeer abundance. The response, vasaos in the beta model denotes the proportion of calves by females.

Comparing the alternative models using pseudo-R2

The R2 values were missing in table 2. In our corrected version we computed the R2 values for the GLS models. They are pseudo-R2 values which cannot be interpreted as the proportions explained by the predictors of the models, merely just for the comparisons between the GLS models, similarly to the models in Table 1 where the R2 values represented the pseudo-R2 for the quasi-likelihood GLMs.

We used R-package rcompanion in the computation of the pseudo-R2 values for GLS. The function nagelkerke in the package gives three different pseudo-R2 values. We used Cox and Snell-values. The other values usually gave rather similar interpretation. In some of the models it might be a little confusing that the values are slightly negative. It happened when the p-value of the only predictor was very or rather near to 1.

5) Your results and discussion: Your main results are 1) negative relationship for calf/female ~ wolf abundance, which might indicate calf predation by wolves; 2) a negative relationship for wolf and positive for moose for reindeer abundance ~ wolf abundance + moose abundance; 3) No relationship between wolf abundance ~ moose abundance. Although the latter two findings give some support to APH, but not to ACH, you still put a lot of emphasis on ACH in the discussion, and you even propose reduction of moose population size as a measure to decrease predation on reindeer. I think evidence for this statement is not present in your results.

RESPONSE: We had highlighted our main findings in Results and Discussion as suggested by the referee. We notify that our results be more supportive on APH than ACH but cautiously because we considered just correlations between populations. In the first chapters of discussion we briefly go through the primary findings and tell that our results are probably more supportive to APH than ACH hypothesis.

6) The English needs to be improved. I have not focused on he language in my review, but can do so in a revised manuscript.

RESPONSE: The English has been checked by a native but may need further improvement.

Specific comments:

L1: The title is misleading. You did not study predation. You assume that predation may be an underlying mechanism of the observed correlations between the time series, that’s all.

RESPONSE: We changed the title to match better with the content.

L10-13: rewrite this first paragraph of the abstract. Include the alternative prey hypothesis and theory of apparent competition, end up with your predictions – see general comments

RESPONSE: Done

L39-40: the message of this sentence is unclear as it is written know. You refer to the paper of Holt and Lawton (1994) and took some of their wording in the abstract into your sentence. They write “…Theory suggests victim-species coexistence depends on particular conditions…”. By using the same wording, the context went missing. I think you can delete this first sentence and rewrite the entire first paragraph of the introduction. Explain the two hypotheses that describe direct and indirect interactions of prey species with a shared predator.

RESPONSE: Now as follows: Background. The alternative prey hypothesis describes a mechanism where that the population size of secondary prey species, endangered wild forest reindeer (Rangifer tarandus fennicus) in our study, may decrease whenever abundance of the principal prey, moose (Alces alces) goes down through dietary shift by a generalist predator, wolf (Canis lupus), from moose to reindeer. Apparent competition is a process where the population size of reindeer can be assumed to decrease as a result of the increased population size of wolf which results from increasing prey biomass that is highly related to moose abundance in our study area in Finland.

L41: Apparent competition – is it a theory or a hypothesis? I see both is used in literature. I personally would prefer hypothesis, so it is on the same line as the alternative prey hypothesis.

RESPONSE: Now both the apparent competition and the alternative prey hypotheses are termed as hypotheses.

L42: I would put APH before ACH, because APH describes a mechanism (prey switching), while ACH is more of a process. This would also change the order in the text of the following lines.

RESPONSE: The order changed.

L68-73: It would make sense to first talk about predator control, so move the last sentence up, and so talk about the regulation of the main prey, and so the combination of both.

RESPONSE: The order changed.

L86: The reproductive rate is so much more than calf/female ratio. Make it more specific and introduce the reader to the wolf being a calf specialist – if this is true.

RESPONSE: We removed the term ‘reproductive rate’ from the ms. We use now only calf/female ratio.

L103: Any chance to incorporate bear data into your models? I think they might be an important predator on neonate calves. What about lynx?

RESPONSE: We do not have proper data about changes in bear and lynx population. We quoted to our unpublished small data about bear predation; we had tracked two bears during early summer and found only moose calves at kill sites. It may be that the very scattered distribution of female forest reindeer during calving time decreases bears’ motivation to actively seek for the calf reindeer that are small compared to moose calves.

L124: Why had collared animals to be snow-tracked? To find pack size? Clarify!

RESPONSE: Yes, clarified.

L126-127: Clarify!

RESPONSE: Described with more details: The moose abundance index based on records by hunters. Moose hunting clubs are operating within their own territory of approximately 5 000 ha. Each hunting club records the number of moose observations per hunting day during the autumn hunting season (from the last Saturday of September to the end of December) and the resultant index provides the annual mean number of moose per hunting day for our study area. We used this index because moose density was not estimated for years 1996-1999.

L132: So the reindeer counts were a sample, not a full count? Or is the number of observations identical with abundance?

RESPONSE: Yes, reindeer counts were full counts. Clarified.

L133: Number of observations: Is this the number of females, the number of animals, or the number of days people have been out observing?

RESPONSE: This means the number of females. We have clarified this.

L138: I guess this needs to be dependent, not independent.

RESPONSE: Yes. Changed.

L160 onwards: GLM with proportional data, GLS and AR-structures are standard methods, and there is no need to show the model equations. However, the calf/female ratio is not really a proportion, it is a ratio. For this, beta-regression would be more adequate than a GLM with binomial family.

Results: I would move the abundance sub-chapter (relationships between species-abundances) up, so it comes right after the first paragraph on trends.

RESPONSE: Changed as the referee proposes. The calves/females are treated as a ratio in beta-regression. The order has been changed in all sections.

L243: Where do you present those pseudo R-squared?

RESPONSE: They are now presented in Table 1 and 2.

L248: See general comment on collinear predictors, so I don’t think you can include both reindeer and moose. But anyhow: What do you define as significant? In Table 2, p-values of those two predictors are > 0.05.

RESPONSE: True, changed, only the p values <0.05 have now been mentioned as significant.

L252: No, you did not study predation on boral wild reindeer! See general comment 1).

RESPONSE: This sentence has been omitted.

L255: …major reason…? Rather: …one of the reasons…

RESPONSE: changed.

L261-264: But the wolf population did not really decrease after moose population decreased. Which brings to my mind that the ACH should test predator numerical response not only as a function of its principal prey, but rather of the total prey base, i.e. the sum of available prey biomass irrespective of species. You however use an index for moose abundance, so it will be difficult to estimate the sum of available biomass of reindeer and moose.

RESPONSE: We mention that moose density is the main determinant of prey biomass and therefore moose abundance is relatively relevant parameter when considering ACH.

L288: I assume the gravel roads are a consequence of forestry. Clarify, maybe call them forest roads. Has their density changed throughout the years? Or has their use changed?

RESPONSE: Yes. This is now changed. The boom in building forest roads was in 1970s and 1980s. Their density has increased just marginally during our study period. This is notified in text.

L292-293: This is a really strong statement that has no direct support from your data, see also general comments 1) and 5).

RESPONSE: True, a sentence has been omitted.

L314: Moose abundance was positively related with reindeer abundance, and I consider the model looking into calf/female ratio ~moose abundance + reindeer abundance as invalid due to collinearity of the predictors. So no support for a negative relationship between moose abundance and “predation risk”.

RESPONSE: Good point, the sentence modified into as follows: The reduction in the abundance of primary prey might (seems) decrease the predation risk although this was not supported in our study.

L525 and else in the text: You use abundance, population, population size and density (e.g. Figure 5b) intermittently. Be consistent.

RESPONSE: Changed to be consistent.

Figure 1: Finland is not labelled, while all other countries are! The map allows for inclusion of more information, which could make the study easier to understand. You could for example add wolf territories, e.g. as overlapping territories from all years. And/or the network of forest roads. Maybe also the area of the other wild reindeer population. Anything to make it more informative!

RESPONSE: A new map has been drawn. The Fig 1. shows Finland as labelled, wolf territory boundaries and road network.

Figure 2: It would make sense to include two similar graphs for calf/female ratio.

RESPONSE: Calf/female ratio is now shown in Fig. 2. Fig. 3 omitted.

Figure 5: The y-axis goes below zero, which does not make sense for wolf abundance. This makes me think that you maybe rather use a Poisson-regression when comparing wolf abundance (a count) to moose abundance index (or reindeer abundance).

RESPONSE: Changed as the referee suggests.

Reviewer #2: The authors have been studying wolf predation on wild forest reindeer in a system where an endangered predator (wolf) has a primary non-endangered prey species (moose) and secondary endangered prey species (wild forest reindeer). In the manuscript the authors have used correlative approach to investigate species abundance interactions on reindeer and wolf population sizes. In addition, they have also used the same approach to investigate calf/female ratio in winter herds of wild forest reindeer. In former studies this ratio has been shown to be related on wolf abundance and in this manuscript the authors have broadened their former study by analysing longer data period and adding the yearly abundance of moose into their models. Their main findings were that the calf/female ratio was negatively associated negatively to both wolf population size and moose abundance. Wild forest reindeer size was only dependent on wolf population size in model where the moose abundance was entered as another independent variable. There was not strong evidence on effect of moose abundance or reindeer population size on wolf population size.

The long term time series for wildlife species used in the study are imposing and the statistical analyses are rather comprehensive, even though there have been some problems in the analyses because of autocorrelations. The study system and the results are very interesting. However, the interpretation of the causal effects behind the species population size variation are difficult because of the correlative approach used in the study. The fact that moose population size is heavily regulated via hunting and wolf population is fluctuating mainly because of poaching makes interpretation of the results even more difficult. However, the observation that the predation by wolves on reindeer might be influenced by moose abundance could still have substantial management implications and the authors are discussing praiseworthy on possible management actions.

The paper is generally well written. However, especially in the abstract the authors should present their results more comprehensively to make the message of the paper clear (see below). And, please, do not use “population” as a substitute of “population size”.

RESPONSE: We have re-analyzed data to avoid errors due to multicollinearity. The term ‘population’ has been changed into ‘population change’ when the sentence does not clearly indicate that there is a question of population size.

Minor points:

Abstract:

Line 15: “…reproductive rate…”. What does this mean? Is it your calf/female ratio. Reproduction is generally measured by gross reproduction rates or net reproduction rates that generally indicate the ratio between the sizes of the daughter's and mother's generations and you are not really measuring them.

RESPONSE: Each ‘reproductive rate’ changed to ‘calf/female’ ratio.

Line 15: “wild reindeer”. Sometimes wild reindeer and wild forest reindeer. Please, be consistent.

RESPONSE: Now only “wild forest reindeer”, because it is an official name of the subspecies.

Line 20: “Reindeer and moose abundances were highly correlated…”. Is this already a result and should be in your Results section of your abstract. You are actually repeating this result also on line

24-25. “The trends in reindeer population size and moose abundance were almost identical”.

RESPONSE: Changed like this. Repeats removed.

Line 27: “Change in reindeer population between consecutive winters”. Should not this be “Change in reindeer population *size* between consecutive winters”.

RESPONSE: Yes, changed.

Line 28: “The calf/female ratio was closely related to wolf population size”. The calf/female ratio was closely related *negatively* to wolf population size.

RESPONSE: Changed.

Line 29: “the reindeer population was related to the wolf population”. I suppose that you mean “the reindeer population *size* was related *negatively” to the wolf population *size*. Other wise this text sounds funny; and tell also that the relationship is negative. …”

RESPONSE: Corrected as the referee suggests.

Line 30: “The wolf population was…”. Should be “The wolf population *size* was

RESPONSE: Yes. Changed.

Introduction: “

Lines 85-90: I suggest that the authors construct a table where they show what are the predictions of their two theoretical models and how their different results support or do not support the two hypotheses.

RESPONSE: In this reviewed version we now express the hypotheses and predictions in much more coincided and clearer fashion in the Abstract, Introduction and Discussion. We think that these changes a bit decrease the need for a table, but if it is seen necessary we are ready to instruct such into the manuscript.

Data:

Lines 132-133: “the calf/female ratio was weighed against the annual number of observations”. In the analyses?

RESPONSE: In the present statistical analysis (betareg) we use calf/female ratio.

Lines 145-146: “The aim of the study was to model the relationships between these populations”. Or: “The aim of the study was to model the interactions in abundances of these populations” or did you really model the relationship between the populations?

RESPONSE: Changed into “

Line 147-148: “In addition, the relationship of the calf/female ratio to the wolf population..” Or: “In addition, the relationship of the calf/female ratio to the wolf population *size*”.

RESPONSE: Correct. Changed.

Result:

Line 217: “Reindeer population and moose abundance.”Or “ Reindeer population *size* and moose abundance “

Line 221: “The annual growth rate of the wild forest reindeer population was related to the number of calves weighted by the number of females (the calves/females ratio) in the linear model Y = -0.16 + 0.473 * X, t = 2.57, p = 0.020).” Is the number of calves weighted by the number of females” really correct here. Thus what are your Y and X here. Please, clarify.

RESPONSE: Y and X added in into the sentence.

Line 225: “but the ratio was related to the wolf population” …. Should be: “but the ratio was related to the wolf population *size*”.

RESPONSE. Yes, changed.

Line 227-228: “the calf/female ratio was related to the wolf population in a highly 28 significant fashion”. Should be: “the calf/female ratio was related to the wolf population *size*” in a highly significant fashion”.

RESPONSE: YES, changed

Line 228-229: “while no relationship between the ratio and the reindeer population *size* existed”. However, there was significant association between reindeer population size and calf/female ratio in your model with three independent variables. Should you mention it, too?

RESPONSE: We have omitted the results based on three independent variables to avoid errors due to multicollinearity because reindeer and moose abundances were so highly correlated, as suggested by the referee 1.

Line 237-241: “but in a model where wolf and moose population *sizes* were entered as independent variables, reindeer population size was negatively related to the wolf population *size* and positively related to the moose population *size*. In models with a one-year lag, the reindeer population *size* was not related to the wolf population *size* alone but was related to both the wolf population *size* and moose*abundance* in a model where both were entered as independent variables”.

RESPONSE: Changed.

Lines 247-250: “Wolf. In a model without a lag, the wolf population size was positively related to moose abundance and negatively related to reindeer population size (Table 2, Fig. 5). In a model where 249 the wolf population size was the dependent variable with a one-year lag, neither the moose nor reindeer population *size* was significantly related to the wolf population (Table 2).

RESPONSE: Changed.

Discussion:

Lines 254-255: “We found that calf/female ratio in reindeer was negatively related to wolf population *size*…

Lines 260-261: “The stabilization of the reindeer population size at the same time the moose population *abundance* decreased…”

Figures:

Fig 4: “Relationship of the calf/female ratio of wild forest reindeer to the wolf population *size*in…”

Fig 5: “Relationships of reindeer population *size*(a) and moose abundance (b) to wolf population *size*…”

RESPONSE: “Population” is now termed as “population size”.

Submitted filename: RESPONSES_PLOS One_Kojola et al.docx

Click here for additional data file.

18 Oct 2021

Calf/female ratio and population dynamics of wild forest reindeer in relation to wolf and moose abundances in a managed European ecosystem

PONE-D-21-02662R1

Dear Dr. Kojola,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Lalit Kumar Sharma

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

3 Dec 2021

PONE-D-21-02662R1

Calf/female ratio and population dynamics of wild forest reindeer in relation to wolf and moose abundances in a managed European ecosystem

Dear Dr. Kojola:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Lalit Kumar Sharma

Academic Editor

PLOS ONE