An increase in detection rates of the critically endangered Baltic Proper harbor porpoise in Swedish waters in recent years

The Baltic Proper harbor porpoise (Phocoena phocoena) is currently listed as critically endangered (CR), with the Static Acoustic Monitoring of the Baltic Sea Harbor Porpoise (SAMBAH) project concluding that only ~500 individuals remain. This population has a distribution that spans the waters of nine countries, making regular abundance estimates and management action challenging. Given the continued decline of other depleted porpoises, namely the vaquita (Phocoena sinus), the question is often raised about whether management action would even have a positive impact, or whether it is too late for population recovery. When abundance estimates are sparse over time, monitoring programs at key sites are likely to serve as the best indication of population trends, and may provide an early indication of changes at the population level. We compared passive acoustic monitoring data from 12 stations that were utilized both in the SAMBAH project (2011–2013) and as a part of the Swedish National Monitoring Program (2017–2020) to determine trends in detection rates. There was a 29% increase in mean daily detection rate during May–October (over the breeding season) between the two study periods. At the three stations with the highest number of detections, log linear regression revealed a yearly increase of 2.4% between 2011 and 2019 (−4.4–9.6, 95% CI). This may be indicative of the beginnings of population recovery, or simply an indication that the decline has stalled. The rate of increase is still well below what is likely to be possible for porpoise populations, and unlikely to buffer against any potential increase in pressures in the future. We therefore call for urgent management action to remove threats and protect this CR population, the only resident cetacean in the Baltic region, in order to give it the best chance of recovery.


| INTRODUCTION
The harbor porpoise is the only resident species of cetacean in the Baltic region. Historically, porpoises were once numerous and more widely distributed in the Baltic Proper (Koschinski, 2001;Lindroth, 1962;Sk ora & Kuklik, 2003), but a wide range of threats have led to a dramatic reduction in abundance over the past century (Koschinski, 2001). Genetic and morphometric studies have both concluded that porpoises in the Baltic Proper are distinct from those in the Kattegat, Skagerrak, and North Sea (Galatius, Kinze, & Teilmann, 2012;Lah et al., 2016;Wiemann et al., 2010). The first and only assessment of the abundance of the Baltic Proper harbor porpoise population occurred between 2011 and 2013, when the Static Acoustic Monitoring of the Baltic Sea Harbor Porpoise (SAMBAH) project estimated that there were only 497 individuals (80-1,091, 95% CI) remaining (SAMBAH, 2016). The Baltic Proper harbor porpoise is currently classified as critically endangered (CR) by both the International Union for Conservation of Nature (IUCN) and the Helsinki Commission (HELCOM) (Hammond et al., 2008;HELCOM, 2013). The species is also listed in Annex IV of the EU Habitats Directive (Council Directive 92/43/EEC, European Commission (1992)), implying that a strict protection regime must be in place across their entire distributional range. Despite this, for three consecutive assessment periods (2007, 2013, and 2019) under Article 17 reporting for the Habitats Directive, all relevant EU Member State assessments and the EU biogeographical assessment have classified the conservation status of the Baltic Proper Harbor porpoise as "unfavorable-bad." Due to the small population size, there are limited data available on key biological and ecological factors affecting the population. This includes a lack of regular abundance estimates, and information on the extent to which the population is exposed to threats. Consequently, there is a high level of uncertainty on population trends and the future viability of the population.
There are many threats to the Baltic Proper harbor porpoise, with bycatch, contaminants and loud impulsive underwater noise sources classified as "high" threats based on evidence or strong likelihood of an impact on individual mortality, health and/or reproduction, likely leading to negative population effects (ICES, 2019). Further, threats such as prey depletion, shipping noise, and habitat degradation have been classified as "medium." For 10 of the 13 species, subspecies and populations ("units") of small cetaceans listed as CR on the IUCN Red List (including the Baltic Proper harbor porpoise), bycatch is the main threat (Brownell et al., 2019). In the Baltic Proper, 97% or more of the harbor porpoise bycatch has been reported to occur in gillnets (Berggren, 1994;EC-DGMARE, 2014;Sk ora & Kuklik, 2003). The limit of allowable human-caused mortality for the Baltic Proper population has been estimated to be 0.7 animals per year (North Atlantic Marine Mammal Commission & Norwegian Institute of Marine Research, 2019) based on the potential biological removal (PBR) approach (Wade, 1998). A bycatch rate close to zero can only be reached by closing all gillnet fisheries within the distribution range of the Baltic Proper harbor porpoise, as pingers do not completely remove the threat of bycatch (Dawson, Northridge, Waples, & Read, 2013;Larsen & Eigaard, 2014;Palka, Rossman, VanAtten, & Orphanides, 2008). In May 2020, ICES released Emergency Measures advice, recommending spatial-temporal closures of fisheries throughout many Natura 2000 sites within the distributional range of the population, and obligatory use of pingers on all static nets (ICES, 2020a). This advice is still currently being considered by the EU and member states.
Unlike other severely depleted species (e.g., vaquita) or populations of harbor porpoises (e.g., U.S. Morro Bay stock) that are able to be managed by the legislation set by one country, the Baltic Proper harbor porpoise has a distribution that spans nine countries. This issue makes regular distribution-wide surveys logistically and politically challenging, increasing the importance of local monitoring programs on the population. Such programs enable investigations of changes in detection rates that could be an early indication of changes in abundance. The SAMBAH project revealed that, during May-October, most detections occurred in Swedish waters (Carlén et al., 2018), suggesting that this region is important for the population over the breeding season. Based on the results of SAMBAH, a new Natura 2000 site was designated specifically for porpoises in Sweden that is the largest Natura 2000 site in the Baltic Sea (area of over 1 million hectares). The Baltic Proper harbor porpoise was also placed onto the Swedish National Red list as CR in 2020 (SLU Artdatabanken, 2020).
In 2017, the Swedish National Monitoring Program (SNMP) for porpoises began in the Baltic Sea. This program was designed to utilize 12 of the stations from the SAMBAH project, in order to facilitate a comparison of detection rates over time. The aim of this study was to determine whether there has been a change in the detection rate of Baltic Proper harbor porpoises in Swedish waters over the past decade that could be indicative of higher-level changes occurring within the population.

| Acoustic data collection and processing
The C-POD was used as the passive acoustic data collection instrument. This device contains an omni-directional hydrophone that records the timing of zero-crossings (accuracy to 200 ns) and the peak amplitude between zero crossings (Tregenza, Dawson, Rayment, & Verfuss, 2016). These data are then used to identify the narrow-band highfrequency (NBHF) clicks of harbor porpoises (Au, Kastelein, Rippe, & Schooneman, 1999;Macaulay, Malinka, Gillespie, & Madsen, 2020;Villadsgaard, Wahlberg, & Tougaard, 2007). Data were collected from 12 stations (average depth 44 m, range 29.0-60.0 m) that were used both as a part of the SAMBAH project (April 2011-July 2013) and the SNMP (April 2017-March 2020) ( Figure 1). C-PODs were anchored with the hydrophone approximately 2 m off the bottom, and the loggers were serviced every 3-6 months for battery and SD card changes, and functionality tests. A different C-POD was deployed at the station each time for logistical reasons, and to facilitate removal of systematic bias that could be caused by different sensitivities of the individual C-PODs.
All files downloaded from the C-PODs (including both the SAMBAH and SNMP data) were cut at midnight after deployment, and midnight prior to retrieval. Custom software (CPOD.exe, v. 2.044, https://www.chelonia. co.uk/) was used to process the data. The average and variance of the instantaneous frequency, click duration, peak amplitude, and two measures of the click envelope were saved for each click. The KERNO classifier was then used to identify click trains and label them as either porpoise (NBHF), other cetacean, boat sonar, or unclassified. A second classifier specifically developed for the Baltic Sea marine region (Hel1, Tregenza, 2014) was also applied to the data to reduce the false positive rate, by removing click trains falsely identified as originating from a porpoise. When the number of detection positive minutes was below 60 a year, the file was visually validated to ensure that there were no false detections.

| Potential sources of bias
Diel patterns have been shown in the vocal behavior of harbor porpoises (Carlström, 2005;Osiecka, Jones, & Wahlberg, 2020;Todd, Pearse, Tregenza, Lepper, & Todd, 2009). As there is a theoretical possibility of changes in vocal behavior over time in response to changes in prey availability or quality, we analyzed a range of acoustic metrics to minimize the risk of the results being influenced by behavioral changes over time. The analyzed metrics were: number of encounters, number of clicks, detection positive seconds (DPS) (all presented in Appendix S1), and detection positive hours (DPH) (presented in results) in a day. Stations shown as black dots were used for the yearly trend analysis, as these three stations had the highest levels of detections, are the furthest removed from the neighboring Belt Sea population, and are located on or close to the Northern Midsea Bank; an area known to be of importance for the Baltic Proper population during May-October, encompassing the breeding season Given the extremely low sighting rate, it is not possible to tag Baltic Proper harbor porpoises to collect data on vocal rate over time. However, the use of a range of acoustic metrics (number of encounters, number of clicks, DPS, and DPH, in a day), and ensuring that the results are consistent over these metrics, should minimize the influence of both changes in the acoustic behavior of the animals and diel patterns in the vocal behavior of porpoises over time. Given the low detection rate of Baltic Proper harbor porpoises (max detection of 11 DPH per day), saturation with detections did not influence the results.
There were gaps in data collection over time (due to equipment failure, delayed battery changes due to inclement weather, and unexpected loss of equipment [i.e., being caught in trawl gear]) resulting in varying effort at each of the stations. Based on the results of the SAMBAH study, the Baltic Proper population is thought to congregate into a major cluster during May-October, which is when breeding takes place. During November to April the population has a more dispersed distribution pattern (Carlén et al., 2018). To account for the varying effort and seasonality, missing data were imputed using a seasonal model fitted separately for each station and only detections over the breeding season were used to calculate a yearly population index (see Section 2.3). Regression imputation with a generalized additive model (Wood, 2011) was used, assuming that the number of DPH (or number of encounters, number of clicks, DPS; analysis was repeated for each of the acoustic metrics investigated, see Appendix S1) per day n jy on Julian day j in year y follow a Poisson-distribution with mean λ jy = exp (s(j) + b y ). Here s (j) defines a cyclical spline function, such that s (0) = s(365), which describes the seasonal pattern common to all years, and b y is a fixed yearly effect. The model was fit in R (R Core Team, 2020) with the mgcv package using default options for automatic selection of basis dimension, providing estimatesλ jy that were used to impute missing counts. Further details of the model fitting can be found in the Supporting Information. The proportion of missing days for each station and year is presented in Table S1.
Temperature affects sound propagation in seawater, so the temperature recorded by the C-PODs was examined over the years. Since temperature varies greatly seasonally, and the porpoise data were examined during May-October, the temperature data were examined over the same time period. An equation that calculates how the absorption of sound (dB/km) is influenced by various acoustic (frequency) and environmental (temperature, salinity, depth, acidity) factors from a previous study (Ainslie & McColm, 1998) was used to investigate the potential effect of temperature changes over the years on the detection rate. To do this, we assumed a frequency of 130 kHz (based on the likely frequency of harbor porpoise signals (Macaulay et al., 2020;Villadsgaard et al., 2007)), salinity of 8 ppt (given the brackish waters of the Baltic Sea), depth of 44 m (average depth of the C-PODs), and acidity of 8 pH.
The Baltic Sea does not have any significant tides (a few centimeters) due the small opening to the North Sea, therefore this factor was not considered a potential source of bias. Additionally, characteristics of the station, such as depth and bottom type may influence detection. However, as the same 12 stations were used over time, and trends were examined at each station, it is unlikely that these factors influenced the detection rates recorded over time.

| Temporal trend analysis
In order to investigate how the detection rate of harbor porpoises changed over the years (2011-2019), data from three stations (1032; 1036; and 1041) with 90% of the DPH in this study (across both SAMBAH and SNMP) were selected, as they are also located in an area of high density for this population during May-October when breeding takes place (Northern Midsea Bank, Figure 1) (Carlén et al., 2018). The selection of these stations also ensured that the detections were likely to be for animals from the Baltic Proper population, as stations closer to the proposed May-October management border further to the west (Carlén et al., 2018) are more likely to contain detections that could be from the neighboring Belt Sea population. Using only the data from May to October, a yearly population index was defined as μ y , the arithmetic mean of (possibly partially imputed-see Section 2.2) counts, a measure of the average number of DPH per day. In order to investigate trends over time, log-linear regressions were fitted to yearly indices for each of the three stations. Only five complete years of data were collected over the course of the two studies (SAMBAH 2011(SAMBAH , 2012SNMP 2017SNMP , 2018SNMP , 2019. For the purposes of this study, data were assumed to meet model assumptions (e.g., normality), even though it was not possible to test with five data points.

| Indicators of population trends in abundance
Under EU legislation (European Commission 1992, all countries with harbor porpoises in their waters are required to set regional or sub-regional indicator thresholds that provide information on whether the species has achieved good environmental status for abundance. For the North Sea population of harbor porpoises and other cetacean populations within the OSPAR region, OSPAR has proposed a threshold for trends in abundance, set as a 5% change over 10 years (significance level α < .05) (CEMP, 2019). An abundance indicator is still currently in development for the Baltic Proper population of harbor porpoises within HELCOM. However, we calculated the power to detect a 5% change over 10 years in the Baltic Proper data (using detection rates, not abundance data) at these Swedish stations. These stations represent the area with the highest detection rates in the May-October distribution range of the Baltic Proper population (Carlén et al., 2018), and are therefore, most likely to be able to detect a change. We also calculated the number of years required to have 80% power to detect a 5% change in this region, as this information may be useful for further indicator development. Although our calculations are based on detection rates (not abundance estimates) it is likely that detection rates will need to be utilized as an index of abundance to be used as an indicator for this population within HELCOM over the next two to three EU reporting cycles, and repeatedly also after that. This is due to the fact that obtaining enough updated estimates of absolute abundance for this population to estimate a trend is still likely to be decades away (assuming one abundance estimate every 10-12 years), and even longer before such surveys can be carried out once per six-year reporting cycle.

| Detection of harbor porpoises
A total of 444 detection positive days (DPD) were recorded as a part of the SAMBAH study over 6,422 days of recording (6.9% of days, mean of 0.157 DPH per day) (Table 1). In comparison, 737 DPD were recorded over 8,117 days during the SNMP (9.1% of days, mean of 0.188 DPH per day) ( Table 1). While there were large differences between the detection rates of each station (Table 1, Figure 2), the distribution pattern of detections did not vary greatly between stations across both studies (Figure 2). During both studies, the highest detection rates were recorded in May-October (Figure 3), validating the decision to only use these months, when examining yearly trends (below). Additionally, the May-October peak appeared to be bimodal in distribution at the stations with the most detections (Figure 3).

| Potential sources of bias
The mean water temperature at the stations increased slightly over the years, ranging from 0.9 ± 1.2 C in 2011 to 3.7 ± 1.9 C in 2019, a total change of 2.8 C. This change likely resulted in an increase in absorption of 0.2 dB/km (12.8 dB/km at 0.9 C vs. 13.0 dB/km at 3.7 C; equation from Ainslie & McColm, 1998). With a maximum detection range of 400 m (https://www. chelonia.co.uk), and an effective detection radius well below 100 m, this change in temperature is likely to have had a negligible effect on absorption, and the detection rate of porpoises over time. Even if there was an effect, an increase in absorption over time is likely to have resulted in a decrease in detections of porpoises over time.

| Temporal trend analysis
The mean DPH per day sampled (DPH/day) across the full year was 20% higher during the SNMP compared to SAMBAH, and nine of the 12 stations had higher detection rates (75% of stations, one station had no change, the remaining two showed a decline-these stations also had exceptionally low detection rates) (Table 1). When only considering May-October, mean DPH per day was 29% higher, and 10 of the 12 stations (83%) had higher detection rate (one no change, one showed a decline) ( Table 1).
For three stations (1032; 1036; and 1041), there were enough detections over the course of both studies to estimate a yearly index for all 5 years of May-October data collection. At these stations, the combined trend was a yearly increase in detection rate of 2.4% (95% CI: À4.4, 9.6) ( Figure 4, Table 2). The station with the most detections (1036) appeared the most stable, with only a 0.6% (95% CI: À7.3, 9.2) increase. In comparison, the other two stations (1041 and 1032) showed a much higher yearly increase of 12.6 (95% CI: À4.2, 32) and 15.9% (95% CI: 0.3, 34), respectively ( Figure 4, Table 2). Note that the common signs of trends at separate stations may be an effect of spatial correlation in data; hence, they may not be interpreted as fully independent evidence of increase. It is further worth noting, that in the overall trend, and for some of the stations, the 95% CI also includes a negative growth, indicating that a continued decline is still possible. However, given the 29% increase in detection rates between SAMBAH and the SNMP, we feel that the results are more indicative of either a stall in the decline or a possible increase in detection rates. Additionally, the pattern in % of detections attributed to each station was similar across the two study periods ( Figure 5). Regardless of which acoustic metric was selected (the number of encounters, the number of clicks, DPS, and DPH), the trend of increasing detections over time was observed (see Appendix S1).

| Indicators of population trends
The power to detect a 5% change in detection rates was highest at the station with the most detections (1036, 73%), yet for the other two stations, the power to detect a change was much lower (Table 2). It is estimated that in order to have 80% power to detect a 5% change at each of these stations, between 11 and 17 years of data are required (Table 2). When combing data from all stations the number of years to detect a change reduces to 10 ( Table 2).

| DISCUSSION
We show that the detection rate of harbor porpoises in the Baltic Proper appears to have increased over the last decade in Swedish waters, though evidence of a persisting trend is still weak. Changes in factors such as prey quality or schooling behavior could have influenced the acoustic behavior of porpoises over time. However, the consistent results across all acoustic metrics suggest that the increase is likely the result of more individuals in the area, rather than changes in acoustic behavior. Additionally, the slight increase in absorption due to temperature changes over time is likely to have resulted in a decrease in detections of porpoises over time, not an increase, so this is also unlikely to have caused the observed increase in detection rates. It is also possible that a change in the vertical distribution of sound speed may have influenced the detection rates; however, we feel that this is unlikely to have resulted in the magnitude of increase observed. Therefore, it is likely that the increase in detection rate may indicate a stall in population decline, or the beginnings of population recovery for the Baltic Proper harbor porpoise.
The area monitored in this study represents the area with the highest detection rates during the SAMBAH project (Carlén et al., 2018). The increase in detection rates could indicate a shift in distribution over time, yet the pattern of detections across the stations was relatively stable in both studies (Figure 2, Figure 5). While the proportion of detections at station 1036 has decreased slightly, the number of detections at this station still increased, indicating that there is likely more animals in the area rather than just the same number of animals spreading out. Movement of individuals from the neighboring Belt Sea population into new areas in the Baltic Sea could also explain the increase in detection rates. However, this is unlikely given that we examined the detection rates during May-October when the distributional range of these two populations is thought to be separated (Carlén et al., 2018). Additionally, a potential increase in detection rates has also been observed in recent years in other parts of the Baltic, including Poland (Swistun et al., 2019) and Denmark (Sveegaard, 2020), suggesting that this increase may also be occurring throughout the distributional range of the population. This provides strong support for the need for a new population-wide abundance estimate, in order to confirm F I G U R E 4 Yearly indices with log-linear fitted regression lines (solid line) of harbor porpoises (Phocoena phocoena) detections for three passive acoustic monitoring stations in the Baltic Sea, and an overall yearly index combining these three stations. Note the differing scale of the y-axes between panels, as the detection rate varied greatly among stations. Grey bars indicate 95% confidence intervals T A B L E 2 The estimated yearly trends (%) with 95% confidence intervals of harbor porpoise (Phocoena phocoena) detections at three stations during May-October between 2011 and 2019 in the Baltic Sea. The estimated power to detect a 5% yearly change over 10 years at a significance level of 0.05 with a lower 95% confidence bound, and the number of years required to achieve 80% power to detect a change in detection rate are also shown for each station. The yearly trend and power estimations are also calculated as a summary across all three stations

Station
Yearly trend % Power to detect trend based on 10 years data Years required for 80% power À5% 5% À5% 5% whether we are beginning to see the start of a recovery of this population. The low power to detect changes over time supports the need for continuous monitoring programs, so that any potential changes in abundance is detected. Although detection rates increased over time, the rate of increase (2.4%) is still very low relative to what is likely possible for porpoise populations in the absence of threats. The annual rate of increase for the Morro Bay stock of harbor porpoises has been estimated to 9.6% (95% credible interval [CI] 6.2-13.0%) for the years 1991-2012, after bycatch was mostly eliminated (Forney, Moore, Barlow, Carretta, & Benson, 2020). Additionally, management using the PBR method assumes a possible annual increase rate of 4% across all cetacean species (Punt et al., 2020;Wade, 1998). In the absence of any additional anthropogenic stressors above the current baseline, the Baltic Proper population of harbor porpoises has been estimated to be capable of an annual growth rate of 2.3% (SD 6.4%) (Cervin, Harkonen, & Harding, 2020); similar to the rate observed in this study. Therefore, to enable recovery rates more in line with what porpoise populations are capable of, further removal of anthropogenic threats is required to ensure population recovery, and that the population achieves good environmental status in the future.
It is unknown what may have driven an increase in detection rates over time. Of the threats classified as "high" for the Baltic Proper harbor porpoise (ICES, 2019), a reduction in bycatch risk is most likely to be the most significant factor as it would directly influence mortality rate of the population. Between 2009 and 2018, gillnet fishing reduced by 45% over the entire Baltic Sea (ICES subdivision 24-32; ICES, 2020b), which may have simultaneously lowered the bycatch risk. However, it should be noted that most of the fishing activity takes place outside the seasonal management range of the Baltic Proper population (ICES, 2020b), implying that the extent of reduction within the seasonal distribution range may be different. The true rates of bycatch over the last decade are unknown and poorly reported. Since 2005, pinger use in the Baltic Sea has been regulated by Council regulation (EC) 812/2004 (until 2019), followed by regulation (EU) 2019/1241. Both regulations specify the same two areas in which pingers are mandatory for gillnetters larger than 12 m. In practice, these specifications have no impact on the harbor porpoise bycatch rate as all or almost all gillnetters in the areas are smaller than 12 m. More thorough and dedicated monitoring, and mandatory reporting of bycatch in this region is required.
The increase in detection rates is less likely to be explained by changes in the other high threats, that is, pollution and loud impulsive noise sources (ICES, 2019). Pollution with contaminants has caused reduced reproductive success in Baltic Sea animals (Sonne et al., 2020), and in harbor porpoise populations (Murphy et al., 2015). However, while contaminant levels of harbor porpoises in the Baltic Proper region were historically high (Berggren et al., 1999;Kannan, Falandysz, Tanabe, & Tatsukawa, 1993), current contaminant levels are unknown. The trends of polychlorinated biphenyl (PCB) concentrations in Baltic biota vary, but are overall decreasing since early 2000 (Nyberg et al., 2015). However, they still remain higher than in for example the North Sea, and PCB concentrations in Baltic Sea sediments appear to be at or near steady-state (Sobek, Sundqvist, Assefa, & Wiberg, 2015). F I G U R E 5 Proportion of total yearly observed detection positive hours (DPH) of harbor porpoises (Phocoena phocoena) during May-October attributed to stations in Sweden. DPH from low-intensity stations have been aggregated into category "Other." Note that imputed values are not included; hence, differences may partly be attributed to missing data This low level of reduction is likely to have limited influence on a top predator within one generation length (estimated to 11.9 years for the species under predisturbed conditions for a stable population, Taylor, Chivers, Larese, & Perrin, 2007). The data on loud impulsive noise sources during the last decade in the Baltic Sea is incomplete (ICES, 2021), but the pressure is likely rapidly increasing due to a raising interest in offshore wind power (4 C Offshore, 2021).
The impact of changes in prey availability is difficult to assess, as there is lack of information on the current diet of the Baltic Proper population. Most knowledge is based on historic stomach content analyses, that indicate that the diet is likely to be highly variable (Koschinski, 2001) and different to that of other populations (Benke & Siebert, 1996;Lindroth, 1962). The collection of any specimens of potential Baltic Proper harbor porpoise is essential to assess contaminant loads, diet, and cause of mortality in order to fully understand which factors may be influencing population dynamics.
The Baltic Proper population is thought to congregate on the Northern Midsea Bank (NMB) (Figure 1) during the summer to breed (Carlén et al., 2018). For the Belt and North Sea populations in Danish and Swedish waters, calving peaks in June and mating occurs in late July or August (Börjesson & Read, 2003;Sørensen & Kinze, 1994). The bimodal peaks observed on the NMB in this study (May followed by September/October) may provide some preliminary insight into the breeding behavior of this population, with calving occurring at the first peak (females with calves are thought to communicate the most (Sørensen et al., 2018)) and the second peak indicating the arrival of males for mating. However, the timing is inconsistent with neighboring populations, so further research is required to understand what caused the bimodal peaks in detection.
Similar to the Baltic Proper harbor porpoise population, the vaquita (Phocoena sinus) is a porpoise species that is CR  and is threatened by gillnetting (mostly illegal gillnetting in the case of the vaquita) (Jaramillo- Taylor et al., 2017). In 1997, the population size was similar to the Baltic Proper population (567 individuals [95% CI 177-1,073], Jaramillo-Legorreta, Rojas-Bracho, & Gerrodette, 1999) but this was followed by a steady annual decline (Gerrodette et al., 2011;Jaramillo-Legorreta et al., 2017;Taylor et al., 2017) resulting in fewer than 19 vaquitas remaining by 2018 (Jaramillo-Legorreta et al., 2019). This decline was despite several conservation efforts aimed at removing the threat of bycatch (Rojas-Bracho & Reeves, 2013), including an emergency gillnet ban that cost $74 million USD to compensate fishers (Jaramillo-Legorreta et al., 2019), and an emergency rescue effort to attempt to place two vaquitas in captivity, involving 90 experts from nine countries at a cost of $5 million USD . The continuing decline was likely the result ineffective management that was too late to reverse the decline, a lack of monitoring and compliance in the designated protection areas (Rojas-Bracho & Reeves, 2013), and illegal fishing remaining an ongoing problem .
In contrast, the Morro Bay stock of harbor porpoises in central California demonstrates a more hopeful example of population recovery. This population had declined in abundance; likely due to bycatch in gillnets (Forney, 1995). In order to reduce this threat, various conservation measures were put in place, including regulatory action on the use of pingers, depths of fishing, and large timearea closures to fishing activities (Barlow & Cameron, 2003;Moore et al., 2009), until set gillnet fishing was permanently prohibited in waters shallower than 110 m off central California (Forney et al., 2020). These enforced regulations allowed the Morro Bay stock of harbor porpoises to increase from 571 individuals in 1990 (95% CI 252-2,666) to a population size of 4,191 (95% CI 1,900-11,971) porpoises by 2012 (Forney et al., 2020). These two examples demonstrate that prompt and enforced conservation measures can have a positive influence on population recovery, and that a lack of enforced regulations and management action in protected areas can lead to extinction.

ACKNOWLEDGMENTS
Funding for the Swedish component of SAMBAH study was provided by the EU LIFE+ Programme (LIFE08 NAT/S/000261), the Swedish Agency for Marine and Water Management (SwAM), the Swedish Environmental Protection Agency, and Kolmården Wildlife Park. Funding for the Swedish National Monitoring program was provided by SwAM. The authors would like to thank the many field teams, data management teams, and those involved in securing the funding for both projects.

CONFLICT OF INTEREST
The authors declare no conflict of interest.
AUTHOR CONTRIBUTIONS Kylie Owen and Julia Carlström: Developed the ideas, and collected and processed the data. Martin Sköld: Completed the data analysis with assistance on interpretation from Julia Carlström and Kylie Owen. Kylie Owen: Wrote the manuscript, with strong input on intellectual content from Julia Carlström and Martin Sköld.

DATA AVAILABILITY STATEMENT
The data supporting the study's findings can be accessed by contacting the corresponding author. All code used for the analyses is available on GitHib (https://github.com/ mskoldSU/Owen_et_al_2021).