Marine Biology

, Volume 160, Issue 1, pp 195–209

Population trends of Steller sea lions (Eumetopias jubatus) with respect to remote sensing measures of chlorophyll-a in critical habitat

Authors

    • National Marine Mammal Laboratory, Alaska Fisheries Science CenterNational Marine Fisheries Service, NOAA
  • Lowell W. Fritz
    • National Marine Mammal Laboratory, Alaska Fisheries Science CenterNational Marine Fisheries Service, NOAA
  • Devin S. Johnson
    • National Marine Mammal Laboratory, Alaska Fisheries Science CenterNational Marine Fisheries Service, NOAA
  • Miles G. Logsdon
    • School of OceanographyUniversity of Washington
Original Paper

DOI: 10.1007/s00227-012-2077-4

Cite this article as:
Lander, M.E., Fritz, L.W., Johnson, D.S. et al. Mar Biol (2013) 160: 195. doi:10.1007/s00227-012-2077-4

Abstract

The recovery plan for Steller sea lions (SSL; Eumetopias jubatus) suggests critical habitat should be enhanced to incorporate the spatio-temporal variation in dynamic oceanographic features that influence the prey and survival of SSL. It is necessary, therefore, to determine which features affect SSL. Demographics for sub-regions of the endangered, western stock of SSL were examined with respect to corresponding average, maximum, and variance of chlorophyll-a data (SeaWIFS), a proxy for primary productivity. Overall, SSL trends (2000–2008) and pup productivity (1999–2009) were related to maximum values of chl-a in critical habitat. Additionally, conditions in critical habitat appeared worse in areas of decline (i.e., dispersed patterns of chl-a hotspots and greater distances from SSL sites to productive areas). Although there may be a low feasibility of mitigating the effects of dynamic features on the recovery of SSL, the interactive effects of primary productivity and other stressors should be investigated for safeguarding their prey.

Introduction

From the late 1970s through the 1990s, the endangered, western distinct population segment (wDPS) of Steller sea lions (SSL; Eumetopias jubatus), which occurs west of 144°W (Cape Suckling, AK, USA), declined by 80 % to less than 75,000 animals (Braham et al. 1980; Loughlin 1998; Calkins et al. 1999). Since 2000, populations of SSL within the wDPS have stabilized or increased east of Samalga Pass in the Aleutian Islands (NMFS 2008, 2010), but trends west of this area have been negative (Fritz et al. 2008). Population trends of SSL continue to vary between geographical sub-regions of the U.S. wDPS, including the western (WAI), central (CAI), and eastern (EAI) Aleutian Islands and the western (WGOA), central (CGOA), and eastern (EGOA) Gulf of Alaska, demonstrating the need to employ a recovery strategy that accounts for the spatial and temporal differences of factors currently impeding the sustainability of the species (NMFS 2008).

Among other factors, including fisheries, predation, or a combination thereof (Loughlin 1998; Loughlin and York 2000; Springer et al. 2003; DeMaster et al. 2006; Hennen 2006), environmental variability was also implicated for the precipitous decline in the wDPS during the 1970s because a rapid climatic regime shift transpired from 1976 to 1977, which dramatically altered the ocean structure and physics in the eastern North Pacific and Bering Sea (Mantua et al. 1997; Zhang et al. 1997; Niebauer 1998). Although controversial (Fritz and Hinckley 2005; Newsome et al. 2007), there was some evidence to indicate that primary production decreased in parts of Alaska after the 1976 regime shift (Sugimoto and Tadokoro 1997; Schell 2000; Hirons et al. 2001; Trites et al. 2007), which may have altered the distribution, abundance, availability, recruitment, or quality of important prey items for SSL (Trites et al. 2007). On a smaller scale, it was also suggested that the distribution of prey patches for SSL may have been altered as a result of reduced localized production (Ferrero and Fritz 2002) coupled with fishery removals (NMFS 2008). For these reasons, environmental variability was identified as a “potentially high threat” to the recovery of the wDPS, and one proposed recovery action within the revised recovery plan for Steller sea lions includes a revised critical habitat designation based on the spatial and temporal variation in essential habitat characteristics (NMFS 2008). The recovery plan also states that in addition to stationary habitat features (i.e., bathymetry and the continental shelf), dynamic oceanographic features that influence sea lion prey should be considered (NMFS 2008). It is necessary, therefore, to determine which oceanographic features influence sea lions and how those features persist over time.

An understanding of any marine ecosystem must be based on appropriate oceanographic features, but all marine communities are influenced by food availability, which ultimately arises from primary productivity via bottom–up trophic interactions through the food web (Dayton et al. 1994). Thus, features associated with primary productivity may affect the integrity of ecological units that constitute critical habitat for marine mammals (Harwood 2001), possibly having indirect consequences on their population demographics. Although predicting the response of populations to changes in primary productivity can be challenging (Brown et al. 2010), population response to chlorophyll-a (chl-a), a proxy for primary productivity, has been documented for sea birds (i.e., reproductive success; Monticelli et al. 2007; Laidre et al. 2008) and fish (Ware and Thomson 2005; Frank et al. 2006). For example, Ware and Thomson (2005) found that primary productivity, as measured using SeaWIFS chl-a data, was positively correlated with secondary productivity and that both variables were also correlated with fish yield within fishing zones along the continental margin of western North America (i.e., Shumagin Islands, AK, to Pt. Conception, CA, USA). However, similar linkages have not been as apparent for SSL, possibly due to limited sample sizes, seasonal effects, lags in response time, uncertain relationships between primary and secondary producers, and the interaction between bottom–up and top–down mechanisms (Lander et al. 2009). The objective of this study was to expand the spatial and temporal scales of previous studies (e.g., Lander et al. 2009) and examine pup production and regional population trends of SSL with respect to remote sensing measures of annual chl-a in designated critical habitat of the wDPS. After identifying maximum chl-a as a possible predictor of population trajectory, the spatial arrangement of this variable within each sub-region of the wDPS was investigated. Lastly, proximity of sea lion sites (i.e., rookeries and haulouts) to productive areas was also examined across the range of the wDPS.

Methods

Remote sensing data

To examine relationships between SSL demographics and primary productivity, level 3 standard mapped images (9 km resolution) of seasonal chl-a composites (summer day of year (DOY): 172–263; fall DOY: 264–354; winter DOY: 355–79; spring DOY: 80–171) collected by the SeaWIFS sensor from 1999 to 2009 (processing versions R5.2 and 2010 for 1999–2008 and 2009, respectively) were obtained from the Ocean Color Discipline Processing System (oceancolor.gsfc.nasa.gov). Chlorophyll-a data were not available for winter 2007. All remote sensing data were converted to raster grids, projected to an Albers equal-area conic projection (ArcInfo, ESRI, Redlands, CA, USA), and clipped to polygons constituting designated SSL critical habitat (50 CFR 226.202), including 20 nautical mile (nmi) buffers around all major haulouts and rookeries, three large offshore foraging areas around Shelikof Strait, Bogoslof, and Seguam Pass, and small protection zones (Fig. 1, ArcMap 9.3, ESRI, Redlands, CA, USA). For each clipped seasonal composite, summary statistics (mean, maximum, and variance) of chl-a were computed for six geographical regions of the wDPS as described in the literature (Fig. 1; NMFS 1992; Fritz et al. 2008). Results (excluding maximum statistic) were then averaged to form annual datasets (summer to spring) to coincide with SSL surveys conducted during the breeding season (June to early July).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-012-2077-4/MediaObjects/227_2012_2077_Fig1_HTML.gif
Fig. 1

Steller sea lion (SSL) critical habitat (50 CFR 226.202, including all protection zones) within six geographical sub-regions of the western distinct population segment of SSL: the western (WAI), central (CAI), and eastern (EAI) Aleutian Islands and the western (WGOA), central (CGOA), and eastern (EGOA) Gulf of Alaska. Critical habitat is outlined in gray, and points indicate SSL rookeries and haulouts (N = 299)

Steller sea lion counts

Counts of SSL non-pups (adults and juveniles) and new-born pups were compiled from four sources (Sease and Gudmundson 2002; Fritz and Stinchcomb 2005; Holmes et al. 2007; Fritz et al. 2008). In summary, non-pups were counted at trend sites during aerial surveys conducted biennially from 2000 to 2008 by the National Marine Fisheries Service, whereas pups were counted at trend sites using a combination of aerial and shore-based surveys during 5 years (2000, 2002, 2004, 2005, and 2009). During 2006, counts of non-pups only were obtained for the EAI and EGOA, whereas during 2000, counts of pups only were obtained for the CGOA, WGOA, and EAI due to logistics. For each year, counts from trend sites were separately pooled within the six geographical regions outlined above.

Annual rates of population change within each region from 2000 to 2008 were calculated from the regression coefficients of log-linear regressions of the natural logarithm of the non-pup counts on the survey years (Fritz and Stinchcomb 2005). These data were then examined with respect to three corresponding measures of annual chl-a concentration (i.e., average, maximum, and variance) averaged over the 9 years using regression analyses. Additionally, annual measures of SSL productivity (ratio of pups to non-pups) for each region were examined with respect to the three measures of corresponding chl-a for the previous year using regression analyses. For example, indices of SSL productivity for 2000 were examined with respect to annual composites of chl-a ranging from June 1999 to June 2000. For significant relationships, Akaike’s information criterion (AIC) was used for model selection (i.e., linear vs. logarithmic or quadratic relationships; R 2.4.1, R Development Core Team 2006).

Hotspot persistence

After identifying maximum chl-a as an important predictor of population regulation (see below), the maximum cell within each region (defined hereafter as “hotspot”) was isolated for each year to examine the spatial patterns of hotspots within critical habitat. If more than one cell contained the maximum value, a cell was randomly chosen for that year. After pooling hotspot cells across years within each region, average nearest neighbor (ANN) indices were calculated as a measure of hotspot persistence among all years using the following metric:
$$ {\text{ANN}} = \frac{{\frac{{\sum\nolimits_{i = 1}^{n} {d_{i} } }}{n}}}{{\frac{0.5}{{\sqrt {{n \mathord{\left/ {\vphantom {n A}} \right. \kern-0pt} A}} }}}} $$
where di is the distance between the centroid of hotspot i and the centroid of its nearest neighbor, n is the total number of hotspots, and A is the total study area of the region. This index is expressed as the ratio of the observed mean distance divided by the expected distance, which is based on a hypothetical random distribution with the same number of features covering the same total area (ArcMap 9.3). A distribution of hotspots was considered clustered, and hence spatially persistent, if the average distance was less than the average for a hypothetical random distribution, whereas a distribution of hotspots was considered dispersed if the average distance was greater than a hypothetical random distribution.

Proximity to productive waters

To examine the proximity of terrestrial SSL sites to productive waters, maximum values of seasonal chl-a composites were used to form annual composites for the 10 years as detailed above and Euclidean distance was measured from each haulout and rookery across the range of the wDPS (N = 17 sites in the WAI, N = 93 sites in the CAI, N = 59 sites in the EAI, N = 42 sites in the WGOA, N = 58 sites in the CGOA, N = 30 sites in the EGOA) to the nearest chl-a “bloom”, where a bloom was defined as 1.0 mg m−3 (Yoo et al. 2008, ArcMap 9.3). A geostatistical maximum likelihood with an exponential covariance function and a common sill and range parameter over all 10 years was used to examine the predictor variable distance with respect to the response variables region and year (R 2.4.1). To further elucidate how chl-a blooms differed between regions, a Kruskal–Wallis test was used to compare the mean proportion of cells designated as blooms (N = 10 year) across regions after conducting a Levene’s test for equal variances (SPSS 13.0). A multiple comparisons test (i.e., Dunnett T3 for unequal variances) was used to examine differences between mean proportions for the six regions.

Results

Population trends, pup production, and chlorophyll-a

Rate of population change of non-pups from 2000 to 2008 ranged from −0.07 for the WAI to 0.05 for the EGOA and was positively related to annual values of maximum chl-a averaged over that time frame (r2 = 0.795, F1,4 = 15.481, P = 0.017; Fig. 2). Additionally, annual rates of population change for SSL displayed a unimodal dome-shaped function of the average variance of chl-a (r2 = 0.964, F2,3 = 39.614, P = 0.007; Fig. 3). From 2000 to 2009, pup production (ratio of pups to non-pups) ranged from 0.30 in the WAI to 0.67 in the CGOA (WAI: \(\overline X\) ratio ± SD = 0.39 ± 0.09; CAI: \(\overline X\) ± SD = 0.57 ± 0.02; EAI: \(\overline X\) ± SD = 0.55 ± 0.06; WGOA: \(\overline X\) ± SD = 0.54 ± 0.01; CGOA: \(\overline X\) ± SD = 0.62 ± 0.04; EGOA: \(\overline X\) ± SD = 0.50 ± 0.04). Indices of annual pup production pooled across regions displayed a significant, positive relationship with respect to annual values of maximum chl-a (r2 = 0.181, F1,25 = 5.539, P = 0.027; Fig. 4), whereas relationships with other measures of chl-a (mean or variance) were not significant (P > 0.050).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-012-2077-4/MediaObjects/227_2012_2077_Fig2_HTML.gif
Fig. 2

Rate of Steller sea lion (SSL) population change (non-pups) from 2000 to 2008 relative to the average of annual maximum values of chlorophyll-a (chl-a) in corresponding sub-regions of critical habitat over that time period. Numbers in parentheses denote the SD of maximum chl-a values, and symbols represent sub-regions of the western distinct population segment of SSL: the western (WAI), central (CAI), and eastern (EAI) Aleutian Islands and the western (WGOA), central (CGOA), and eastern (EGOA) Gulf of Alaska

https://static-content.springer.com/image/art%3A10.1007%2Fs00227-012-2077-4/MediaObjects/227_2012_2077_Fig3_HTML.gif
Fig. 3

Rate of Steller sea lion (SSL) population change (non-pups) from 2000 to 2008 relative to the average variance of chlorophyll-a (chl-a) in corresponding sub-regions of critical habitat over that time period. Sub-regions include the western (WAI), central (CAI), and eastern (EAI) Aleutian Islands and the western (WGOA), central (CGOA), and eastern (EGOA) Gulf of Alaska

https://static-content.springer.com/image/art%3A10.1007%2Fs00227-012-2077-4/MediaObjects/227_2012_2077_Fig4_HTML.gif
Fig. 4

Relationship between annual pup productivity of Steller sea lions and annual values of maximum chlorophyll-a (chl-a) in corresponding sub-regions of critical habitat: the western (WAI), central (CAI), and eastern (EAI) Aleutian Islands and the western (WGOA), central (CGOA), and eastern (EGOA) Gulf of Alaska. Data points (N = 27) were from 2000, 2002, 2004, 2005, and 2009

Nearest neighbor analyses

After pooling the ten annual maximum chl-a locations (i.e., hotspots; Fig. 5) within each region, nearest neighbor analyses indicated hotspots were significantly dispersed (P < 0.050) in the WAI and CAI, whereas the spatial distribution of hotspots for the other regions displayed a random distribution (P > 0.050; Table 1). We assumed results were biased negative because the equation used to calculate the ANN index, Z score, and P value are based on the assumption that there are no barriers (i.e., land) and the points being measured are free to locate anywhere within the study.
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-012-2077-4/MediaObjects/227_2012_2077_Fig5a_HTML.gif
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-012-2077-4/MediaObjects/227_2012_2077_Fig5b_HTML.gif
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-012-2077-4/MediaObjects/227_2012_2077_Fig5c_HTML.gif
Fig. 5

Ten annual composites (summer to spring) of maximum chlorophyll-a (chl-a; mg m−3) data corresponding to Steller sea lion (SSL) critical habitat. Pink stars represent regional hotspots (absolute maximum cell), and gray hatch marks (plus) indicate cells defined as a chl-a bloom (>1 mg m−3). Sub-regions of the western distinct population segment of SSL are indicated on the map, including the western (WAI), central (CAI), and eastern (EAI) Aleutian Islands and the western (WGOA), central (CGOA), and eastern (EGOA) Gulf of Alaska

Table 1

Nearest neighbor indices (ratio of observed mean distance (OMD) to expected mean distance (EMD)) and associated Z score and P values for chlorophyll-a (chl-a) hotspots in six sub-regions of the western distinct population segment of Steller sea lions, including the western (WAI), central (CAI), and eastern (EAI) Aleutian Islands and the western (WGOA), central (CGOA), and eastern (EGOA) Gulf of Alaska

Region

OMD/EMD

Z score

P

WAI

1.489

2.957

0.003*

CAI

1.870

5.262

0.000*

EAI

1.161

0.976

0.329

WGOA

1.300

1.809

0.070

CGOA

1.264

1.597

0.110

EGOA

0.939

−0.368

0.713

Asterisks denote significance (P < 0.050)

Distance analysis

Chlorophyll-a blooms occurred within critical habitat every year (Fig. 5), but distance measures were more variable in the Aleutian Islands than in the Gulf of Alaska (Fig. 6). The geostatistical regression model indicated distances from SSL sites to nearest cell defined as a bloom were significantly greater in the CAI than in any other region (Table 2). Additionally, the proportion of cells in critical habitat defined as blooms (WAI: \(\overline X\) proportion ± SD = 0.73 ± 0.17; CAI: \(\overline X\) ± SD = 0.57 ± 0.15; EAI: \(\overline X\) ± SD = 0.89 ± 0.08; WGOA: \(\overline X\) ± SD = 0.96 ± 0.09; CGOA: \(\overline X\) ± SD = 0.96 ± 0.08; EGOA: \(\overline X\) ± SD = 0.95 ± 0.13; Fig. 7) differed significantly between regions (H5 = 39.153, P = 0.000). A post hoc analysis indicated mean proportion of cells in critical habitat containing blooms in the CAI was significantly less than data for all other regions except the WAI (P = 0.353). In turn, data for the WAI were significantly less than data for the WGOA (P = 0.026) and CGOA (P = 0.022). Data did not differ significantly among EAI, WGOA, CGOA, and EGOA and were fairly consistent across years (Fig. 7).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-012-2077-4/MediaObjects/227_2012_2077_Fig6_HTML.gif
Fig. 6

Mean distance (km) from Steller sea lion (SSL) sites to nearest chlorophyll-a (chl-a) bloom (>1.0 mg m−3) for each sub-region of the western distinct population segment of SSL (denoted by different colors) over 10 years. Data are based on annual composites of maximum chl-a, and error bars indicate SD

Table 2

Results of a geometric regression used to examine distance from Steller sea lion sites to nearest chlorophyll-a (chl-a) bloom with respect to sub-regions of the western distinct population segment of Steller sea lions and year

Covariate

Estimate

SE

Z

P

(Intercept)

8.827

0.122

72.618

0*

Region CGOA

−0.705

0.095

7.388

1.490E−13*

Region EAI

−0.385

0.104

3.700

0.000*

Region EGOA

−0.573

0.116

4.962

6.980E−07*

Region WAI

−0.350

0.156

2.249

0.024*

Region WGOA

−0.749

0.108

6.917

4.610E−12*

Year 2 (2000–2001)

−0.265

0.154

1.723

0.085

Year 3 (2001–2002)

−0.182

0.154

1.185

0.236

Year 4 (2002–2003)

−0.228

0.154

1.481

0.139

Year 5 (2003–2004)

−0.208

0.154

1.351

0.177

Year 6 (2004–2005)

−0.255

0.154

1.656

0.098

Year 7 (2005–2006)

−0.275

0.154

1.784

0.074

Year 8 (2006–2007)

−0.035

0.154

0.229

0.819

Year 9 (2007–2008)

−0.017

0.154

0.109

0.913

Year 10 (2008–2009)

−0.020

0.154

0.131

0.896

Regions include the western (WAI), central (CAI), and eastern (EAI) Aleutian Islands and the western (WGOA), central (CGOA), and eastern (EGOA) Gulf of Alaska

Asterisks denote significance (P ≤ 0.050)

https://static-content.springer.com/image/art%3A10.1007%2Fs00227-012-2077-4/MediaObjects/227_2012_2077_Fig7_HTML.gif
Fig. 7

Proportion of cells defined as chlorophyll-a (chl-a) blooms (>1.0 mg m−3) in critical habitat (CH) for each sub-region of the western distinct population segment of Steller sea lions (denoted by different colors) over 10 years. Data are based on annual composites of maximum chl-a

Discussion

Over the 10-year time series examined for this project, the spatial structure of chl-a data varied geographically within designated critical habitat for SSL. Similar to other patterns reported for the North Pacific and Bering Sea, spatial patterns of chl-a in critical habitat displayed a longitudinal dipole (Ladd et al. 2005a; Mordy et al. 2005), with surface chl-a concentrations in the Aleutian Islands being less than those in the Gulf of Alaska. Temporal patterns of phytoplankton production reported for the Bering Sea as a whole are generally characterized by a spring bloom (April–June), a summer decrease (July–August), and a fall bloom (September–October; Iida and Saitoh 2007). In contrast, maximum chl-a values in critical habitat primarily occurred during spring (38 %), followed by summer (33 %), fall (27 %), and winter (2 %), supporting the notion that seasonal breeding of SSL is synchronized with times of greater productivity, which may coincide with peak food availability for adults (Ims 1990; Boyd 1991). Collectively, it appeared that large pulses of primary production in critical habitat, as indicated by annual maximum chl-a values, were likely beneficial for SSL population growth over the past decade.

Variations in the environment can potentially lead to contrasting ecological effects over relatively small geographical regions (Stenseth et al. 2002). To some extent, regional population trends of SSL reflected the longitudinal pattern of primary productivity in critical habitat, but this pattern was more apparent for regions in the Aleutian Islands than it was for regions in the Gulf of Alaska. In general, population trends of SSL and the environment as characterized by primary productivity were markedly different in the west than in the east, with conditions appearing better for sea lions in the east. Chlorophyll-a concentrations as well as the proportion of chl-a blooms in the Gulf of Alaska exceeded those in the Aleutians. Furthermore, areas defined as hotspots over time displayed dispersed patterns in regions of population decline (WAI and CAI) in the west. Distances from rookeries and haulouts to other areas defined as chl-a blooms also were greater for the CAI. Although the functional relevance of the productive areas we identified was unknown, sea lions from western regions were at a disadvantage if in fact those productive regions served as a source of prey aggregation.

Primary production is necessary for maintaining biodiversity, and supporting offspring production and overall population production (Monticelli et al. 2007; Witman et al. 2008; Brown et al. 2010); thus, declines in primary productivity have been implicated as causes of population decline for many species, including pinnipeds. For example, it was suggested that a divergence in population recovery rates between two sub-species (Arctocephalus spp.) was due to the large differences in primary productivity of the marine environments the two species occupied (Warneke and Shaughnessy 1985). A more recent study found that pup survival and population success of two different Zalophus spp. corresponded to latitudinal differences and seasonal variation of primary production (Villegas-Amtmann et al. 2011). Our results also indicated that SSL population trends and pup production (albeit marginal) were related to primary productivity in critical habitat, supporting larger-scale studies conducted for this species (Guénette et al. 2006; Trites et al. 2007; Nesse 2009). For example, Guénette et al. (2006) used the Pacific decadal oscillation (PDO) as an index of ocean productivity in an ecosystem model and found that changes in ocean productivity explained a large part of the SSL decline in the Aleutian Islands from 1963 to 2000. Additionally, Nesse (2009) found the pup-to-adult survivorship at six sites within the wDPS through the 1990s was partially predicted by the June PDO. It remains important, therefore, to understand the effects of oceanographic bottom–up processes on SSL populations in the face of climate change.

Population decline coupled with lower pup production in the WAI coincided with lower primary production in this area. These data are suggestive of a decrease in natality for this region, though proximate and ultimate causes are unknown. However, mammals generally have lower rates of production with less abundant or reliable food sources (Sibly and Brown 2009), and food limitation has long been suspected for the decline of the wDPS of SSL. NMFS (2010) found SSL population trends in the WAI from 2000 to 2009 were negatively associated with the relative intensity of fisheries and concluded that low ecosystem productivity in that area only exacerbated the availability of important prey. Because fishing reduces the age, size, and geographical diversity of fish populations, they may become more sensitive to additional stressors such as environmental change (Brander 2007). Unfortunately, this appears to be the case for Atka mackerel (Pleurogrammusmonopterygius) in the western range of the Aleutian archipelago, where size at age, growth rates, and prey quality were less than those in eastern areas (Lowe et al. 1998; Rand et al. 2010). Because Atka mackerel is the primary prey of SSL in the WAI (Sinclair and Zeppelin 2002; Lander et al. 2009), the interrelated effects of fishing and oceanographic change (including that of primary production) on this species should be examined in more detail.

After excluding data for the WAI, chl-a did not explain the variation in pup production very well, which was fairly consistent spatially (i.e., across regions) and temporally (i.e., no significant change in pup production from 2000 to 2009 for each region) across the range of the wDPS. When these data were cross-examined with non-pup data, they appeared somewhat inconsistent with hypothesized causes of recent decline (i.e., decreased natality; Holmes et al. 2007). For example, although the non-pup population of the CAI declined over the past decade, pup productivity in the CAI was similar to the other increasing sub-populations. Likewise, pup productivity was greatest in the CGOA, which also did not experience any population growth. One explanation for these discrepancies may be that the index of pup production we used (i.e., ratio of pups to non-pups) was not an actual measure of natality because juveniles and adult males were included in the index and the abundance of these age classes may have differed among regions. Another explanation may be that adult females, which are tending pups during the peak breeding season, are capable of modifying their short-term performance (via physiological or behavioral-mediated responses) to a range of primary production beyond a lower limit. Published accounts of telemetry data indicated that adult female SSL did not push their physiological limits during the breeding season, nor did they push the fasting limits of their pups (Merrick and Loughlin 1997; Brandon 2000; Rehberg et al. 2009), supporting the idea that females may have the ability to alter their behavior to accommodate changing conditions (Rehberg et al. 2009). However, the extent of this plasticity is unknown (Rehberg et al. 2009), especially in relation to oceanographic features and the long-term effects on sea lion dynamics.

Although there is growing evidence that effects of primary production can be propagated up food webs to higher marine predators (Frederiksen et al. 2006), predicting the response of populations to changes in primary production can be complicated by nonlinear, threshold responses to prey availability, predation, and competition interactions (Piatt et al. 2006; Brown et al. 2010). For example, Lander et al. (2009) examined regional SSL trends with respect to corresponding metrics of summer chl-a data in study areas demarcated from satellite-tagged individuals and proposed that competition may have been responsible for the hump-shaped pattern observed; this pattern was driven by the decline of sea lions in the CGOA, despite being a highly productive area. With the addition of two regions during this study, different patterns emerged with different measures of chl-a, but the CGOA still appeared to be an outlier when relating SSL population trends to primary productivity. These findings tend to coincide with inconsistencies observed for walleye pollock (Theragra chalcogram), the dominant prey of Steller sea lions in the CGOA (Sinclair and Zeppelin 2002; Lander et al. 2009). For example, Hollowed et al. (2007) found the distribution of walleye pollock around Kodiak Island coincided with primary productivity, yet Walline et al. (2012) found that high density patches of pollock in similar areas were rare. Although it is believed that oceanographic processes drive the distribution of pollock and their prey in this area, Logerwell et al. (2010) proposed a better understanding of interspecific competition among forage fish species would lead to a better understanding of pollock productivity. Ultimately, an assessment of spatio-temporal match–mismatch from primary producers to higher trophic levels, including sea lions and their prey, will be necessary for predicting the impact of environmental change and extreme perturbations on these populations (Grémillet et al. 2008).

From an ecological standpoint, extreme events in climate indices (e.g., natural event regimes, thresholds of environmental variables, and periodic pulses of productivity) often are more relevant in influencing life history thresholds than are fluctuations in mean indices (Mantzouni and MacKenzie 2010) because extreme events can accelerate system changes by reducing inertia, which is represented in long-lived organisms (Jentsch et al. 2007). In addition to having the potential to shift populations and ecosystems into new equilibria or states (Easterling et al. 2000), extreme events may also weaken or strengthen food web dynamics by impacting key prey or predators of a given species (Mantzouni and MacKenzie 2010). During this study, SSL population trends and pup production both appeared buffered when examined with respect to the mean of chl-a, possibly because the conditions represented by this statistic were within tolerable ranges of sea lions, enabling them to readily adapt. However, the adaptive capacity of sea lions may have been compromised or heightened when exposed to conditions outside the range of average primary productivity. Because the IPCC (2007) forecasts an increase in magnitude and/or frequency of climatic events in the future, the magnitude (including fluctuations in magnitude) of primary productivity and corresponding effects on Steller sea lions should be investigated in more detail because the maximum values used for our analyses were obtained from seasonal composites (averaged over three monthly means) and did not reflect the true magnitude of extreme conditions. Retrospectively, when non-pup population trends and pup productivity were examined with respect to chl-a measures using the annual composites based on the maximum values of seasonal composites (i.e., those used for distance analyses), nonlinear (i.e., quadratic) relationships were evident for the average variance (r2 = 0.901, F2,3 = 13.581, P = 0.031) and annual mean (r2 = 0.242, F2,24 = 3.823, P = 0.036) of chl-a, respectively.

Conservation implications

It has been suggested that the condition of some populations may be dependent on their sub-population connectivity (Bishop et al. 2010). Connectivity is broadly defined by the spatial continuity of corridors, which act as conduits for movement of organisms, and can be measured at different levels ranging from genetic connectivity of species to seascape habitat components (Forman and Godron 1986). Habitat connectivity generally refers to the functional linkage among habitat patches either because those patches are of similar habitat type, physically adjacent, or because the dispersal abilities of the organisms of interest effectively connect patches (O’Neill et al. 1988; Gardner et al. 1993). In the case of protected areas, habitat connectivity should support natural dispersal and ensure a network resilient to changes (Sundblad et al. 2011). Because many fish (i.e., prey) species use separate locations for spawning, larval development, and juvenile or adult feeding, it is necessary that spatially segregated locations are successfully connected to alleviate the effects of fishing pressure and/or climate on those critical links (Bakun 2010). Taking this into consideration, it is interesting to note that population declines experienced by SSL in the WAI and CAI west of 178°W coincide with the fragmentation of critical habitat, which otherwise is spatially connected across the range of the wDPS. Breaks in critical habitat occur in the western portion of the wDPS at Near Strait, Buldir pass, and Amchitka Pass (Fig. 8). However, Near Strait and Amchitka Pass were generally characterized by chl-a blooms. Atka mackerel are known to form dense aggregations in island passes and areas of strong currents (McDermott et al. 2005). Even if Atka mackerel do not congregate in island passes as a direct response to the primary productivity that typically characterizes these areas, they can be easily exploited by fishers with access to real-time ocean color data that are commercially available (Simpson 1992). High movement rates of tagged Atka mackerel from inside to outside trawl exclusion zones at Amchitka Island (Lowe and Fritz 1997) suggest they are especially vulnerable to the fishery in this area (NMFS 2010).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-012-2077-4/MediaObjects/227_2012_2077_Fig8_HTML.gif
Fig. 8

Locations within Alaska referenced in the “Discussion

Prey resources, which are highly dynamic and influenced by environmental variability and oceanographic conditions, are the most essential feature of marine critical habitat for SSL (NMFS 2010), and management of areas containing environmental features that are important to prey will ultimately entail having a clear understanding of their underlying processes. It has been suggested that there are three major classes of physical processes that combine to yield favorable fish habitat including enrichment processes, concentration processes, and processes favoring retention within appropriate habitat (Bakun 2010). In some regions of the wDPS, critical habitat manages to capture some of these processes. In the Gulf of Alaska, productive areas in critical habitat were likely associated with cross-shelf exchange induced by eddies, downwelling episodes, and tidal mixing over banks and through channels (Ladd et al. 2005b). In the EAI, chlorophyll blooms of substantial concentration were commonly observed north of Ogchul and Unalaska in the three forage boxes that straddle the division between the Bering Sea Shelf and the Bering Sea Basin (Fig. 8). In this area, summer production and walleye pollock productivity also tend to be associated with fronts that characterize the shelf-break region north of Unimak Pass in critical habitat (Sambrotto et al. 2008). However, unlike Unimak Pass and other shallow eastern passes, phytoplankton production is depressed in the western Aleutian passes due to extensive cloud cover and deep wind-mixing (Schell et al. 1998) and vertical mixing sometimes ceases to exist in Amchitka Pass during summer (Stabeno et al. 2009). Hence, in the absence of enrichment processes (e.g., upwelling, mixing), dynamic processes (e.g., concentration processes, such as convergence and frontal formation) or retention processes that produce favorable fish habitat tend to be vital (Bakun 2010) and should possibly be considered for protection. For example, in the WAI, areas around Buldir where the Bering Sea shelf intercepts the Bering Sea Basin in the north and the North Pacific shelf to the south appeared to be more productive. Additionally, chl-a composites indicated that areas outside of critical habitat associated with the Alaska Stream south of Attu and Agattu were richer in chl-a; findings by Himes Boor and Small (2012) and some telemetry data (Lander et al. 2009; Lander et al. 2011) support the idea that these areas and the area south of Buldir Pass may be important habitat for SSL. Similar to the WAI, areas in the CAI associated with the Alaska stream just south of critical habitat appeared richer in chl-a (e.g., blooms and greater values). Results warrant caution, however, due to the problems associated with satellite remote sensing data in coastal areas (Ruddick et al. 2000) and the uncertainty of sub-surface processes undetected by satellites.

Overall, results support the idea that increased primary production was probably advantageous for SSL, but mechanistic links between the physical forcing processes responsible for chl-a hotspots and their spatial associations with higher trophic levels remain uncertain. However, it is likely that primary production influenced the demographics of SSL over broader temporal scales, and features associated with this process warrant additional research. Although the recovery plan for SSL specifies there is a low feasibility of mitigating the effects of environmental variation on their recovery, some recommendations proposed by other researchers can be applied if considering the inclusion of dynamic oceanographic features in a revised critical habitat designation. For example, Hyrenbach et al. (2000) provided a comprehensive viewpoint detailing some guidelines for protecting static, predictable, and ephemeral oceanic features. Specific to our area of interest, Piatt et al. (2006) found a link between the distribution of short-tailed albatross (Phoebastria albatrus) to permanent topographic, yet also dynamic, features such as shelf-edges and passes in the Aleutian Islands (i.e., Near Strait, Buldir Pass, Samalga Pass, and Seguam Pass) and recommended that threatening activities (e.g., fishing, oil and gas development) be regulated in those areas. Given the sensitivity of the current conservation measures (i.e., fishery closures) currently enforced in some areas of the wDPS, an adaptive management approach (Hooker et al. 2011) that is timely, flexible, and based on the interactive effects of fishing and environmental variability (Brander 2007) may be an option for supplementing the precautionary principles governing the protection of SSL habitat.

Acknowledgments

We thank the entire staff of the Alaska Ecosystems Program at the National Marine Mammal Laboratory for their assistance with data collection, especially K. Sweeney. This manuscript was improved with constructive reviews by G. Duker, E. Gurarie, B. Fadely, T. Gelatt, K. Laidre, E. Sinclair, S. McDermott, and one anonymous reviewer. This work was conducted under Federal Marine Mammal Permits 358–1564 and 782–1532. The use of trade, product, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government. The findings and conclusions in the paper are those of the authors and do not necessarily represent the views of the National Marine Fisheries Service.

Copyright information

© Springer-Verlag (outside the USA) 2012