Study Area

Gulf Coast. The Laguna Madre of Texas, USA, is a 365,800-ha hypersaline lagoon, averaging ∼1 m in depth (Fig. 1A). It contains extensive areas of seagrass (Halodule wrightii, Ruppia maritima, Syringodium filiforme, Thalassia testudinum) and wind tidal flats (Tunnell and Judd 2002). Laguna Madre is bordered on the east side by a barrier island, including the Padre Island National Seashore and South Padre Island (Tunnell and Judd 2002). On the mainland side, undeveloped rangeland dominates the landscape, but other land use includes agricultural fields, wind farms, and suburban developments. The Rio Grande River Delta is located between Laguna Madre, Texas, USA, and Laguna Madre de Tamaulipas, Tamaulipas, Mexico, and is composed of coastal wetlands, distributary channels from the Rio Grande, oxbow and playa lakes, and wind tidal flats (Tunnell and Judd 2002). The Laguna Madre of Tamaulipas is 570,000 ha (Fig. 1A), is a Natural Protected Area and supports breeding and wintering Reddish Egrets.

The Laguna de Términos is a 750,032-ha lagoon in Campeche, Mexico (Fig. 1B). It was designated a nature reserve in 1994 and is listed as an Important Bird Area (Vidal et al. 2009) and as a wetland of international importance by the Ramsar Convention. Although the reserve is largely protected, the conversion of forest to cattle (Bos taurus) ranches has increased along its borders. It is also threatened with degradation due to cutting of mangroves (Rhizophora mangle, Avicennia germinans), road construction, and other human-related causes (Bach et al. 2005).

Hagens Cove is located on the Gulf Coast of Florida, USA (Fig. 2A). It is nearly 54 ha of hardpan, shallow, tidal cove with a variety seagrasses. The site is especially important for Reddish Egrets in the fall and winter months, where as many as 22 individuals have been seen foraging at one time. Sanibel Island and Pine Island are located on the central Gulf Coast of Florida (Fig. 2B). The area has a series of tidal impoundments where Reddish Egrets gather at the outflow culverts at low tide as fish concentrate over the hardpan sand fan created by the concentrated water currents. Boca Grande Key is a 74-ha island that is in the Florida Keys (Fig. 2C). An extensive tidal lagoon runs north to south on the west side of the key creating ideal forage for many species of birds that may nest in the mangroves on the eastern side or that use the key during the non-breeding season. Cudjoe Key, also part of the Florida Keys, is a 1,450-ha island located ∼35 km east of Boca Grande Key (Fig. 2C). Much of Cudjoe Key has shallow salt water ponds in remote areas.

Pacific Coast. Laguna del Mar Muerto is on the southern coast of the Tehuantepec Isthmus in Oaxaca and Chiapas, Mexico, and encompasses 70,000 ha (Fig. 3A). A survey of breeding colonies along the western coast of Mexico found that Laguna del Mar Muerto supported the largest breeding colony of Reddish Egrets on the Pacific Coast, with ∼258 pairs (Palacios et al. 2010). The Rio Verde is a 342-km-long river located in Oaxaca, Mexico (Fig. 3A). The Rio Verde basin covers 2,586 km² and spans ∼64 km from east to west between two mountain ranges. The habitat was described by Binford (1989) as arid tropical scrub, tropical deciduous forest, and arid pine-oak (Pinus spp., Quercus spp.) forest. There is little information available on habitat or avifauna for the river system, although Binford (1989) describes bird sightings in the area. Lastly, one individual in this study wintered in the Bahía de Jiquilisco, Usulután, El Salvador (Fig. 3B). This bay is 63,500 ha and is designated as a national conservation area. Further, it includes the Xirihualtique-Jiquilisco Biosphere Reserve within its boundaries and is also a Ramsar site.

Capture and Transmitters

In Texas, adult Reddish Egrets (n = 30) were trapped in late May–June during the 2010–2012 and 2014 breeding seasons. Reddish Egrets (hereafter, egrets) were trapped on nine colonies using noose mats (Sutherland et al. 2004; McGowan and Simons 2005) placed within 1 m of the nest. Specific colonies and nests were chosen based on the number of other birds nesting on the island, location of the nest in relation to other species, and accessibility of the nest to minimize disturbance to other nesting birds. Egrets were trapped during mid to late incubation to minimize nest abandonment. In Florida, three adult egrets and one immature egret were trapped during the nonbreeding seasons (October–February) of 2009–2010 and 2013–2014, and one adult was captured during the breeding season (July) of 2010. Egrets were captured at foraging areas using a flip trap (Herring et al. 2008) or CO₂-powered net launcher (Advanced Weapons Technology). We recorded mass (g), tarsus length (mm), middle toe length (mm), culmen length (mm), and color morph, and affixed an aluminum leg band and a colored, alphanumeric plastic leg band to each egret. Solar GPS satellite platform terminal transmitters (PTT-100, Microwave Telemetry, Inc.) weighing 22 g were attached with Teflon ribbon as a backpack harness. Satellite PTTs were < 3% of total body weight (n = 26) or 3.00–3.24% (n = 9) and were accurate to 18 m. Total handling time was < 30 min. The satellite PTTs were set to record locations six times per day, at 08:00, 09:00, 16:00, 17:00, 24:00, and 01:00 hr, coinciding with peak foraging times and nocturnal roosting. Two satellite PTTs placed on individuals in Florida recorded locations at 09:00, 10:00, 17:00, 18:00, 01:00, and 02:00 hr, two recorded locations at 05:00, 08:00, 10:00, 12:00, 14:00, 16:00, 18:00, and 21:00 hr and one recorded location at 09:00, 14:00, 16:00, 18:00, 20:00, and 22:00 hr. All times are in local time; Central Time for egrets marked in Texas and Eastern Time for egrets marked in Florida. Locations were downloaded once weekly from the Argos system (Argos 2011) and uploaded to Movebank (Wikelski and Kays 2016). We determined the sex of Reddish Egrets using blood samples (n = 13) and genetic samples from collected carcasses of marked individuals that died during the study period (n = 7), as blood samples had not been taken for those individuals. Discriminant analysis was used to determine sex of 15 egrets for which we did not have a DNA sample (Koczur et al. 2015).

Winter Season

In Texas, the winter season was October–February. Nest initiation dates have been recorded as late as mid-May to mid-June with re-nesting occurring as late as mid-July (McMurry 1971). Reddish Egrets incubate for ∼26 days, and the time from hatching to fledging (without provisioning from adults) is ∼9 weeks. Therefore, we assumed that Reddish Egrets ceased provisioning young by October. For each individual that migrated, the winter season began when it reached its wintering area and ended when it left the area. Colony formation in the Laguna Madre begins as early as March (Paul 1991); therefore, the winter season ended in February for resident egrets. In Florida, the winter season was determined for each individual because breeding can occur throughout the year (Paul 1991). GPS locations were used to determine that Reddish Egrets were no longer exhibiting nesting or provisioning behavior, as they no longer returned to the colony. Data were used until 25 February 2016 (Texas) and 12 November 2015 (Florida).

Nocturnal roosting locations were those recorded at 24:00 and 01:00 hr in Texas and 01:00, 02:00, 04:00, 05:00, and 21:00 hr in Florida (dependent on the duty cycle of the transmitter). An individual was considered roosting if the distance between the successive locations within a night was < 50 m, which takes into account the accuracy of the PTTs (± 18 m). If an egret moved > 50 m between the two locations, it was likely flying and those locations were excluded from roost site analyses.

Roosting and Foraging Habitat

The distance from roosting locations and foraging locations to the mainland or barrier island was measured in ArcMAP (Environmental Systems Research Institute 2015) using the near tool, and the mean (± SD) and least square mean (LSM) are reported. Benthic habitat at roosting and foraging locations in the Laguna Madre, Texas, were delineated using a benthic habitat dataset (National Oceanic and Atmospheric Administration 2007). Habitat types included land (areas above high-tide line), unconsolidated sediment (< 10% seagrass coverage; unvegetated tidal flats), patchy seagrass (10–75% coverage), continuous seagrass (76–100% coverage), reef, emergent marsh, mangrove, and unknown (areas that could not be classified) (Finkbeiner et al. 2009). A land cover dataset and a benthic habitat dataset (National Oceanic and Atmospheric Administration 2007) were used to delineate habitat at sites in Florida. Habitat types included tidal flats, salt marsh, mangrove, oyster beds, coral, hard bottom, hardwood forest, and seagrass. The intersect tool in ArcMap was used to extract habitat features at each location. We created convex hulls in ArcMAP using the minimum bounding geometry tool and winter locations for each individual to delineate available area. Continuous seagrass and patchy seagrass habitat types were combined because of small sample sizes within these categories and because of the relative similarity in their structure. Benthic habitat datasets were not available for sites in Mexico and El Salvador; therefore, Google Earth (Google, Inc.) imagery was used to describe the roosting and foraging habitat of those sites. We used the package adehabitatHS v. 0.3.12 (Calenge 2011) in statistical program R (R Development Core Team 2014) to calculate Manly selection ratios (w_i; Manly et al. 2007). We used the Design III model, which measures both habitat availability and use of habitat by each individual.

To further assess whether foraging and roosting habitat were similar, we used the kerneloverlap function in the adehabitatHR package (Calenge 2015) in statistical program R (R Development Core Team 2014), which quantified the degree of overlap between roosting areas and foraging areas (i.e., home ranges) in each winter season. Bhattacharyya's Affinity Index (BAI) was used to measure overlap and provided a value ranging from 0–1; a value of zero indicated no overlap, whereas a value of one indicated complete overlap (Bhattacharyya 1943; Fieberg and Kochanny 2005).

Distances Flown

Straight-line distances between successive locations were calculated using the Pythagorean Theorem in Microsoft Excel Office 2013 (Microsoft Corporation). We calculated the mean and standard error (SE) of distances moved from roost sites to morning foraging sites, from morning to evening foraging sites, and from evening foraging sites to roost sites for each individual. Three egrets from Florida had transmitters with different duty cycles, so times were matched as closely as possible to make analyses comparable. Because of a limited number of marked individuals at some sites (i.e., one egret in Campeche, Mexico), we grouped winter sites into regions: Texas, Tamaulipas, and Campeche were grouped as “Gulf Coast”, and Oaxaca and Usulután were grouped as “Pacific Coast”. Because of the marked differences in habitats available from other regions, Florida was not grouped with other sites and was treated as a distinct region. We used a three-way analysis of variance in SAS (SAS Institute, Inc. 2008; Proc MIXED) with time of day as a repeated measure to determine if time of day, gender, migratory status, region, and their interactions influenced mean distance moved. Years were pooled for analyses and the Shapiro-Wilk test was used to assess distribution of variances for each treatment combination.

Site Fidelity and Home Range

We assessed roost site and foraging site fidelity by calculating the area (ha) of 95% utilization distributions using the adehabitatHR package (Calenge 2015) in program R (R Development Core Team 2014). We then used Bhattacharyya's Affinity Index to measure the overlap of utilization distributions across successive and nonsuccessive winters. The averages of least square means for roosting fidelity and foraging fidelity were calculated. We used the least square cross validation method and all locations within a season for each individual. This metric was used to assess whether there were significant differences in areas of use among individuals. Analysis of variance in SAS (Proc GLM) was used to determine if home range size varied among sexes, color morphs, migratory status, or winter sites (grouped as above). The Shapiro-Wilk test was used to assess distribution of variances for each treatment combination.

Wintering Areas

Sixteen of the marked Reddish Egrets were resident in the Laguna Madre of Texas, USA, and 11 wintered at sites in Mexico and El Salvador. Two of the 16 residents made northward movements out of Laguna Madre during one winter, but not in subsequent winters; one female flew ∼190 km from Laguna Madre to Galveston Bay (29° 32′ 0″ N, 94° 46′ 0″ W) for approximately 3.5 months and another moved ∼100 km north of Laguna Madre into Aransas Bay (28° 2′ 33″ N, 96° 59″ 37.72″ W) for ∼2 months. The other 14 resident egrets in Texas remained within the Laguna Madre through the duration of the study. Four Reddish Egrets dispersed to winter sites in Tamaulipas, Mexico. Three of the four egrets that wintered in Tamaulipas occurred just south of the border (< 32 km) within the Rio Grande Delta (Fig. 1A). The fourth was slightly farther inland within the northern Laguna Madre de Tamaulipas (∼20 km from the coastline). One egret wintered in the Laguna de Términos, Campeche, Mexico (Fig. 1B), and four egrets wintered in the Laguna del Mar Muerto, Oaxaca, Mexico (Fig. 3A). One egret wintered in Usulután, El Salvador, within a relatively small range in the Bahía de Jiquilisco (Fig. 3B). One egret died during autumn migration west of Tampico, Mexico, prior to reaching its wintering area. Three egrets either died or the transmitter failed before their first winter season. The five egrets that were monitored in Florida remained resident and showed no migratory movements.

Figure 1.

Winter sites (2010–2016) of adult Reddish Egrets marked with satellite transmitters in the Laguna Madre, Texas, USA, during the breeding season: (A) the Laguna Madre of Texas, USA, to the north and the Laguna Madre de Tamaulipas, Mexico, to the south and (B) the Laguna de Términos, Campeche, Mexico.

Roosting and Foraging Habitat

Roost locations of adult Reddish Egrets during winter occurred from 0.12 (0.10) to 3.20 (2.29) km from the mainland (overall LSM = 1.37 + 0.71 km) and distances were relatively consistent among sites. Benthic habitat at roost sites was assessed for 15 egrets that wintered within the Laguna Madre, Texas, totaling 2,746 locations. Nearly half (47.5%) of roost locations occurred on unconsolidated sediment, whereas ∼22% of locations were in areas dominated by seagrasses (Fig. 4). Most egrets used unconsolidated sediment for roosting, which was more than expected (w_i = 2.44, SE = 0.31), whereas land (w_i = 0.77, SE = 0.32) and seagrass (w_i = 0.45, SE = 0.05) were used slighdy less than expected. A total of 917 locations were used to examine benthic habitat at roost sites for five egrets wintering in Florida. Salt marsh (w_i = 5.64, SE = 2.57) was used more than expected, whereas tidal flats (w_i = 0.88, SE = 0.21) and seagrass (w_i = 0.76, SE = 0.45) were used slightly less than expected. However, habitats used varied among individuals (Fig. 5).

Figure 2.

Winter sites on (A) the upper coast of Florida, USA (2014-2015), (B) the central coast of Florida, USA (2014–2015), and (C) the Florida Keys, USA (2010–2011) for Reddish Egrets marked with satellite transmitters.

Figure 3.

Winter sites of adult Reddish Egrets marked with satellite transmitters in the Laguna Madre, Texas, USA during the breeding season: (A) the Rio Verde, Oaxaca, Mexico, to the west and the Laguna Mar Muerto, Oaxaca and Chiapas, Mexico, to the east (2010-2016) and (B) the Bahía de Jiquilisco, Usulután, El Salvador (2014–2016).

Figure 4.

Percent or roosting locations at benthic habitat types for 15 Reddish Egrets wintering in the Laguna Madre, Texas, USA, during 2010–2016.

The mean distance from foraging locations to the mainland/barrier island ranged from 0.04 to 3.82 km among individuals, and the LSMs for each winter area ranged from 0.04 km in Campeche to 2.04 km in Texas. The overall LSM was 1.71 km. Benthic habitat was assessed at foraging sites for 15 egrets that wintered in the Laguna Madre, Texas, totaling 12,784 foraging locations. Similar to roosting habitat, Reddish Egrets in Texas showed preference for unconsolidated sediment (w_i = 2.71, SE = 0.45) and avoided foraging areas with seagrass (w_i = 0.25, SE = 0.07). Sixty-three percent of foraging locations in Texas were on unconsolidated sediment, whereas ∼13% were in areas dominated by seagrass (Fig. 6). Benthic habitat was assessed for five egrets that wintered in Florida using 4,706 foraging locations. Similar to Texas, Reddish Egrets in Florida showed a tendency to avoid foraging in areas dominated by seagrass (w_i = 0.44, SE = 0.04). Egrets in Florida showed preference for tidal flats (w_i = 7.96, SE = 1.51) and salt marsh (w_i = 2.08, SE = 0.96), yet habitat use varied considerably among individuals (Fig. 7).

Figure 5.

Percent of roosting locations at benthic habitat types of five Reddish Egrets wintering in Florida. USA, during 2010–2015.

Sixty-six percent of Reddish Egrets showed considerable overlap (BAI > 0.50) of foraging and roosting areas during each winter (Table 1). Forty-seven percent of the 58 individual*year combinations exhibited nearly complete overlap (BAI > 0.97). Egret 49148 was a resident in Texas and had no overlap of foraging and roosting areas in the first winter, and a low degree of overlap in subsequent winters. This egret primarily foraged along a chain of islands in the Laguna Madre. It roosted ∼3 km north in shallow water and on a small island, and on a cabin in the middle of the water ∼0.5 km west of foraging sites. Egret 49422b wintered in Tamaulipas, adjacent to what appears to be abandoned salt or aquaculture ponds. Foraging and roosting areas overlapped (Table 1); however, there were distinct roost sites that occurred in the center of a tidal flat and shallow water of a wetland. In Campeche, Mexico, egret 49154 exhibited extensive overlap in foraging and roosting areas in an estuary surrounded by mangroves near an inlet (Table 1). In Florida, all egrets exhibited overlap between foraging and roosting areas; however, egret 49154b exhibited the smallest degree of overlap (Table 1). This egret foraged around Hagens Cove and roosted approximately 3.5 km south on Big Grass Island (Fig. 2A). All roost locations were on this island, and some foraging locations occurred near the island.

Figure 6.

Percent of foraging locations at benthic habitat types for 15 Reddish Egrets wintering in the Laguna Madre, Texas, USA, during 2010–2016.

Distances Flown

The mean straight-line distance between roosting and morning foraging locations, morning and afternoon foraging locations, and afternoon foraging and roosting locations ranged from 0.1 to 4.8 km across all birds. The main effects of migratory status (F₂₂ = 27.22, P < 0.001) and location (F₄₈= 27.94, P < 0.001) were significant. Reddish Egrets that were resident flew greater distances between foraging and roosting areas (2.28 km, SE = 0.15) than individuals that were migratory (0.99 km, SE = 0.18). Also, egrets wintering in Florida flew shorter distances (0.41 km, SE = 0.24) on average than egrets wintering on the western Gulf (2.27 km, SE = 0.12) and Pacific Coasts (2.23 km, SE = 0.25). There was a significant interaction between sex and time of day (F₄₈= 3.63, P = 0.034). Overall, females flew greater distances than males between the morning and evening foraging locations (1.70 ± 0.21 vs. 1.08 ± 0.17 km), whereas distances from the roost to morning foraging locations (2.28 ± 0.21 vs. 1.88 ± 0.17 km) and from evening foraging to roost locations (1.37 ± 0.21 vs. 1.49 ± 0.17 km) were similar. Overall, the LSM distance between successive daily roosting locations was 2.93 (± 2.61) km and ranged from 0.10 (± 0.13) to 8.62 (± 15.61) km among individuals and across all winters. The median distance ranged from 0.01 (± 0.01) to 3.46 (± 3.34) km.

Figure 7.

Percent of foraging locations at benthic habitat types of five Reddish Egrets wintering in Florida, USA, during 2010–2015.

Fidelity

Reddish Egrets exhibited a high degree of fidelity to roosting areas between successive years. Eighteen of the transmitters were active for 2 or more years and were therefore used to assess site fidelity across years. All individuals exhibited some degree of overlap in roosting areas, with 79% of 28 possible individual*year combinations having > 50% overlap (Table 2). The average least square mean BAI was 0.65 for roosting areas and 0.71 for foraging areas across all individuals. Egret 49164 had very low overlap between the first two years; it flew to Galveston Bay (188 km) during one wintering period, but did not make the same trip in subsequent winters. Reddish Egrets also exhibited relatively high fidelity to foraging areas across successive years; 82% of 28 individual*year combinations had ≤ 50% overlap (Table 2). The average least square mean BAI was 0.71 for foraging areas across all individuals. Four of five egrets had ≤ 45 % overlap across all nonsuccessive winters (Table 3). Egret 49164 had low overlap in nonsuccessive winters that included the winter in which it moved to Galveston Bay.

Table 1.

Bhattacharyya's Affinity Index of Reddish Egret foraging and roosting area overlap in each winter (Year 1 = winter of 2010–2011, Year 2 = winter of 2011–2012, etc.; bold text indicates where the individual occurred for the majority of the winter).

Home Range Size

Year was not a significant variable explaining variation in home range size; therefore, home range areas were averaged across years. An ANOVA was used to assess variation between sexes, morphs, migratory status, and winter locations. Treatment combinations were not normally distributed, so the natural logarithm of home range size was used. There was a significant interaction between sex and location (F_2,24 = 7.23, P = 0.004). Males and females had similar average home range areas in Florida (39 ha, 95% CI = 7-224 vs. 75 ha, 95%, CI = 9-639, P = 0.659) and at Pacific Coast sites (1,920 ha, 95%, CI = 334-10,938 vs. 1,636 ha, 95%, CI = 194-13,905, P = 0.943). At western Gulf Coast sites, males had a considerably smaller average home range area than females (116 ha, 95%, CI = 40-337 vs. 11,849 ha, 95%, CI = 4,946-28,282; P < 0.001).